Organizing Committee

General Chair Zalman J. PalmorFaculty of Mechanical EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: palmor@tx.technion.ac.il

Co-Chair Howard KaufmanElectrical, Computer and Systems Engineering DepartmentRensselaer Polytechnic InstituteTroy, New York 12180–3590

Program Chair Arie FeuerDepartment of Electrical EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: feuer@ee.technion.ac.il

Finance Chair Nahum ShimkinDepartment of Electrical EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: shimkin@ee.technion.ac.il

Publicity Chair Hector RotsteinRAFAEL—Armament Development AuthorityandDepartment of Electrical EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: hector@ee.technion.ac.il

Publication Chair Leonid MirkinFaculty of Mechanical EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: mersglm@tx.technion.ac.il

Local Arrangements Chair Per-Olof GutmanFaculty of Agricultural EngineeringTechnion—Israel Institute of TechnologyTechnion City, Haifa 32000, IsraelE-mail: peo@tx.technion.ac.il

Conference Secretariat Palex Tours59 Ha’atzmaut Rd.Haifa 33033, IsraelTel: 972-4-8524254Fax: 972-4-8522491E-mail: pl@netvision.net.il

International Program Committee

Ailon, A. (IL) Friedland, B. (USA) Ozcelik, S. (USA)Antsaklis, P. J. (USA) Goodwin, G. C. (AU) Ozguner, U. (USA)Athans, M. (USA) Groumpos, P. P. (GR) Praly, L. (FR)Banks, T. (USA) Haddad, A. H. (USA) Reich, S. (IL)Barkana, I. (USA) Ioannou, P. A. (USA) Sandberg, I. (USA)Bartolini, G. (IT) Isidori, A. (IT) Shaked, U. (IL)Berman, N. (IL) Kumar, P. (USA) Sideris, A. (USA)Bodson, M. (USA) Kucera, V. (CZ) Simaan, M. (USA)Cassandras, C. G. (USA) Landau, I. D. (FR) Spanias, A. (USA)Christodoulou, M. A. (GR) Levine, W. (USA) Valavanis, K. P. (USA)Chrysanthis, P. (USA) Lewis, F. L. (USA) Voulgaris, P. G. (USA)Demetriou, M. A. (USA) Marino, R. (IT) Wen, J. (USA)Djaferis, T. E. (USA) Middleton, R. H. (AU) Yaniv, O. (IL)Fradkov, A. L. (RU) Morse, A. S. (USA) Yee, J. (USA)Franklin, J. (USA) Ortega, R. (FR) Zeheb, E. (IL)

Sponsoring Organizations

IEEE Control Systems SocietyIsrael Association for Automatic Control (IBA)

The Institute for Advanced Studies in Mathematics at the TechnionTechnion— Israel Institute of Technology

Haifa Convention BureauBen-Gurion University of the Negev

Tel-Aviv UniversityIsrael’s Ministry of Science

In cooperation with

Omikron Delta (1927) LtdThe Israel Electric CorporationKOLLMORGEN Servotronix

Official Carrier

EL-AL Israel Airlines

Official Winery

Tishbi Estate Winery—Baron Wine Cellars Ltd

Contents

In Memoriam: Howard Kaufman 3

Introduction 5

Greetings from the General Chair 7

Technical Program Overview 7Plenary Lecture I (Graham C. Goodwin) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Plenary Lecture II (Michael Heymann) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Plenary Lecture III (Stephen Boyd) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Plenary Lecture IV (Yaakov Bar-Shalom) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9Plenary Lecture V (David S. Bayard) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10Presentation by representative of the European Commission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Social events 11Informal Get-Together Party . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Welcome to Haifa Reception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Festive Banquet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Good-Bye Coffee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Conference information 13Registration Desk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Courtesy Internet Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Exhibits and Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Hotel Floor Plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Haifa Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Message from the MCA President 15The MED List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Technical Program 17Program-at-a-Glance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18Sessions in MA Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Sessions in MM Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Sessions in MP Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22Sessions in TA Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Sessions in TM Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26Sessions in TP Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Sessions in WA Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30Sessions in WM Slot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Abstracts 35Session MA1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

1

Session MA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Session MA3 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Session MA4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40Session MA5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Session MM1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Session MM2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Session MM3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Session MM4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Session MM5 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Session MP1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Session MP2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Session MP3 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Session MP5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51Session TA1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Session TA2 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Session TA3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Session TA4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Session TA5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Session TM1 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Session TM2 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Session TM3 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Session TM4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Session TM5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Session TP1 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Session TP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Session TP3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Session TP4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Session TP5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Session WA1 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Session WA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Session WA3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Session WA4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Session WA5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Session WM1 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Session WM2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Session WM3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Session WM4 (I) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78Session WM5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Index of Authors, Chairpersons, and Organizers 81

Cooperating Organizations 87

2

IN MEMORIAM

Howard Kaufman (1940–1999)

Professor Howard Kaufman died suddenly on January 31, 1999 while cross-country skiing. He was 58 yearsold. Survivors include his wife Eve, two sons David and Jeffrey, and a daughter Deborah.

Professor Kaufman received his B.S., M.S., and Ph.D. in 1962, 1964, and 1965, respectively, all from Rensse-laer Polytechnic Institute, Troy, New York. After working in Cornell Aeronautical Laboratory and GE Researchand Development Laboratory, he rejoined RPI in 1969, where he was promoted in 1980 to full professor in theDepartment of Electrical, Computer, and Systems Engineering. In the summer of 1972, he was a NASA Sum-mer Faculty Fellow at NASA – Langley Research Center, and in the summer of 1982, he was an NSF IndustrialResearch Participant at General Electric Corporate Research and Development. He was on sabbatical leave atKnolls Atomic Power Laboratory when this untimely death occurred.

Professor Kaufman was active in the research of adaptive systems. One of his specialties was the applicationof adaptive control to drug delivery systems and other biomedical applications. He co-authored the populartext “Direct Adaptive Control Algorithms: Theory and Application.” He supervised 23 Ph.D. students andnumerous MS students. He had published more than 100 papers in various journals and control conferences.He was also active in conference organizations and a senior member of IEEE.

Professor Kaufman was an avid outdoorsman. He enjoyed hiking, skiing, walking, bicycling, and kayaking.He was also deeply religious. He was one of the founders of Congregation Beth Shalom in Clifton Park, NewYork, twenty-five years ago. He served as the head of religious education for several years. Professor Kauf-man also was key to leading the organizations affiliated with the Northeastern New York federation of JewishAgencies to develop applications on the worldwide web.

Professor Kaufman had the utmost respect for and interest in his multi-cultural colleagues and students, andenjoyed his many trips overseas. He looked forward to the IEEE Mediterranean Conference on Control andAutomation that he was to co-chair in Israel this summer. Going to Israel, as he had frequently done in the pastyears, was very special to him since it was a merging of his private Jewish and public professional life.

In memory for his dedicated service to RPI, the faculty at RPI has set up a Kaufman fellowship fund, whichwill be used to support graduate students.

Joe H. ChowRensselaer Polytechnic Institute.

3

4

MED99

Introduction

GREETINGS FROM THE GENERAL CHAIR

On behalf of the MED99 Organizing Committee and the co-sponsors, I welcome you to Haifa, Israel and theMED99 Conference. MED99 is the 7th of a very successful series of conferences, which take place every summeron the shores of the Mediterranean. In view of the unparalleled quality and number of contributions and thelarge attendance expected, I am convinced that MED99 will be the best MED Conference ever.

MED99 will be taking place in Haifa, one of the marvelous cities in Israel. Perched along the green slopes andpeaks of the Carmel Mountains, it affords beautiful vistas of the Haifa Bay spread out below, Acre silhouettedagainst the curling coastline, long beaches and the expansive sea beyond. Looking inland, the vistas are filledwith the wooded Carmel hilltops. Haifa is a city of many facets. A fascinating tourism town, it offers museums,a busy port and Druze villages. Haifa has the largest R&D high-tech center in Israel and is also an academic citycontaining both the Technion- Israel Institute of Technology and the Haifa University.

The organizing committee has worked hard to make the theme:

“MED99 — a conference you will never forget”

a reality. In addition to the excellent technical program assembled, the organizing committee has gone out ofits way to provide an exceptional social program at extraordinary locations. MED99 will be highlighted by agourmet banquet, sumptuous buffet lunches and excellent Tishbi wines. The organizing committee has alsoplanned pre and post-Conference tours to promote your enjoyment of the Conference and enable you to learnmore about the wonderful land of Israel.

Organizing the MED99 Conference involved a lot of work by many dedicated volunteers. I would like tothank the members of the organizing committee for their diligent and professional work.

Our work as organizers, though essential, does not make the Conference. It is made by the contributions ofthe authors, the session organizers and the leading specialists who deliver the plenary talks. I am thankful toall of them. My gratitude also goes to the Foundations, Institutions and Companies that gave their generousfinancial support to MED99. Their support enabled us to grant twenty-four fellowships to participating authorsfrom Eastern Europe, Asia and other Developing Countries.

Sadly, my co-chair, Howard Kaufman, died suddenly in the midst of the preparations for the conference. Itwas his initiative to have MED99 take place in Israel. It is really unfortunate that Howard didn’t live to see hiswish fulfilled.

With the strong technical program and the special social events, I believe that everyone will enjoy this MEDConference. I wish you a successful visit affording you both a professional and a cultural experience that willbe long remembered.

Zalman J. PalmorGeneral Chair – MED99.

TECHNICAL PROGRAM OVERVIEW

We have received over 260 manuscripts as a result of our call for papers and for invited sessions. All thesemanuscripts, including the invited sessions, have been reviewed and the decision for their acceptance wasbased on reviewers’ recommendations. The final program includes 202 papers to be presented in 39 sessions,11 out of which are invited sessions. The papers cover a broad range of control and systems related areas andrepresent the state of the art in research and practice conducted in these areas.

We wish further to highlight some points in our program. There is a whole day (three sessions) dedicated toprocess control and we look forward to active participation from our local process industry. At the other endof the spectrum, there are four (invited) sessions dealing with the mathematical aspects of infinite dimensionalsystems theory. A further review of the program reveals the many other areas, such as, robotics, aerospacecontrol, adaptive systems, manufacturing, and many more, included in the conference. We look forward to anexciting and stimulating experience with many exchanges of results and ideas.

7

This is a good opportunity to thank the colleagues who organized the invited sessions and to all the reviewerswho very diligently read the submitted manuscripts and helped in evaluating them.

The program includes five plenary presentations by leading researchers and specialists in the control area.In these presentations we have tried, and we believe with success, to create a mix of research and practice,traditional control and cutting edge new areas.

Plenary Lecture I Monday, June 28Graham C. Goodwin, The University of Newcastle 8:30 – 9:30“Identification and Robust Control: Bridging the Gap” Alon Hall

The topics of Identification and Robust Control have a rich history and have reached a level of considerablematurity. However, a major difficulty is that the two fields have evolved along different lines and now havemany incompatibilities. The aim of this paper is to raise awareness to this problem for researchers. We suggestthat alternative formulations may be desirable in both fields to obtain a satisfactory match. One possible line ofattack on the problem is proposed together with illustrations showing the potential merits in gaining a betterunderstanding of this problem.

Graham C. Goodwin was born in Broken Hill, Australia in 1945. He obtained a B.Sc. (Physics), B.E.(Electrical Engineering), and Ph.D. from the University of New South Wales. From 1970 until 1974 hewas a lecturer in the Department of Computing and Control, Imperial College, London. Since 1974he has been with the Department of Electrical and Computer Engineering, The University of Newcas-tle, Australia. He is the co-author of seven books: Control Theory (Oliver and Boyd, 1970), DynamicSystem Identification (Academic Press, 1977), Adaptive Filtering, Prediction and Control (Prentice Hall,1984), Digital Control and Estimation (Prentice Hall, 1989), Sampling in Digital Signal Processing and Con-trol (Birkhauser, 1996), Fundamental Limitations in Filtering and Control (Springer, 1997), Control SystemDesign (Prentice Hall, 1999), as well as several hundred technical papers.

Graham Goodwin is the recipient of several international prizes including a best paper award by IEEE Trans. AutomaticControl, and best engineering text book award from the International Federation of Automatic Control. He is currentlyProfessor of Electrical Engineering and Director of the Centre for Integrated Dynamics and Control at the University ofNewcastle. Graham Goodwin is a Fellow of IEEE; an Honorary Fellow of Institute of Engineers, Australia; a Fellow of theAustralian Academy of Science; a Fellow of the Australian Academy of Technology, Science and Engineering; and a memberof International Statistical Institute.

Plenary Lecture II Monday, June 28Michael Heymann, Technion — IIT 13:15 – 14:15“Control of Hybrid Systems and Some Applications” Alon Hall

Hybrid systems, in which discrete and continuous behaviors coexist and interact, have been receiving increasingattention in recent years both in the control theory community and in the computer science community. Whilefrom the control theory viewpoint, hybrid systems are dynamical systems endowed with discontinuities anddiscrete changes, from the computer science viewpoint, hybrid systems are viewed as discrete systems endowedwith dynamic timing constraints. These two viewpoints led to very different research agendas.

The present talk will review some of the main issues associated with analysis and control of a class of hybridsystems called hybrid machines so as to satisfy safety and liveness specifications. Several real-life practicalapplications of hybrid systems theory will be discussed.

Michael Heymann received the B.Sc. and M.Sc. degrees from the Technion, Haifa, Israel, in 1960 and1962, respectively, and the Ph.D. degree from the University of Oklahoma, Norman, in 1965, all inChemical Engineering.During 1965–1966 he was on the Faculty of the University of Oklahoma. From 1966 to 1968 he was withMobil Research and Development Corporation, engaged in research in control and systems theory.From 1968 to 1970 he was with the Ben-Gurion University of the Negev, Beer-Sheva, where he estab-lished and headed the department of Chemical Engineering. Since 1970 he has been with the Technionwhere he is currently Professor in the Department of Computer Science and Director of the Center forIntelligent Systems which he also founded. He is holder of the Carl Fechheimer Chair in Electrical

Engineering. He has previously been with the Department of Electrical Engineering and with the department of AppliedMathematics of which he was Chairman during 1972–1973. He held visiting positions at various institutes, including theUniversity of Toronto, the University of Florida, the University of Eindhoven, Concordia University, CSIR, Yale University,the University of Bremen and the University of Newcastle. Since 1983 he has been associated with NASA Ames ResearchCenter, where he has spent many summers as well as the years 1983–1984, 1988–1989, 1995–1996 as an NRC-Senior Research

8

Associate and more recently as a Jan-Jose-State University visiting Scientist.His research covered topics in the areas of linear system theory, differential games, optimization, and adaptive control. Hiscurrent interests are chiefly in the areas of discrete-event control, hybrid systems, the theory of concurrent processes andtheir applications.He has been on the editorial boards of the SIAM Journal of Control and Optimization and Systems & Control Letters.

Plenary Lecture III Tuesday, June 29Stephen Boyd, Stanford University 8:30 – 9:30“Optimization over Linear Matrix Inequalities” Alon Hall

The recent development of efficient interior-point algorithms for convex optimization problems involving linearmatrix inequalities (LMIs) has spurred research in a wide variety of application fields, including control systemanalysis and synthesis, combinatorial optimization, circuit design, structural optimization, experiment design,and geometrical problems involving ellipsoidal bounding and approximation.

In the first part of the talk, I will describe the basic problems, semidefinite programming (SDP) and deter-minant maximization, discuss their basic properties, and give a brief description of interior-point methods fortheir solution. In the second half of the talk I will survey applications from several areas.

Stephen Boyd received the A.B. degree in Mathematics, from Harvard University in 1980, and thePh.D. in Electrical Engineering and Computer Science from the University of California, Berkeley, in1985. In 1985 he joined the Electrical Engineering Department at Stanford University, where he is nowProfessor and Director of the Information Systems Laboratory. His interests include computer-aidedcontrol system design, and convex programming applications in control, signal processing, and circuits.

Plenary Lecture IV Wednesday, June 30Yaakov Bar-Shalom, University of Connecticut 8:30 – 9:30“Target Tracking and Data Fusion: How to Get the Most Out of Your Sensors” Alon Hall

This talk describes the evolution of the technology of tracking objects of interest (targets) in a cluttered envi-ronment using remote sensors. Approaches for handling target maneuvers (unpredictable motion) and falsemeasurements (clutter) are discussed. Advanced (“intelligent”) techniques with moderate complexity are de-scribed. The emphasis is on algorithms which model the environment and the scenarios of interest in a realisticmanner and have the ability to track low observable (LO) targets. The various architectures of informationprocessing for multisensor data fusion are discussed. Applications are presented from Air Traffic Control (datafusion from 5 FAA/JSS radars for 800 targets) and underwater surveillance for a LO target.

Yaakov Bar-Shalom was born on May 11, 1941. He received the B.S. and M.S. degrees from the Tech-nion — Israel Institute of Technology in 1963 and 1967 and the Ph.D. degree from Princeton Universityin 1970, all in electrical engineering. From 1970 to 1976 he was with Systems Control, Inc., Palo Alto,California. Currently he is Professor of Electrical and Systems Engineering and Director of the ESPLab (Estimation and Signal Processing) at the University of Connecticut. His research interests are inestimation theory and stochastic adaptive control and has published over 220 papers in these areas. Inview of the causality principle between the given name of a person (in this case, “(he) will track,” inthe modern version of the original language of the Bible) and the profession of this person, his interestshave focused on tracking. His other interests are stochastic control of vertical airfoils and of pairs of

inclined foot supports on crystals. He co-authored the monograph Tracking and Data Association (Academic Press, 1988), thegraduate text Estimation and Tracking: Principles, Techniques and Software (Artech House, 1993), the text Multitarget-MultisensorTracking: Principles and Techniques (YBS Publishing, 1995), and edited the books Multitarget-Multisensor Tracking: Applicationsand Advances (Artech House, Vol. I 1990; Vol. II 1992). He has been elected Fellow of IEEE for “contributions to the theoryof stochastic systems and of multitarget tracking.” He has been consulting to numerous companies, and originated theseries of Multitarget-Multisensor Tracking short courses offered via UCLA Extension, at Government Laboratories, privatecompanies and overseas. He has also developed the commercially available interactive software packages MULTIDATTM forautomatic track formation and tracking of maneuvering or splitting targets in clutter, PASSDATTM for data association frommultiple passive sensors, BEARDATTM for target localization from bearing and frequency measurements in clutter, IMDATTM

for image segmentation and target centroid tracking and FUSEDATTM for fusion of possibly heterogeneous multisensor datafor tracking. During 1976 and 1977 he served as Associate Editor of the IEEE Transactions on Automatic Control and from

9

1978 to 1981 as Associate Editor of Automatica. He was Program Chairman of the 1982 American Control Conference, Gen-eral Chairman of the 1985 ACC, and Co-Chairman of the 1989 IEEE International Conference on Control and Applications.During 1983–87 he served as Chairman of the Conference Activities Board of the IEEE Control Systems Society and during1987–89 was a member of the Board of Governors of the IEEE CSS. In 1987 he received the IEEE CSS Distinguished MemberAward. Since 1995 he is a Distinguished Lecturer of the IEEE AESS. He is co-recipient of the M. Barry Carlton Award for thebest paper in the IEEE Transactions on Aerospace and Electronic Systems in 1995.

Plenary Lecture V Wednesday, June 30David S. Bayard, Jet Propulsion Laboratory 13:15 – 14:15“Deep Space Control Challenges of the New Millennium” Alon Hall

The exploration of deep space presents a variety of significant control challenges. Long communication delayscoupled with challenging new science objectives require high levels of system autonomy and increasingly de-manding pointing and control capabilities. Historically, missions based on the use of a large single spacecrafthave been successful and popular since the early days of NASA. However, these large spacecraft missions arecurrently being displaced by more frequent and more focused missions based on the use of smaller and lessexpensive spacecraft designs. This trend drives the need to design smart software and good algorithms whichtogether with the miniaturization of control components will improve performance while replacing the heavierand more expensive hardware used in the past.

NASA’s future space exploration will also include mission types that have never been attempted before,posing significant challenges to the underlying control system. This includes controlled landing on small bod-ies (e.g., asteroids and comets), sample return missions (where samples are brought back from other planets),robotic exploration of planetary surfaces (e.g., intelligent rovers), high precision formation flying, and deepspace optical interferometry,

While the control of planetary spacecraft for traditional flyby and orbiter missions are based on well-understood methodologies, control approaches for many future missions will be fundamentally different. Thisparadigm shift will require completely new control system development approaches, system architectures, andmuch greater levels of system autonomy to meet expected performance in the presence of significant environ-mental disturbances, and plant uncertainties. This paper will trace the motivation for these changes and willlayout the approach taken to meet the new challenges. Emerging missions will be used to explain and illustratethe need for these changes.

Dr. David S. Bayard is a Senior Research Scientist at the Jet Propulsion Laboratory, California Insti-tute of Technology. He received the B.A. degree in mathematics and chemistry from Queens Collegeof the City University of New York in 1977, and the M.S. and Ph.D. degrees in electrical engineeringfrom the State University of New York at Stony Brook in 1979 and 1984, respectively. During the period1980–1983 he served as an industrial consultant to Norden Systems, General Instruments, and ComtechLaboratories, involved in adaptive filtering for radar tracking systems, nonlinear system identification,and optimal adaptive control for countermeasures. In 1984, he served on the faculty at SUNY StonyBrook as a Visiting Professor, and in 1985 joined the Guidance and Control section of the Jet PropulsionLaboratory.

Dr. Bayard is currently a Guidance and Control Research Group Leader at JPL. He has been the technical lead for research inthe identification and control of large flexible structures, adaptive control of robotic manipulators, attitude estimation andcontrol for miniature spacecraft, adaptive vibration control, nonlinear estimation for robotic balloons, development of Marsprecision landing technology, reconfigurable control design for the Space Infra-Red Telescope Facility (SIRTF), and the appli-cation of modern estimation and control techniques to numerous emerging spacecraft and planetary missions. Dr. Bayard’sresearch interests include adaptive control, signal processing, system identification, dynamic programming and optimiza-tion, stochastic control, and optimal experiment design, and he has published over 100 journal and conference papers inthese areas. In 1996 he received the NASA Exceptional Service Medal for fundamental contributions to autonomous space-craft control systems. Dr. Bayard is a member of IEEE, SIAM, AIAA, Phi Beta Kappa and Beta Delta Chi.

Presentation by representative of the European Commission Monday, June 28Dr.-Ing. Alkis Konstantellos 18:00 – 18:30“Control Systems and European R&D Programmes: Past Experiences and New Challenges” Alon Hall

In this presentation two topics are addressed: (a) the content and results of the last four European R&D Frame-work programmes since 1984 and the spectrum of control systems applications funded and (b) the 5th R&DFramework Programme (1999–2002) which was launched in February 1999 and the mechanisms and opportu-nities for collaborative research.

10

Up to now, although there has not been an obvious central area for control systems, the European Union pro-grammes have supported a significant number of related research projects, working groups, trials and take upactions affecting vital parts of the production and services innovation processes. These have addressed (a) eithergeneric and advanced control algorithms, architectures and systems including safety critical and complexity is-sues or (b) sectoral applications (primarily in automotive/aerospace, process including batch, semiconductorand electronics industries). Other projects dealt with controls combined with decision, concurrent engineering,visualization, simulation and industrial communications from design through operations to the integration andmanagement of controls with instrumentation and business systems (MES and higher levels).

The new EU R&D programmes and in particular the Information Society Technologies programme (IST) areuser-driven with specific support for radical innovation and high risk ideas. IST is an integrated 3.6 Bi EUROprogramme, running for 4 years and encompassing the previous programmes ESPRIT(IT), TELEMATICS andACTS (telecomms).

Based on experience with the technological developments of the last few years and a set of strategic objectives,the IST programme is designed along four focused “key actions,” two additional activities and flexible cross-programme topics:

I: Systems and services for the citizen (e.g., transport, health care, environment and administration);

II: New methods of work and e-commerce;

III: Multimedia content and tools;

IV: Essential technologies and infrastructures (IT and comms);

FET: Future and emerging technologies;

RN: Research networking;

CAP: Cross-programme actions.

Within this context several new challenges could be posed to control systems developers and users in thisprogramme, e.g.: 1) encourage the creation of revolutionary control methods and facilitate implementationof affordable and dependable solutions, 2) stimulate a response to expected new and promising paradigmsfrom/to other disciplines such as quantum computing and communications, information ecosystems and nano-technologies, 3) create synergies between the ubiquitous control domains and related enabling technologiessuch as real time systems, next generation instrumentation, fast interfaces and active networks as well as inter-actions with user industries, 4) enhance micro and macro level collaborations and awareness among interestedparties and 5) conduct market and socio-economic analyses and impact assessments of control systems researchand related projects.

The new R&D programmes of the European Union and in particular the IST programme provide a broadportfolio of themes and could facilitate the exploition of the rich ideas of the control systems and automationcommunities.

SOCIAL EVENTS

Last December, at the IEEE Conference on Decision and Control in Tampa, Florida, one of the MED99 contribut-ing authors confided that in general IEEE sponsored conferences are located too far from the right places. Thisconference, however, is located too near the wrong places, he continued, pointing to the one-storey buildingopposite the conference hotel that was adorned with a large, pink, neon sign proclaiming: “The best women inTampa”!

In contrast, the IEEE co-sponsored 7th Mediterranean Control Conference is located exactly in the right place:the luxury hotel Dan Panorama in the middle of the Carmel Center with its shops and entertainment, on theverge of the Carmel Mountain with a splendid view of the cultured city of Haifa and the beautiful Haifa Bay.

The Organizing Committee has made a great effort to embellish the conference and the beautiful surroundingswith attractive social events for the participants and accompanying guests. Our aim is to make reality of theconference motto: “MED99 — a conference you will never forget.”

We are particularly proud to announce that MED99 is the first control conference with an Official Winery:

11

Official Carrier Official Winery

Baron Wine Cellars Ltd.

The Tishbi Estate Winery — Baron Wine Cellars Ltd.

This high quality boutique winery from Binyamina delivers premier label “Jonathan Tishbi Special Reserve”Chardonnay 1997 and “Tishbi Estate” Sauvignon Blanc 1998, Cabernet Sauvignon 1997, and Merlot 1997 winesfor our enjoyment during the lunches, the reception, and the festive banquet.

Informal Get-Together Party Sunday, June 27 (19:30 – 21:30)

The evening before MED99 gets started, we meet at the Dan Panorama Nightclub for an informal get-togetherparty. This party provides a good occasion for relaxing after a long flight, and meeting with other conferenceparticipants. Everybody will certainly find partners to dance with and drink with! And by the way, to make theatmosphere really convival and unforgettable — all drinks, except hard liquor, are on the house!

Welcome to Haifa Reception Monday, June 28 (19:30 – 21:30)

Wrapping up the first day of MED99, there is a “Welcome to Haifa Reception,” in part sponsored by the City ofHaifa. The reception is held at the Clandestine Immigration and Naval Museum. The charming mayor of Haifa,Mr. Amram Mitzna, will give a speech. Professional guides will tour the museum with us. We will be told andshown the story of the clandestine Jewish immigration to the then British Mandate of Palestine during the 30’sand 40’s before Israel’s independence in 1948. The landing on the Mediterranean shore was often dramatic —if the British authorities discovered the immigrants they were often deported to Cyprus or to Europe. Theseevents shaped history, and have been described in several books and films, e.g. Leon Uris’s “Exodus.” A bandwill play while we eat and drink. And of course, Prof. Zalman Palmor, the General Chair of MED99 will give awelcoming address.

Please note that the buses taking us to the Reception will leave the Dan Panorama Hotel and the Holiday Innat 19:10 sharp. If you miss the bus, please turn to the Conference Registration Desk in the lobby of the DanPanorama Hotel, where you will get instructions how to take a local taxi to reach the Museum.

Festive Banquet Tuesday, June 29 (19:00 – 23:00)

After the second day of MED99, the participants and their accompanying guests will ride to the heart of theGalilee for a festive and elegant Banquet at the Yehiam Fortress in the Yehiam National Park.

The earliest remains at Yehiam are from Roman times, but the main building is an impressive Crusader fort,named “Judin” in Arabic. The Teutonic order strengthened the fortress in 1208 but it fell anyway in 1265 toMammeluke Sultan Baybars who destroyed it. The fortress was rebuilt in the 18th century by a local Beduinruler who used it as a stronghold. Recently, the Israel National Park Authority reconstructed parts of the fortressand prepared the cite for visitors.

The view from the Yehiam Fortress over the Galilee is splendid. To the west the Mediterranean meets thehorizon, and we will make sure to be on time to see the romantic sunset accompanied with a welcoming cocktail.A concert of Renaissance music will follow, played on original instruments by the renowned Modus VivendiEnsemble. We will enjoy the gourmet dinner a fresco under the full moon on the upper terrace of the fortress. Ashort after-dinner stroll in the park is recommended as the conclusion of the evening.

Please note that the busses leave the Dan Panorama Hotel and the Holiday Inn at 17:50 sharp. If you missthe bus, there is probably no way you will reach the Yehiam Fortress on time for the sunset, unless you have arented car and a good sense of directions.

Additional banquet tickets may be purchased at the Registration Desk (see p. 13) for $58.5 each.If you cannot participate in the Festive Banquet and hold a ticket, would you please return it to the Registra-

tion Desk so that we may donate it to one of the MED99 student participants.

Good-Bye Coffee Wednesday, June 30 (16:30 – 17:00)

The third and final day will be concluded with a cup of coffee and cake at the Dan Panorama Hotel. We hopethat the technical program has been rewarding, the exhibitions interesting, that you have learned many newthings, met new and old colleagues, and, above all, that you have had fun. The Good-Bye Coffee is the timeto say Shalom ve’Lehitraot, Bye and See you again. But also to plan the remainder of the visit to Israel, maybetogether with a new friend.

12

CONFERENCE INFORMATION

Registration Desk

The Registration Desk is located at the entrance hall and is open during the following hours:

• Sunday, June 27 16:00 – 20:00

• Monday, June 28 7:30 – 19:30

• Tuesday, June 29 8:00 – 18:30

• Wednesday, June 30 8:00 – 17:00

Participants may register on-site during the opening hours of the Registration Desk. The on-site registration feeamounts to $433 ($175.5 for students).

Additional CD-ROM proceedings may be purchased at the Registration Desk for $29 each.

Courtesy Internet Communication

A courtesy communication station which includes a PC with access to Internet and a telnet program is availablefor usage by MED99 participants. The station is located in the Brosh Hall (entrance level) and will be active forthe three days of the conference during the Registration Desk opening hours (see above).

Exhibits and Displays

MED99 will host an exhibit by Israeli manufacturers and distributors of control-related products, as well as ascientific book display. The exhibits are located at the Brosh Hall (entrance level).

Hotel Floor Plans

ÔÂÈÏچȷÂφ˙Ó˜

ÏӯΆ‰Ù˜

È·ÂÏ

ү‚˙ÈÙ

MAIN LOBBY LEVEL

LOBBY

PYTHAGORAS

CAFE CARMEL

≤∞†‰Ó˜

Ìχ¯ÓÂ˙

ʯ‡†Ìχ

ԯ‡†Ìχ

20th FLOOREREZ HALL

TOMERHALL

ORENHALL

Ú˜¯˜†˙Ó˜

ENTRANCE LEVEL

ENTRANCE

Dekel HallAlon Hall

Brosh Hall

ÔÂÈÏچȷÂφ˙Ó˜

ÏӯΆ‰Ù˜

È·ÂÏ

ү‚˙ÈÙ

MAIN LOBBY LEVEL

LOBBY

PYTHAGORAS

CAFE CARMEL

≤∞†‰Ó˜

Erez Hall 20th FLOOR

TomerHall

OrenHall

Ú˜¯˜†˙Ó˜

‰ÒÈÎ

˘Â¯·†Ìχ

Ϙ„†ÌχÔÂχ†Ìχ

ENTRANCE LEVEL

ENTRANCE

DEKEL HALLALON HALL

BROSH HALL

ÔÂÈÏچȷÂφ˙Ó˜

Cafe Carmel

MAIN LOBBY LEVEL

Lobby

Pythagoras

≤∞†‰Ó˜

Ìχ¯ÓÂ˙

ʯ‡†Ìχ

ԯ‡†Ìχ

20th FLOOREREZ HALL

TOMERHALL

ORENHALL

Ú˜¯˜†˙Ó˜

‰ÒÈÎ

˘Â¯·†Ìχ

Ϙ„†ÌχÔÂχ†Ìχ

ENTRANCE LEVEL

ENTRANCE

DEKEL HALLALON HALL

BROSH HALL

13

Haifa Map

14

MESSAGE FROM THE MCA PRESIDENT

The MED99 conference in Haifa, Israel is the 7th Mediterranean Conference on Control and Automation(MCCA). Information about all previous Med conferences is included below. The first Med conference wasin Crete, Greece in 1993 and there has been strong interest in the MCCA every year since then. The success ofthe present conference in Haifa bears witness to this fact. What does the future hold? In 2000 MCCA returns toGreece in Patras. In 2001 it goes to Dubrovnik, Croatia and in 2002 MCCA will be Lisbon, Portugal.

In June 1998 the group of researchers who had been supervising the MCCA decided to form a parent orga-nization named the MEDITERRANEAN CONTROL ASSOCIATION (MCA). MCA’s interests are in the broadtechnical fields of Systems, Control and Automation and its main goals are to promote initiatives aiming atenhancing scientific exchanges, to disseminate information, to coordinate research networks and to facilitatetechnology transfer primarily within the countries of the Mediterranean region. Its main functions are thecoordination and supervision of the Mediterranean Conference on Control and Automation (MCCA) and thedissemination of information primarily via electronic means. We are currently in the process of establishing aWeb site for easier dissemination of information with mirror sites in Greece and the US. Keep an eye out for uson the Internet in the near future.

I hope to see you next year in Greece!

Panos AntsaklisMCA President.

The MED List

1ST IEEE MEDITERRANEAN SYMPOSIUM ON NEW DIRECTIONS IN CONTROL THEORY AND APPLICATIONSJune 21–23, 1993, Handris Hotel, Maleme, Chania, Crete, GreeceGeneral and Program Chair: Manolis A. ChristodoulouProgram co-Chair: Petros A. Ioannou

2ND IEEE MEDITERRANEAN SYMPOSIUM ON NEW DIRECTIONS IN CONTROL AND AUTOMATIONJune 19–21, 1994, Louis Maleme Beach Hotel, Chania, Crete, GreeceHonorary Chair: George N. SaridisGeneral Chair: Kimon P. ValavanisProgram Chair: Frank L. Lewis

3RD IEEE MEDITERRANEAN SYMPOSIUM ON CONTROL AND AUTOMATIONJuly 11–13, 1995, Sheraton Hotel, Limassol, CyprusGeneral Chair: Petros A. IoannouProgram Chair: Frank L. Lewis

4TH IEEE MEDITERRANEAN SYMPOSIUM ON CONTROL AND AUTOMATIONJune 10–13, 1996, Louis Maleme Beach Hotel, Chania, Crete, GreeceHonorary Chair: Panos J. AntsaklisGeneral Chair: Frank L. LewisGeneral Co-Chairs: Petros P. Groumpos and Paris N. ParaskevopoulosProgram Co-Chairs: Kostas Kyriakopoulos and Petros G. Voulgaris

5TH IEEE MEDITERRANEAN CONFERENCE ON CONTROL AND SYSTEMSJuly 21–23, 1997, Phaethon Beach Hotel & Club, Paphos, Cyprushttp://www.ecs.umass.edu/ece/djaferis/5thMED/General and Program Chair: Theodore E. DjaferisFinance and Special Events Chair: Petros A. Ioannou

6TH IEEE MEDITERRANEAN CONFERENCE ON CONTROL AND AUTOMATIONJune 9–11, 1998, Hotel Carlos V, Alghero, Sardinia, Italyhttp://www.disp.utovrm.it/˜med98/General Chair: Antonio TornambeProgram Chair: Giuseppe ContePublications Chair: Anna Maria Perdon

15

16

MED99

Technical Program

Program-at-a-Glance

Track 1 2 3 4 5Room Dekel Hall

(Entrance level)Alon Hall

(Entrance level)Oren Hall(20th floor)

Tomer Hall(20th floor)

Erez Hall(20th floor)

27.06 19:30 – 21:30 Informal Welcoming Reception (at the Dan Panorama Nightclub)

MO

ND

AY,

Jun

e28

8:15 – 8:30 Opening Remarks – Alon Hall

8:30 – 9:30 Plenary talk (Graham C. Goodwin) – Alon Hall

9:30 – 10:00 Coffee break

MA10:00 – 12:00

Optimal Control:H2/H∞/`1 Discrete Event and

Hybrid Systems

Automotive Controland Energy

Conversion SystemsLinear Systems 1 Adaptive Control 1

12:00 – 13:15 Lunch

13:15 – 14:15 Plenary talk (Michael Heymann) – Alon Hall

MM14:30 – 16:30

Optimization MethodsIntelligent Control and

Neural NetworksControl Applications Linear Systems 2

Nonlinear SystemIdentification in

Practice

16:30 – 16:50 Coffee break

MP16:50 – 18:30

Flexible StructuresFault Detection

European R&D Programs∗

(18:00)

Target Tracking Adaptive Control 2

19:30 – 21:30 Opening Reception (at the Clandestine Immigration and Naval Museum)

TU

ES

DA

Y,Ju

ne

29

8:30 – 9:30 Plenary talk (Stephen Boyd) – Alon Hall

9:30 – 9:50 Coffee break

TA9:50 – 11:50

Sampled-Data SystemsRecent Innovations in

Process ControlStability andStabilization

Nonlinear Systems 1 Filtering

11:50 – 13:10 Lunch

TM13:10 – 15:10

Control of DistributedParameter Systems

Integration of ProcessDesign and Process

Control

Advances to Meet theMissile Guidance

Challenge at the Vergeof the New Millennium

Nonlinear Systems 2 Identification 1

15:10 – 15:30 Coffee break

TP15:30 – 17:10

Dynamical Models Process Control Aerospace Control Time Delay Systems Identification 2

19:00 – 23:00 Festive Banquet (at the Yehiam Crusader Fortress)

WE

DN

ES

DA

Y,Ju

ne

30

8:30 – 9:30 Plenary talk (Yaakov Bar-Shalom) – Alon Hall

9:30 – 10:00 Coffee break

WA10:00 – 12:00

Optimal Control Manufacturing Robotics 1 Robust ControlSignal and Image

Processing

12:00 – 13:15 Lunch

13:15 – 14:15 Plenary talk (David S. Bayard) – Alon Hall

WM14:30 – 16:30

Parameter Estimation Fuzzy Logic Methods Robotics 2 Sliding Mode ControlComputer Networksand Queing Systems

16:30 – 17:00 Good-bye coffee

∗Presentation by representative of the European Commission Dr.-Ing. Alkis Konstantellos “Control Systems and European R&D Programmes: Past Experiences and New Challenges.”

18

MONDAY, June 28

Plenary Session I Alon Hall

IDENTIFICATION AND ROBUST CONTROL:Bridging the Gap

Graham C. GoodwinCentre for Integrated Dynamics & Control

Dept. of Electrical and Computer EngineeringThe University of Newcastle

Chair: Palmor, Zalman J. Technion — IIT

MA1 Dekel HallOptimal Control: H∞/H2/`1Chair: Rotstein, Hector RAFAEL — ADACo-chair: Bar-Gil, Aharon RAFAEL — ADA

1000 37

A combined QFT/H∞ Design Technique for TDOFUncertain Feedback SystemsSidi, Marcel Center of Technolog. Edu., Holon

1020 37

Computation of `1 Optimal Controllers using H2 ProjectionsRotstein, Hector RAFAEL — ADA & Technion — IITDesages, Alfredo Univ. Nacional del Sur

1040 37

(J, J0)-dissipative matrices and singular H∞ controlBaramov, Lubomır Univ. of Southampton

1100 37

On the Existence of Nash Equilibrium Solution for MixedH2/H∞ ControlFlorentino, Helenice O. UNESP, BotucatuSales, Roberto M. USP, Sao Paulo

1120 37

SVD H∞ Controller Design for an Active Horizontal SprayBoom SuspensionAnthonis, Jan K.U.LeuvenRamon, Herman K.U.Leuven

1140 37

A Linear Matrix Inequality Approach towards H∞ Control ofDescriptor SystemsRehm, Ansgar Univ. of StuttgartAllgower, Frank ETH Zurich

MA2 Alon HallDiscrete Events and Hybrid SystemsChair: Antsaklis, Panos J. Univ. of Notre DameCo-chair: Colantonio, M. C. Imperial College

1000 38

On Readily Available Supervisory Control Policies thatEnforce Liveness in a Class of Completely Controlled PetriNetsSreenivas, Ramavarapu S. Univ. of Illinois at Urbana-Ch.

1020 38

Firing Sequences Estimation for Timed Petri NetsLefebvre, Dimitri Univ. de Technologie de Belfort

1040 38

Short and Long-term Scheduling in SemiconductorManufacturingColantonio, M. C. Imperial CollegePapageorgiou, L. University College LondonShah, N. Imperial College

1100 38

Hybrid Control of a Robotic Manufacturing SystemKoutsoukos, Xenofon D. Univ. of Notre DameAntsaklis, Panos J. Univ. of Notre Dame

1120 38

Discrete-Event State Equations and Petri NetsCanuto, Enrico Politecnico di TorinoBalduzzi, Fabio Politecnico di Torino

1140 39

Stabilizing a Linear System with Finite-State Hybrid OutputFeedbackLiberzon, Daniel Yale Univ.

MA3 (I) Oren HallAutomotive Control and Energy ConversionSystemsOrganizer: Rizzo, Gianfranco Univ. of SalernoChair: Rizzo, Gianfranco Univ. of SalernoCo-chair: Dambrosio, Lorenzo Polytechnic of Bari

1000 39

Identification of Manifold Two-Phase Fuel Flow Model in aSpark Ignition Engine with Kalman Filter and Least SquareMethodsArsie, I. Univ. di SalernoPianese, C. Univ. di SalernoRizzo, Gianfranco Univ. di Salerno

19

1020 39

Optimal Idle Speed Control with Induction-to-Power FiniteDelay for SI EnginesGlielmo, Luigi Univ. di Napoli Federico IISantini, Stefania Univ. di Napoli Federico IISerra, Gabriele Magneti Marelli Engine Control Div.

1040 39

Estimator-Based Adaptive Fuzzy Logic Control Techniquefor a Wind Turbine-Induction Generator SystemDadone, Andrea Politecnico di BariDambrosio, Lorenzo Politecnico di Bari

1100 39

Active Suspension Control of Ground Vehicle Heave andPitch MotionsCampos, Javier The Univ. of Texas at ArlingtonDavis, Leo Davis Technologies Int., Inc.Lewis, Frank L. The Univ. of Texas at ArlingtonIkenaga, Scott The Univ. of Texas at ArlingtonEvans, Mark Davis Technologies Int., Inc.

1120 40

An Object-Oriented Modular Simulation Model forIntegrated Gasoline Engine and Automatic TransmissionControlHong, Keum-Shik Pusan National Univ.Yang, Kyung-Jinn Pusan National Univ.

1140 40

A Comprehensive Model for ICE Oriented to the ElectronicControl of the InjectionAnatone, Michele Univ. of L’AquilaCarapellucci, Roberto Univ. of L’AquilaCipollone, Roberto Univ. of L’AquilaSciarretta, Antonio Univ. of L’Aquila

MA4 Tomer HallLinear Systems 1Chair: Yaniv, Oded Tel-Aviv Univ.Co-chair: Kaczorek, Tadeusz Warsaw Univ. of Technology

1000 40

Stable Inversion of MIMO Linear Discrete TimeNon-Minimum Phase SystemsGeorge, Koshy Delft Univ. of TechnologyVerhaegen, Michel Delft Univ. of TechnologyScherpen, Jacquelien M.A. Delft Univ. of Technology

1020 40

State Space and Internal Models in Discrete-time LQRegulator DesignGessing, Ryszard Politechnika Slaska

1040 41

Modified Internal Model Control for Unstable SystemsYamada, Kou Yamagata Univ.

1100 41

The Wiener-Hopf Standard Control Problem: A StableFractional ApproachXie, Li Univ. of Posts and Telecom.Xue, Dingyu Northeastern Univ.

1120 41

Reduction of Singular 2D Models to Equivalent StandardModelsKaczorek, Tadeusz Warsaw Univ. of Technology

1140 41

Some New Results in Theory of ControllabilityBashirov, Agamirza Eastern Mediterranean Univ.Mahmudov, Nazim Eastern Mediterranean Univ.

MA5 Erez HallAdaptive Control 1Chair: Kosut, Robert SC Solutions Inc.Co-chair: Pait, Felipe M. Univ. de Sao Paulo

1000 41

Iterative Adaptive (Unfalsified) ControlKosut, Robert SC Solutions Inc.

1020 41

On the Design of Direct Adaptive ControllersPait, Felipe M. Univ. de Sao Paulo

1040 41

Tuning via Measurements of the Squared ErrorPait, Felipe M. Univ. de Sao Paulo

1100 42

Adaptive Generalized Predictive Control Subject to InputConstraintsKrolikowski, Andrzej Technical Univ. of Poznan

1120 42

Decentralized Adaptive Controller with Zero ResidualTracking ErrorsMirkin, Boris M. Academy of Sc. of Kyrgyz Republic

1140 42

Advanced Adaptive Control for Complex NonlinearProcessesConstantin, Nicolae Univ. Politehnica of BucharestDumitrache, Ion Univ. Politehnica of Bucharest

20

Plenary Session II Alon Hall

CONTROL OF HYBRID SYSTEMSAND SOME APPLICATIONS

Michael HeymannDept. of Electrical Engineering

Technion — IIT

Chair: Feuer, Arie Technion — IIT

MM1 Dekel HallOptimization MethodsChair: Rusnak, Ilan RAFAEL — ADACo-chair: Guez, Allon Drexel Univ.

1430 42

PI Controller Tuning via Multiobjective OptimizationKookos, I. K. Imperial CollegeArvanitis, K. G. National Tech. Univ. of AthensKalogeropoulos, G. Univ. of Athens

1450 43

Decomposition-Coordinated Optimization of Large-ScaleDiscrete Systems with Parallel-Sequential CoordinatedSchemeLychenko, Nataly M. Academy of Sc. of Kyrgyz Republic

1510 43

Generalized PID ControllerRusnak, Ilan RAFAEL — ADA

1530 43

Lagrange Problem for Non-Standard Nonlinear SingularlyPerturbed SystemsFridman, Emilia Tel-Aviv Univ.

1550 43

Optimization for Part Nesting and Layout Using aDistributed SPMD ArchitectureWetstein, Joseph P. Drexel Univ.Guez, Allon Drexel Univ.

1610 43

Computing Resources Dynamic Optimization of DigitalMultichannel Control SystemsFrid, Arkadi I. Ufa Aviation Technical Univ.Enikeev, Adel K. Ufa Aviation Technical Univ.Novikov, Boris A. Ufa Aviation Technical Univ.

MM2 Alon HallInteligent Control and Neural NetworksChair: Taylor, James H. Univ. of New BrunswickCo-chair: Anghelea, Marius Univ. of Gent

1430 44

The Basic Ideas of Neural Predictive ControlSchnitman, Leizer Aeronautics Inst. of TechnologyFontes, Adhemar de B. Bahia Federal Univ.

1450 44

A Rule-Based Neuro-Optimal Controller for NonlinearMIMO SystemsTuncay, Serhat Middle East Technical Univ.Leblebicioglu, Kemal Middle East Technical Univ.Ozgen, Canan Middle East Technical Univ.Halici, Ugur Middle East Technical Univ.

1510 44

Neural Network Based Softsensor for a Tubular ReactorAnghelea, Marius Univ. of GentDeclercq, Filip Univ. of GentDe Keyser, Robin Univ. of GentDecoster, Martin EXXON Chemical Comp.

1530 44

An Expert-Aided Implementation Interface for IndustrialProcess Control SystemsTaylor, James H. Univ. of New BrunswickChan, Cheney Univ. of New Brunswick

1550 45

A Self-Organizing Neurocontroller for Vibration SuppressionMoshou, Dimitrios K.U.LeuvenAnthonis, Jan K.U.LeuvenJancsok, Pal K.U.LeuvenRamon, Herman K.U.Leuven

MM3 Oren HallControl ApplicationsChair: Hong, Keum-Shik Pusan National Univ.Co-chair: Dayan, Yehoshua Technion — IIT

1430 45

A New Modeling of the Macpherson Suspension Systemand its Optimal Pole-Placement ControlHong, Keum-Shik Pusan National Univ.Jeon, Dong-Sub Pusan National Univ.Sohn, Hyun-Chull Pusan National Univ.

1450 45

A Numerical Algorithm for the Design of a DecentralizedController for Open-Channel NetworksSeatzu, Carla Univ. of Cagliari

21

1510 45

Flight Control Design for a Missile: An ApproximateFeedback Linearization ApproachTsourdos, Antonios Cranfield Univ.Blumel, Anna L. Cranfield Univ.White, Brian A. Cranfield Univ.

1530 46

Robust Quasi NID Current and Flux Control of an InductionMotor for Position Controlvan Duijnhoven, Marc Eindhoven Technical Univ.Błachuta, Marian J. Silesian Technical Univ.

1550 46

Contact Elimination in Mechanical Face Seals Using ActiveControlDayan, Joshua Technion — IITZou, Min SeagateGreen, Itzhak Georgia Tech

1610 46

Design, Simulation & Control of a Segmented ReflectorTest-bedMorales, Mauricio California State Univ., LAMirmirani, Majdedin California State Univ., LABoussalis, Helen California State Univ., LA

MM4 Tomer HallLinear Systems 2Chair: Kucera, Vladimır Trnka Lab. & UTIACo-chair: Ferreira, Pedro M. G. PUC-Rio

1430 46

Generalized Versions of Bode’s TheoremGera, Amos E. Elta

1450 46

On a Conjecture and the Internal ModelFerreira, Pedro M. G. PUC-Rio

1510 47

Reliable Computation of the Input-State-Output Relations inAutoregressive Representations of Multivariable SystemsKraffer, Ferdinand Inst. of Inf. Theory and Automation

1530 47

The Suboptimal Tracking Problem in Linear SystemsDostal, Petr Technical Univ. BrnoBobal, Vladimir Technical Univ. Brno

1550 47

Margins and Bandwidth Limitations of NMP SISO FeedbackSystemsSidi, Marcel Center of Technolog. Edu., HolonYaniv, Oded Tel-Aviv Univ.

1610 47

Reachability and Controllability of Positive Linear Systemswith State FeedbacksKaczorek, Tadeusz Warsaw Univ. of Technology

MM5 (I) Erez HallNonlinear Systems Identification in PracticeOrganizer: Sjoberg, Jonas E. Chalmers Univ. of TechnologyChair: Sjoberg, Jonas E. Chalmers Univ. of TechnologyCo-chair: Gutman, Per-Olof Technion — IIT

1430 47

Adaptive Hybrid Physical/Neural Network Modeling and itsApplication to Greenhouse Climate OptimizationLinker, Raphael Technion — IITSeginer, Ido Technion — IITGutman, Per-Olof Technion — IIT

1450 47

Initialization and Model Reduction for Wiener ModelIdentificationHagenblad, Anna Linkoping Univ.

1510 48

Nonlinear Identification of Automobile Vibration DynamicsWestwick, David T. Delft Univ. of TechnologyGeorge, Koshy Delft Univ. of TechnologyVerhaegen, Michel Delft Univ. of Technology

1530 48

Generalization: A Hidden Agenda in System IdentificationLarsen, Jan Technical Univ. of DenmarkHansen, Lars Kai Technical Univ. of Denmark

1550 48

Nonlinear Identification of the Position Sled Dynamics of aCD PlayerSjoberg, Jonas E. Chalmers Univ. of TechnologyGutman, Per-Olof Technion — IIT

1610 48

A Global Optimization Approach to Nonlinear SystemIdentificationTiano, A. Univ. of PaviaPizzocchero, F. Univ. of PaviaVenini, P. Univ. of Pavia

22

MP1 Dekel HallFlexible StructuresChair: Halevi, Yoram Technion — IITCo-chair: Menini, Laura Univ. di Roma Tor Vergata

1650 49

Design of a Multivariable Pole-Placement Controller for thePrimary Mirror of the 10m Grantecan TelescopeAcosta, L. La Laguna Univ.Sigut, M. La Laguna Univ.Hamilton, A. La Laguna Univ.Mendez, J. A. La Laguna Univ.Marichal, G. N. La Laguna Univ.Moreno, L. La Laguna Univ.

1710 49

Application of a Classical PD Regulator to the Control of aFlexible Planar Closed Chain LinkageGasparetto, Alessandro Univ. of UdineMiani, Stefano Univ. of Udine

1730 49

Control of Flexible Structures Using Models with Dead TimeRaskin, Natalya Technion — IITHalevi, Yoram Technion — IIT

1750 49

Balanced Realization of Flexible Structures with GeneralDamping: A Power Series ApproachHalevi, Yoram Technion — IIT

1810 49

Computation in closed form of the equations of motion for aflexible beam with lumped masses and rotational inertiasMenini, Laura Univ. di Roma Tor VergataTornambe, Antonio Univ. di Roma TreZaccarian, Luca Univ. di Roma Tor Vergata

MP2 Alon HallFault DetectionChair: Speyer, Jason L. Univ. of California, LA

1650 50

Improved Observer for Sensor Fault Diagnosis of a PowerPlantSimani, Silvio Univ. di FerraraFantuzzi, Cesare Univ. di FerraraBeghelli, Sergio Univ. di Ferrara

1710 50

Residual-Sensitive Fault Detection FilterChen, Robert H. Univ. of California, LASpeyer, Jason L. Univ. of California, LA

1730 50

Catastrophic Failure EvaluationMacdonald, John M. Los Alamos National Lab.Nekimken, Howard Los Alamos National Lab.Picard, Rick Los Alamos National Lab.Olson, Keith Los Alamos National Lab.Bates, Adam Los Alamos National Lab.Ortiz, Augustine Los Alamos National Lab.

MP3 (I) Oren HallTarget TrackingOrganizer: Bar-Shalom, Yaakov Univ. of ConnecticutChair: Bar-Shalom, Yaakov Univ. of ConnecticutCo-chair: Madan, Rabi ONR, Arlington

1650 50

Nonlinear Filters with Virtual MeasurementsDaum, Frederick E. Raytheon Co.

1710 50

System Level Performance of Radar WaveformsNiu, Ruixin Univ. of ConnecticutWillett, Peter Univ. of ConnecticutBar-Shalom, Yaakov Univ. of Connecticut

1730 50

A Radar Power Multiplier Algorithm for Acquisition of LowObservable Ballistic Missiles Using an ESA RadarSivananthan, Sivaloganathan ARCON Corp.Kirubarajan, Thiagalingam Univ. of ConnecticutBar-Shalom, Yaakov Univ. of Connecticut

1750 51

Trajectory and Launch Point Estimation for Ballistic Missilesfrom Boost Phase LOS MeasurementsLi, Yicong Comverse Network SystemsKirubarajan, Thiagalingam Univ. of ConnecticutBar-Shalom, Yaakov Univ. of Connecticut

1810 51

Artificial Neural Network Embedded Kalman Filter BearingOnly Passive Target TrackingSurendra Rao, Alladi Naval Sc. and Technological Lab.

MP5 Erez HallAdaptive Control 2Chair: Feuer, Arie Technion — IITCo-chair: Arvanitis, K. G. National Tech. Univ. of Athens

1650 51

Adaptive Pole Placement Control of Linear Systems UsingPeriodic Multirate-Input ControllersArvanitis, K. G. National Tech. Univ. of AthensKalogeropoulos, G. Univ. of Athens

23

1710 52

Development of a Self-Tuning PID Controller Based onNeural Network for Nonlinear SystemsHan, Woo-yong Jeonju Technical CollegeHan, Jin-wook Chonbuk National Univ.Lee, Chang-goo Chonbuk National Univ.

1730 52

Multi-Drug Infusion Control Using a Robust Direct AdaptiveController for Plants with Time DelaysOzcelik, Selahattin Texas A&M Univ.–KingsvillePalerm, Cesar C. Rensselaer Polytechnic Inst.Kaufman, Howard Rensselaer Polytechnic Inst.

1750 52

Indirect Adaptive Control of Drug Infusion for a CirculatorySystem ModelAchuthan, G. Rensselaer Polytechnic Inst.Alekseyenko, Y. Rensselaer Polytechnic Inst.Ishihara, A. Rensselaer Polytechnic Inst.Kaufman, Howard Rensselaer Polytechnic Inst.

1810 53

Optimal Adaptive Control of Uncertain Stochastic DiscreteLinear SystemsRusnak, Ilan RAFAEL — ADA

TUESDAY, June 29

Plenary Session III Alon Hall

OPTIMIZATION OVER LINEAR MATRIXINEQUALITIES

Stephen BoydInformation Systems Lab.

Electrical Engineering Dept.Stanford University

Chair: Rotstein, Hector RAFAEL — ADA

TA1 Dekel HallSampled-Data SystemsChair: Palmor, Zalman J. Technion — IITCo-chair: Weller, Steve Univ. of Newcastle

950 53

Improved Wiener-Hopf Method for H2-Design ofSampled-Data SystemsLampe, Bernhard P. Univ. of RostockRosenwasser, Yephim N. St.Petersburg U. Ocean Techn.

1010 53

H∞ Design of Generalized Sampling and Hold Functionswith Waveform ConstraintsKahane, Allan C. Technion — IITMirkin, Leonid Technion — IITPalmor, Zalman J. Technion — IIT

1030 53

Sampling Zeros and Robust Sampled-Data Control DesignWeller, Steven Univ. of Newcastle

1050 54

Self-Tuning PID Controller Using δ-Model IdentificationBobal, Vladimir Technical Univ. BrnoDostal, Petr Technical Univ. BrnoSysel, Martin Technical Univ. Brno

TA2 (I) Alon HallRecent Innovations in Process ControlOrganizer: Lewin, Daniel R. Technion — IITChair: Ogunnaike, Babatunde A. E.I. DuPont de NemoursCo-chair: Lewin, Daniel R. Technion — IIT

950 54

Multiple Model Control of a Pilot Distillation ColumnRodriguez, Julio A. Sydney Univ.Goodwin, Graham C. Univ. of NewcastleRomagnoli, Jose A. (“Cacho”) Sydney Univ.

24

1010 54

Online Outlier Detection and RemovalMenold, Patrick H. ETH ZurichPearson, Ronald K. ETH ZurichAllgower, Frank ETH Zurich

1030 54

Identification for Control Purposes by Relay Techniques:Achievable Performance versus ComplexityMarchetti, G. Univ. di PisaScali, Claudio Univ. di Pisa

1050 55

Algorithmic Internal Model Control of Unstable SystemsBerber, Ridvan Univ. of AnkaraBrosilow, Coleman Case Western Reserve Univ.

1110 55

Robust Stability Analysis of Nonlinear Processes UsingEmpirical State Affine Models and LMI’sBudman, Hector Univ. of WaterlooKnapp, Timothy Univ. of Waterloo

1130 55

Model Predictive Control of a Continuous GranulationProcessAdetayo, Anthony A. DuPont Central R&DPottmann, Martin DuPont DacronOgunnaike, Babatunde A. DuPont Central R&DEnnis, Brian J. E&G Associates

TA3 Oren HallStability and StabilizationChair: Keel, L. H. Tennessee State Univ.Co-chair: Johansson, Mikael Lund Inst. of Technology

950 56

Analytic Conditions for StabilizabilityKeel, L. H. Tennessee State Univ.Bhattacharyya, S. P. Texas A&M Univ.

1010 56

Stability of Dynamical Systems with ParameterPerturbationsLiberzon, Mark R. Moscow State Aviation Techn. Univ.

1030 56

BIBO Stability of NARX ModelsDzielinski, Andrzej Warsaw Univ. of Technology

1050 56

Improving Efficiency in the Computation of PiecewiseQuadratic Lyapunov FunctionsJohansson, Mikael Lund Inst. of TechnologyGhulchak, Andrey Lund Inst. of TechnologyRantzer, Anders Lund Inst. of Technology

1110 56

Practical Stability of Synchronized Chaotic Attractors and itsControlKapitaniak, Tomasz Technical Univ. of LodzCzolczynski, Krzysztof Technical Univ. of LodzBrindley, John Univ. of Leeds

1130 56

Remarks on Open-Loop Stabilizability of LinearInfinite-Dimensional Time-Varying Discrete-Time SystemsPrzyłuski, K. Maciej Polish Academy of Sc.

TA4 Tomer HallNonlinear Systems 1Chair: Fradkov, Alexander L. Russian Acad. of Sc.Co-chair: Krasnosel’skii, Alexander M.Russian Acad. of Sc.

950 57

On Oscillations in Resonant Equations with ComplexNonlinearitiesKrasnosel’skii, Alexander M. Russian Academy of Sc.

1010 57

Feedback Resonance in 1-DOF and 2-DOF NonlinearOscillatorsFradkov, Alexander L. Russian Academy of Sc.Andrievsky, Boris R. Russian Academy of Sc.

1030 57

Input-Output Models for a Class of Nonlinear Systems:Questions and AnswersKotta, Ulle Tallinn Technical Univ.

1050 57

Aspects of Traction ControlFriedland, Bernard New Jersey Inst. of Technology

1110 57

Energy Control of Hamiltonian Systems under DisturbancesPolushin, Ilya G. Russian Academy of Sc.Fradkov, Alexander L. Russian Academy of Sc.

1130 57

Nonlinear Systems Admitting Hybrid Feedback ControlStabilizationLitsyn, Elena The College of Judea and SamariaNepomnyashchikh, Yurii V. Perm State Univ.Ponosov, Arcady Inst. for Matematiske Fag

25

TA5 Erez HallFilteringChair: Colaneri, Patrizio Politecnico di MilanoCo-chair: Yaesh, I. Taas Israel Industries

950 58

Robust H∞ Filtering of Stationary Discrete-Time LinearSystems with Stochastic UncertaintiesGershon, E. Tel-Aviv Univ.Shaked, Uri Tel-Aviv Univ.Yaesh, I. Taas Israel Industries

1010 58

The J-Spectral Interactor Matrix in the Discrete-TimeSingular H∞ Filtering ProblemColaneri, Patrizio Politecnico di MilanoMaroni, Massimo Politecnico di Milano

1030 58

Kalman Bucy Filtering for Singular Output-NoiseCovarianceCarravetta, Francesco CNR-IASIGermani, Alfredo L’Aquila Univ.Manes, Costanzo L’Aquila Univ.

1050 58

On the Feasibility and Convergence of H∞ MultistepPredictorsMaroni, Massimo Politecnico di MilanoBolzern, Paolo Politecnico di Milano

1110 58

Nonlinear Observers for a Class of Differential DelaySystemsAggoune, Woihida Univ. Henri Poincare, INRIA CONGEDarouach, Mohamed Univ. Henri Poincare

TM1 (I) Dekel HallControl of Distributed Parameter SystemsOrganizer: Reich, Simeon Technion — IITOrganizer: Demetriou, Michael A. Worcester Poly. Inst.Chair: Demetriou, Michael A. Worcester Poly. Inst.Co-chair: Reich, Simeon Technion — IIT

1310 59

Boundary Control of the Korteweg–de Vries–BurgersEquation: Further Results on Stabilization and NumericalDemonstrationBalogh, Andras Univ. of California, San DiegoKrstic, Miroslav Univ. of California, San Diego

1330 59

Finite Horizon H∞ Control of Systems with State DelaysFridman, Emilia Tel-Aviv Univ.Shaked, Uri Tel-Aviv Univ.

1350 59

Numerical Criterion for Stabilizing Steady State Solutions ofthe Navier-Stokes EquationsTiti, Edriss S. Univ. of California, IrvineCao, Chongsheng Univ. of California, IrvineKevrekidis, Yannis Univ. of California, Irvine

1410 59

Lax-Phillips Scattering and Well-Posed Linear SystemsStaffans, Olof J. Abo Akademi Univ.

1430 59

Identification and Adaptive Control of Some StochasticDistributed Parameter SystemsPasik-Duncan, Bozenna Univ. of Kansas

TM2 (I) Alon HallIntegration of Process Design and Process ControlOrganizer: Lewin, Daniel R. Technion — IITChair: Lewin, Daniel R. Technion — IITCo-chair: Ogunnaike, Babatunde A. E.I. DuPont deNemours

1310 60

Interaction of Design and ControlLewin, Daniel R. Technion — IIT

1330 60

Simultaneous Process Design and Process Control:Application to Complex Separation SystemsRoss, Roderick Imperial CollegeBansal, Vikrant Imperial CollegePerkins, John D. Imperial CollegePistikopoulos, Efstratios N. Imperial College

1350 60

Process Design with Complex NonlinearitiesSeider, Warren D. Univ. of Pennsylvania

1410 60

Towards Integration of Controllability into Plant DesignJørgensen, S. Bay Technical Univ. of DenmarkGani, R. Technical Univ. of DenmarkAndersen, T. R. Technical Univ. of Denmark

1430 61

Controllability and Resiliency Analysis for a Heat-IntegratedC3-SplitterSolovyev, Boris M. Technion — IITLewin, Daniel R. Technion — IIT

26

1450 61

On the Generation of the Most Promising Control Structurefor Large Dimensional SystemsKookos, I. K. Imperial CollegeArvanitis, K. G. National Tech. Univ. of AthensKalogeropoulos, G. Univ. of Athens

TM3 (I) Oren HallAdvances to Meet the Missile Guidance Challengeat the Verge of the New MillenniumOrganizer: Shinar, Josef Technion — IITOrganizer: Ben-Asher, Josef Z. Technion — IITOrganizer: Gurfil, Pini RAFAEL — ADAChair: Davidovitz, Avraham RAFAEL — ADA

1310 61

Integrated Design of Agile Missile Guidance and ControlSystemsMenon, P. K. Optimal Synthesis Inc.Ohlmeyer, E. J. Naval Surface Warfare Center

1330 62

Optimal Guidance with Time Delay for Continuous TimeSystemsGitizadeh, R. Taas Israel IndustriesYaesh, I. Taas Israel IndustriesBen-Asher, Josef Z. Technion — IIT

1350 62

Optimal Guidance Laws with Uncertain Time-of-FlightRusnak, Ilan RAFAEL — ADA

1410 62

Design of Non-Saturating Guidance SystemsGurfil, Pini RAFAEL — ADAJodorkovsky, Mario RAFAEL — ADAGuelman, Moshe Technion — IIT

1430 62

Robust Missile Guidance Law against Highly ManeuveringTargetsShinar, Josef Technion — IITShima, Tal Technion — IIT

TM4 Tomer HallNonlinear Systems 2Chair: Friedland, Bernard New Jersey Inst. of TechnologyCo-chair: Kotta, Ulle Tallin Technical Univ.

1310 63

From Physical Realizations to Nonlinear Stability, Passivityand OptimalityMargaliot, Michael Tel-Aviv Univ.Langholz, Gideon Tel-Aviv Univ.

1330 63

Nonlinear State Estimation for Rigid Body Motion withLow-Pass SensorsRehbinder, Henrik KTH, SwedenHu, Xiaoming KTH, Sweden

1350 63

Control Systems with Actuator Saturation and Bifurcationsat InfinityPonce, Enrique Univ. SevillaAracil, Javier Univ. SevillaPagano, Daniel Univ. Federal de Santa Catarina

1410 63

An Antiwindup Control Using µ-SynthesisLu, E. Univ. Bordeaux IBergeon, B. Univ. Bordeaux IYgorra, S. Univ. Bordeaux I

1430 63

A High Gain Observer for Robust State Feedback ControllerUcar, Ahmet Firat Univ.

1450 63

On the Role of Invariance in the Theory of Systems andControl — An Intelligible Introduction for the BeginnersShima, Masasuke Hokkaido Univ.

TM5 Erez HallIdentification 1Chair: Pachter, Meir Air Force Inst. of TechnologyCo-chair: Ninness, Brett Univ. of Newcastle

1310 64

A Directional Forgetting Algorithm Based on theDecomposition of the Information MatrixCao, Liyu Carleton Univ.Schwartz, Howard M. Carleton Univ.

1330 64

A Parameter Estimation Method for a Special Class ofSystems of Ordinary Differential EquationsSeatzu, Carla Univ. of Cagliari

1350 64

An Algorithm for Control System Loop Gain IdentificationPachter, Meir Air Force Inst. of Technology

1410 64

Real-Time Identification Using a Classical NonlinearOptimization Algorithm and the Flatness Properties of aSystem: Application to an Intensity/Pressure ConverterSanchez, Augustin National Inst. Appl. Sc., ToulouseMahout, Vincent National Inst. Appl. Sc., Toulouse

27

1430 64

Stopping of Algorithms and Faults Detection in KalmanFilter ApplicationHajiyev, Chingiz Istanbul Technical Univ.

1450 65

Estimation Variance is not Model Structure IndependentNinness, Brett Univ. of NewcastleHjalmarsson, Hakan Royal Inst. of TechnolgyGustafsson, Fredrik Linkoping Univ.

TP1 (I) Dekel HallDynamical ModelsOrganizer: Reich, Simeon Technion — IITOrganizer: Demetriou, Michael A. Worcester Poly. Inst.Chair: Ackleh, Azmy S. Univ. of Southwestern LouisianaCo-chair: Fridman, Emilia Tel-Aviv Univ.

1530 65

Stability, Euler Approximations of Dynamical Systems andFixed Point IterationsFarkhi, Elza Tel-Aviv Univ.

1550 65

Asymptotic Behavior of Infinite Products ofOrder-Preserving Mappings in Banach SpaceReich, Simeon Technion — IITZaslavski, Alexander J. Technion — IIT

1610 66

Exponential Stabilization of Vibrating Systems byCollocated FeedbackWeiss, George Imperial CollegeCurtain, Ruth F. Univ. of Groningen

1630 66

A Composite Semigroup for the Infinite-DimensionalDifferential Sylvester EquationEmirsajlow, Zbigniew Technical Univ. of Szczecin

1650 66

Input-Output Stability of Systems Governed by NonlinearSecond Order Evolution Equations in Hilbert SpacesGil’, Michael I. Ben-Gurion Univ.

TP2 Alon HallProcess ControlChair: Dumont, Guy A. Univ. of British ColumbiaCo-chair: Oisiovici, Ronia M. Univ. Estadual de Campinas

1530 66

Wood Chip Refiner ControlIsmail, Ahmed A. Univ. of British ColumbiaDumont, Guy A. Univ. of British Columbia

1550 66

Automatic Tuning of the Window Size in the Box CarBackslope Data Compression AlgorithmPettersson, Jens Royal Inst. of TechnologyGutman, Per-Olof Technion — IIT

1610 67

Experimental Tests of Digital Filters for Control of aPilot-Scale Batch Distillation ColumnOisiovici, Ronia M. Univ. Estadual de CampinasCruz, Sandra L. Univ. Estadual de CampinasPereira, Joao A. F. R. Univ. Estadual de Campinas

1630 67

A Linear Time-Varying State-Space Model of BatchDistillation Columns for Control ApplicationsOisiovici, Ronia M. Univ. Estadual de CampinasCruz, Sandra L. Univ. Estadual de Campinas

1650 67

Neuro-Fuzzy Modeling in Petrochemical IndustryBucolo, Maide Univ. di CataniaGraziani, Salvatore Univ. di CataniaFortuna, Luigi Univ. di CataniaSinatra, Mario ERG Petroli Siracusa

TP3 Oren HallAerospace ControlChair: Ben-Asher, Josef Z. Technion — IITCo-chair: Idan, Moshe Technion — IIT

1530 67

An Integrated Algorithm for Path Planning and FlightController Scheduling for Autonomous HelicoptersEgerstedt, Magnus Royal Inst. of TechnologyKoo, T. K. John Univ. of California, BerkeleyHoffmann, Frank Univ. of California, BerkeleySastry, Shankar Univ. of California, Berkeley

1550 67

Actuator Design for Aircraft Robustness Versus Category IIPIOAmato, Francesco Univ. of NaplesIervolino, Raffaele Univ. of NaplesScala, Stefano Italian Aerospace Research CentreVerde, Leopoldo Italian Aerospace Research Centre

1610 68

On Algorithms for Attitude Estimation Using GPSBar-Itzhack, Itzhack Y. Technion — IITNadler, Assaf Technion — IIT

28

TP4 Tomer HallTime Delay SystemsChair: Mirkin, Leonid Technion — IITCo-chair: Yaniv, Oded Tel-Aviv Univ.

1530 68

Every Stabilizing Dead-Time Controller has anObserver-Predictor-Based StructureMirkin, Leonid Technion — IITRaskin, Natalya Technion — IIT

1550 68

The Structure at Infinity of Linear Delay Systems and theRow-by-Row Decoupling ProblemRabah, Rabah Inst. de Rech. en Cyb. de NantesMalabre, Michel Inst. de Rech. en Cyb. de Nantes

1610 69

Stabilization of Singularly Perturbed Linearly Systems withDelay and Saturating ControlIonita, Achim National Inst. of Aerospace Res.Dragan, Vasile Romanian Academy

1630 69

Near Optimal PLL Design for Decision Feedback Carrierand Timing RecoveryYaniv, Oded Tel-Aviv Univ.Raphaeli, Dan Tel-Aviv Univ.

TP5 Erez HallIdentification 2Chair: Balakrishnan, S. N. Univ. of Missouri-RollaCo-chair: Nagurka, Mark Marquette Univ.

1530 69

Modelling and Identification of a High Temperature ShortTime Pasteurization Process Including DelaysAlastruey, Carlos F. Public Univ. of NavarraDe la Sen, Manuel Univ. of the Basque CountryGarcia-Sanz, Mario Public Univ. of Navarra

1550 69

Parameter Identification In Nonlineaer Systems UsingHopfield Neural NetworksHu, Zhenning Univ. of Missouri-RollaBalakrishnan, S. N. Univ. of Missouri-Rolla

1610 69

Optimal Combination of Identification and Control forBounded-Noise ARX SystemsKrolikowski, Andrzej Technical Univ. of Poznan

1630 69

Closed-loop model-free subspace-based LQG-designFavoreel, Wouter Katholieke Univ. LeuvenDe Moor, Bart Katholieke Univ. LeuvenGevers, Michel Univ. Catholique de LouvainVan Overschee, Peter Katholieke Univ. Leuven

1650 70

Measurement of Impedance Characteristics of ComputerKeyboard KeysNagurka, Mark Marquette Univ.Marklin, Richard Marquette Univ.

29

WEDNESDAY, June 30

Plenary Session IV Alon Hall

TARGET TRACKING AND DATA FUSION:How to Get the Most Out of Your Sensors

Yaakov Bar-ShalomEstimation and Signal Processing Lab.

Electrical and Systems Engineering Dept.University of Connecticut

Chair: Mirkin, Leonid Technion — IIT

WA1 (I) Dekel HallOptimal ControlOrganizer: Reich, Simeon Technion — IITOrganizer: Demetriou, Michael A. Worcester Poly. Inst.Chair: Malanowski, Kazimierz Polish Academy of Sc.Co-chair: Ackleh, Azmy S. Univ. Southwestern Louisiana

1000 70

Structure of Optimal Solutions of Infinite DimensionalControl ProblemsZaslavski, Alexander J. Technion — IIT

1020 70

Lipschitz Stability of Solutions to Parametric OptimalControl for Parabolic EquationsMalanowski, Kazimierz Polish Academy of Sc.Troltzsch, Fredi Technische Univ. Chemnitz-Zwickau

1040 71

Optimal Control of Differential Inclusions Involving PartialDifferential OperatorsIoffe, Alexander Technion — IIT

1100 71

On the Existence of Optimal Strategies for MultichainMarkov Decision ProcessesLeizarowitz, Arie Technion — IIT

1120 71

Feedback Control for Descriptor SystemsKurina, Galina A. Voronezh State Forestry Academy

WA2 Alon HallManufacturingChair: Fatikow, Sergej Univ. of KarlsruheCo-chair: Dahan, Marc Inst. de Productique, Besancon

1000 71

Optimal Design of Transfer Lines and MultipositionMachinesDolgui, Alexandre Univ. of Technology of TroyesGuschinsky, Nikolai N. Academy of Sc. of BelarusLevin, Genrikh M. Academy of Sc. of Belarus

1020 72

Conrol Architecture of a Flexible Microrobot-BasedMicroassembly StationFatikow, Sergej Univ. of KarlsruheSeyfried, J. Univ. of Karlsruhe

1040 72

The Relationship between Planning and ProductionActivities in Process IndustriesJovan, Vladimir Jozef Stefan Inst.

1100 72

Strategies for Integrating Preparation andRealisation — The Case of Product ModelsLarsen, Michael Holm The Technical Univ. of DenmarkKirkby, Phillip The Technical Univ. of DenmarkVesterager, Johan The Technical Univ. of Denmark

1120 72

Aided Decision and Authentication of Lamellated WoodFrameworksImberdis, Claude Inst. de productique, BesanconGendreau, Dominique IUFM de Franche ComteDahan, Marc Inst. de productique, Besancon

WA3 Oren HallRobotics 1Chair: Ailon, Amit Ben-Gurion Univ.Co-chair: Michalska, Hannah McGill Univ.

1000 72

An Estimate to the Energy Function of a Rigid Robot with aStabilizing PD ControllerAilon, Amit Ben-Gurion Univ.Gil’, Michael I. Ben-Gurion Univ.

1020 72

Hierarchical Fuzzy Behavior-Based Control of a Multi-AgentRobotic SystemBerman, Sigal Ben-Gurion Univ.de Oliveira, Marco A. A. The Univ. of New MexicoEdan, Yael Ben-Gurion Univ.Jamshidi, Mohammad The Univ. of New Mexico

30

1040 73

Geometric and System Decomposition Techniques inApplication to Control of a Mobile Robot with TrailerMichalska, Hannah McGill Univ.

1100 73

Following a Path of Varying Curvature as an OutputRegulation ProblemAltafini, Claudio Royal Inst. of Technology

1120 73

On Enhancing GJK Algorithm for Distance ComputationBetween Convex Polyhedra: Comparison of ImprovementsShiang, Shen-Po National Taiwan Univ.Chien, Yu-ren National Taiwan Univ.Liu, Jing-Sin Academia Sinica

WA4 Tomer HallRobust ControlChair: Zeheb, Ezra Technion — IITCo-chair: Rodrıguez-Palacios, Alejandro Cenidet

1000 73

Links Between Robust and Quadratic Stability of UncertainDiscrete-Time PolynomialsHenrion, Didier LAAS-CNRS, ToulouseSebek, Michael Trnka Lab. & UTIAKucera, Vladimır Trnka Lab. & UTIA

1020 73

Development of the Modal Regulator Design Method for aPlant with Interval ParametersSmagina, Yelena IsraelBrewer, Irina USA

1040 74

Robust Stability Condition for the System with FeedbackConnected Uncertainty and Uncertain Number of UnstablePolesYamada, Kou Yamagata Univ.

1100 74

Real and Complex Stability Radii in AutomaticLoad-Frequency Control Systems via LQG/LTR and LMITerra, Marco H. Univ. of Sao PauloMasca, Gregoria M. T. Univ. of Sao Paulo

1120 74

Robust Control for a Class of Linear Infinite DimensionalSystems with Multiplicative DisturbancesRodrıguez-Palacios, Alejandro CenidetFernandez-Anaya, Guillermo UIA

1140 74

About Some Interconnection Between LTR and RPISde Larminat, Philippe Inst. de Rech. en Cyb. de NantesLebret, Guy Inst. de Rech. en Cyb. de NantesPuren, Sophie Ing. Pour Signaux et Systemes

WA5 Erez HallSignal and Image ProcessingChair: Shimkin, Nachum Technion — IIT

1000 74

A Novel Architecture for Digital Pulse Height Analysis withApplication to Radiation SpectroscopyElhanany, Itamar Nuclear Research Center NegevJacobi, Shimshon Nuclear Research Center NegevKahane, Michael Nuclear Research Center NegevMarcus, Eli Nuclear Research Center NegevTirosh, Dan Nuclear Research Center NegevBarak, Dov Nuclear Research Center Negev

1020 75

Real-Time Adaptive Filtering for Nonstationary ImageRestoration Using Gaussian InputAbilov, Abdulriza Ankara Univ.Tuzunalp, Onder Ankara Univ.Telatar, Ziya Ankara Univ.

1040 75

The Edge Point Detection Problem in Image Sequences:Definition and Comparative Evaluation of Some 3D EdgeDetecting SchemesJetto, L. Univ. of AnconaOrlando, G. Univ. of AnconaSanfilippo, A. Univ. of Ancona

Plenary Session V Alon Hall

DEEP SPACE CONTROL CHALLENGES OF THENEW MILLENNIUM

David S. Bayard & Garry M. BurdickJet Propulsion Laboratory

California Institute of Technology

Chair: Gutman, Per-Olof Technion — IIT

WM1 (I) Dekel HallParameter EstimationOrganizer: Reich, Simeon Technion — IITOrganizer: Demetriou, Michael A. Worcester Poly. Inst.Chair: Reich, Simeon Technion — IITCo-chair: Demetriou, Michael A. Worcester Poly. Inst.

1430 75

Model-Based Detection Observer of Component Failuresfor Distributed Parameter SystemsDemetriou, Michael A. Worcester Polytechnic Instit.

31

1450 75

Parameter Estimation Problem for a Nonlinear ParabolicEquation with a Singular Nonlocal Diffusion TermAckleh, Azmy S. Univ. of Southwestern Louisiana

1510 75

Parameter Identification in a Nonautononous NonlinearVolterra Integral EquationAckleh, Azmy S. Univ. Southwestern LouisianaAizicovici, Sergiu Ohio Univ.Ferdinand, Robert R. East Central Univ.Reich, Simeon Technion — IIT

1530 75

Adaptive Control of a Time-Varying Parabolic System:Averaging AnalysisHong, Keum-Shik Pusan National Univ.Solo, Victor Macquarie Univ.Bentsman, Joseph Univ. of Illinois at Urbana

1550 75

Approximation of High-Order Lumped Systems by usingNon-Integer Order Transfer FunctionsFortuna, Luigi Univ. di CataniaGraziani, Salvatore Univ. di CataniaMuscato, Giovanni Univ. di CataniaNunnari, Giuseppe Univ. di CataniaPorto, Domenico Univ. di Catania

WM2 Alon HallFuzzy Logic MethodsChair: Jamshidi, Mohammad The Univ. of New Mexico

1430 76

Closed-Loop Robust Controllers with Fuzzy GainScheduling for FNS Assisted Walking of ParaplegicsMoulin, Mark Technion — IITInbar, Gideon F. Technion — IIT

1450 76

Mathematical Formulation of Fuzzy Cognitive MapsStylios, Chrysostomos D. Univ. of PatrasGroumpos, Peter P. Univ. of Patras

1510 76

An Outline for a Universal Logic System: A Logic System inEight Truth ValuesHeald, Graeme RMIT Univ.

1530 76

Using Soft Computing Methodologies for MultistageSupervisory Control of Complex SystemsStylios, Chrysostomos D. Univ. of PatrasChristova, Nikolinka Univ. of PatrasGroumpos, Peter P. Univ. of Patras

WM3 Oren HallRobotics 2Chair: Berman, Nadav Ben-Gurion Univ.Co-chair: Schwartz, Howard Carleton Univ.

1430 77

Variable Structure Control with Varying Bounds of RobotManipulatorsNigrowsky, Pierre M. B. Brunel Univ.Turner, Peter J. Brunel Univ.

1450 77

An MRAC Output Feedback Controller for RobotManipulatorsSchwartz, Howard M. Carleton Univ.

1510 77

Basic Fairing Principles of Fiberglass Pits and PatchesOliver, Glen C. Univ. of Texas at ArlingtonShiakolas, Panayiotis S. Univ. of Texas at ArlingtonLawley, Tommy J. Univ. of Texas at Arlington

1530 77

RobSurf: A Near Real Time OLP System for RoboticSurface FinishingShiakolas, Panayiotis S. Univ. of Texas at ArlingtonLabalo, Dragan Univ. of Texas at ArlingtonFitzgerald, J. Mick Univ. of Texas at Arlington

1550 77

Adaptive Nonlinear Visual Servoing Using Lyapunov-BasedDesignConticelli, Fabio Scuola Superiore Sant’AnnaAllotta, Benedetto Scuola Superiore Sant’Anna

1610 78

On the Inclusion of Robot Dynamics in Visual ServoingSystemsConticelli, Fabio Scuola Superiore Sant’AnnaAllotta, Benedetto Scuola Superiore Sant’Anna

WM4 (I) Tomer HallSliding Mode ControlOrganizer: Zinober, Alan S. I. The Univ. of SheffieldChair: Zinober, Alan S. I. The Univ. of Sheffield

1430 78

Partial Lipschitz Nonlinear Sliding Mode ObserversKoshkouei, Ali J. Univ. of SheffieldZinober, Alan S. I. Univ. of Sheffield

32

1450 78

Dynamical Adaptive First and Second Order Sliding ModeControl of Nonlinear Non-Triangular Uncertain SystemsZinober, Alan S. I. Univ. of SheffieldScarratt, Julie C. Univ. of SheffieldFerrara, Antonella Univ. of PaviaGiacomini, Luisa Aston Univ.Rios-Bolıvar, Miguel Univ. de Los Andes

1510 78

Adaptive Sliding Backstepping Control of NonlinearSemi-Strict Feedback Form SystemsKoshkouei, Ali J. Univ. of SheffieldZinober, Alan S. I. Univ. of Sheffield

1530 79

A Feedforward-Feedback Interpretation of a Sliding ModeControl LawMonsees, Govert Delft Univ. of TechnologyGeorge, Koshy Delft Univ. of TechnologyScherpen, Jacquelien M.A. Delft Univ. of TechnologyVerhaegen, Michel Delft Univ. of Technology

1550 79

Nonminimum Phase Output Tracking via Sliding ModeControl: Stable System Center TechniqueShkolnikov, Ilya A. The Univ. of Alabama in HuntsvilleShtessel, Yuri B. The Univ. of Alabama in Huntsville

1610 79

2-Sliding Mode with AdaptationBartolini, Giorgio Univ. of CagliaryLevant, Arie Inst. for Industrial MathematicsPisano, Alessandro Univ. of CagliaryUsai, Elio Univ. of Cagliary

WM5 Erez HallComputer Networks and Queing SystemsChair: Shimkin, Nachum Technion — IIT

1430 79

The Optimal Markov Strategy for Access in ISDNs withReserves of ChannelsMelikov, Agassy Z. International American Univ.Deniz, Dervis Z. Eastern Mediterranean Univ.

1450 79

Tbit/sec Switching Scheme for ATM/WDM High-SpeedComputer NetworksElhanany, Itamar Ben-Gurion Univ.Sadot, Dan Ben-Gurion Univ.

1510 80

Grid-based ATM Switch Architecture: A New Fault-TolerantSpace-Division Switch Fabric ArchitectureLaskaridis, Haralampos S. Aristotle Univ. of ThessalonikiVeglis, Andreas A. Aristotle Univ. of ThessalonikiPapadimitriou, Georgios I. Aristotle Univ. of ThessalonikiPomportsis, Andreas S. Aristotle Univ. of Thessaloniki

33

34

MED99

Abstracts

MONDAY, June 28

MA1Optimal Control: H∞/H2/`1

MA1-1 1000

A combined QFT/H∞ Design Technique for TDOFUncertain Feedback Systems

Marcel Sidi Center of Technolog. Edu., Holon

Summary: The present paper presents a way to incorporateQFT principles to the H∞ control design technique to solvethe TDOF Feedback Problem with Highly Uncertain Plants.The proposed design procedure is illustrated with SISO andMIMO design examples for highly uncertain plants.

Keywords: QFT, H∞, uncertain plant, sensitivity.

MA1-2 1020

Computation of `1 Optimal Controllers using H2Projections

Hector Rotstein RAFAEL — ADA & Technion — IITAlfredo Desages Univ. Nacional del Sur

Summary: Although the `1 or peak-to-peak norm can beused to capture a number of desirable closed-loop specifica-tions, few practical applications of the criterion have beenreported to date. Arguably, the single main reason for this isthat, unlike the situation with the closeH2 andH∞ relatives,efficient numerical algorithms for `1 are still not available.The purpose of the present paper is to present an algorithmfor computing sub-optimal `1 controllers using sequentialH2 projections. As opposed to previous approaches, the al-gorithm does not use interpolations constraints nor attemptsto solve an infinite optimization problem via finite approxi-mation. Instead, sequential projections onto convex sets areperformed to decide whether a given sub-optimal `1-normlevel can be achieved or not. The present algorithm has sev-eral key advantages over previous methods:

1. At each stage, a finite optimization problem must besolved. This finite dimensionality is not due to trunca-tion but results from the exact application of the algo-rithm.

2. The finite optimization problems are H2 projectionsand can be solved efficiently.

3. The approach does not rely on interpolation con-straints. The same algorithm that works for the sim-plest version of the `1 problem (e.g., 1-block, SISO), canbe modified in a straightforward manner to yield a so-lution to the general linear time-invariant case (e.g., 4-block, MIMO).

MA1-3 1040

(J, J0)-dissipative matrices and singular H∞ control

Lubomır Baramov Univ. of Southampton

Summary: This paper deals with a class ofH∞ control prob-

lems where the transfer matrix from the external input tothe measured output is invertible on the imaginary axis in-cluding infinity whil e there is no assumption about the infi-nite and/or imaginary-axis zeros of the transfer matrix fromthe control input to the penalized output. Our approachis based on the chain-scattering representation and a newlyproposed (J, J0)-dissipative factorization extending thus thewell-known approach of H. Kimura, while preserving itssimplicity. We provide also a characterization o f the set ofcontrollers solving the given problem.

Keywords: Singular H∞ control, (J, J0)-dissipative matri-ces, controller parametrization, chain-scattering representa-tion.

MA1-4 1100

On the Existence of Nash Equilibrium Solution forMixed H2/H∞ Control

Helenice O. Florentino UNESP, BotucatuRoberto M. Sales USP, Sao Paulo

Summary: The aim of this work is the application of thegame theory in mixed H2/H∞ control problems, using con-vex optimization. We use the formulation of the mixedH2/H∞ control problem as a Nonzero-Sum NASH Game,where the two pay-off functions are associated with twoplayers, which represent the H2 and H∞ criteria. We showthat the necessary and sufficient conditions for the existenceof a NASH equilibrium solution are related to the existenceof a global optimal solution to a convex optimization prob-lem. The plant is assumed linear and time-invariant and theresulting controller is a state-feedback law.

Keywords: Nash Game, H2/H∞ Control, Convex Opti-mization.

MA1-5 1120

SVD H∞ Controller Design for an Active HorizontalSpray Boom Suspension

Jan Anthonis K.U.LeuvenHerman Ramon K.U.Leuven

Summary: An active suspension, acting as a band stop fil-ter, to reduce the horizontal motions of an agricultural sprayboom, is designed. Because the translational and the ro-tational behaviour of the system can be separated, a superoptimal SVD H∞ controller is achieved by two single SISOdesigns. Black box frequency domain identification meth-ods render continuous models. Accelerometer drift, perfor-mance and robustness issues are tackled. The final activehorizontal suspension is validated on a commercial avail-able spray boom.

Keywords: SVD H∞ controller, black box frequency do-main identification, active suspension, spray boom.

37

MA1-6 1140

A Linear Matrix Inequality Approach towards H∞Control of Descriptor Systems

Ansgar Rehm Univ. of StuttgartFrank Allgower ETH Zurich

Summary: In this paperH∞ control of high index and non-regular linear descriptor systems is addressed. Based on ageneralization of the bounded real lemma (BRL) to indexone systems, all controllers solving the H∞ control problemcan be characterized via biaffine matrix inequalities (BMIs).These inequalities imply a certain structure of candidate ma-trix solutions. Making use of this structure, standard linearalgebra tools can be used in order to show the equivalenceof the BMI synthesis conditions to a numerically appealingcharacterization of the solution of the H∞ control problemvia linear matrix inequalities (LMIs). We also address thecomputation of full- and reduced order controllers.

Keywords: State-space H∞-control, singular systems, Lin-ear Matrix Inequalities.

MA2Discrete Events and Hybrid Systems

MA2-1 1000

On Readily Available Supervisory Control Policiesthat Enforce Liveness in a Class of CompletelyControlled Petri Nets

Ramavarapu S. Sreenivas Univ. of Illinois at Urbana-Ch.

Summary: A Petri Net (PN) is said to be live if it is pos-sible to fire any transition from every reachable marking,although not necessarily immediately. Under appropriateconditions, a non-live PN can be made live via supervision.Under this paradigm an external-agent, the supervisor, pre-vents the firing of certain transitions at each reachable mark-ing so as to enforce liveness. A PN is said to be completelycontrollable if the supervisor can prevent the firing of anytransition. Testing the existence of a supervisory policy thatenforces liveness in a completely controlled Petri net can becomputationally expensive. In this paper we present a newclass of PNs for which there is a readily available supervi-sory policy that enforces liveness. This observation obviatesthe aforementioned test for the specific class of PNs intro-duced in this paper.

Keywords: DEDS, supervisory control, Petri nets, liveness.

MA2-2 1020

Firing Sequences Estimation for Timed Petri Nets

Dimitri Lefebvre Univ. de Technologie de Belfort

Summary: This work deals with the firing sequences esti-mation for transitions - timed Petri nets by measurement ofthe places marking. Firing durations are unknown, but sup-posed not to be null. In fact, the Petri net marking is mea-sured, on line, with a sampling period Dt small enough suchthat each transition is fired, at the most, one time during Dt.

The estimation problem has exact and approximated solu-tions that are described. Sufficient conditions are given onthe accuracy of the marking measurement, such that the es-timation of the firing sequences is an exact one. If the esti-mation provides several solutions, the Petri net is completedin order to give a unique solution.

Keywords: Timed Petri nets, manufacturing systems, esti-mation, firing sequences, Moore-Penrose inverse.

MA2-3 1040

Short and Long-term Scheduling in SemiconductorManufacturing

M. C. Colantonio Imperial CollegeL. Papageorgiou University College LondonN. Shah Imperial College

Summary: This paper addresses the scheduling problemin semiconductor manufacturing. A two level hierarchicalstructure is considered to take into account different hori-zons in the decision making process. Long-term planning issolved by means of an `1-norm Model Predictive Controllerwhich gives the release policy to a short-term controller. Thelatter is based on a State-Task-Network representation of thebatch recipe and provides the detailed operation of the fab.

Keywords: Semiconductor fabs, discrete-events, planning,scheduling, optimization.

MA2-4 1100

Hybrid Control of a Robotic Manufacturing System

Xenofon D. Koutsoukos Univ. of Notre DamePanos J. Antsaklis Univ. of Notre Dame

Summary: In this paper, a new approach for control of hy-brid systems is introduced and illustrated using a roboticmanufacturing system. Hybrid systems, which are used tomodel the physical process, and their controllers are viewedas system components of an intelligent control frameworkand they are modeled as set-dynamical systems. The centralconcept studied in hybrid system modeling is quasideter-minism and it is used to address the problems that arise be-cause of the nondeterministic nature of the discrete approx-imations of the continuous dynamics. Decision algorithmsare derived based on supervisory control of discrete-eventsystems described by Petri nets. The notions of abstractionand multirate time scales play an important role in the de-sign of decision algorithms to supervise the operation of thesystem.

Keywords: Hybrid control systems, quasideterminism,Petri nets.

MA2-5 1120

Discrete-Event State Equations and Petri Nets

Enrico Canuto Politecnico di TorinoFabio Balduzzi Politecnico di Torino

Summary: In this paper we present a novel formulation forthe modeling and control of discrete event dynamic systems.This original approach leads to a discrete-event state equa-tions formulation satisfying Kalman axioms, where the stateis defined as the sequence of potential events (enabled tran-

38

sitions in terms of Petri net language) forced by the occur-rence of state events (free evolution) or by arbitrary inputevents (forced evolution). The proposed formulation is con-sidered to be very general and appropriate to any discreteevent systems. This conviction is supported by the analysisperformed by comparing discrete-event state equations withclassical discrete-event models like untimed and timed Petrinets, finite-state timed and untimed automata. We showthat all these models can be formulated as a sub-class of thediscrete-event state equations.

Keywords: State equations, discrete-event, Petri Nets, Au-tomata.

MA2-6 1140

Stabilizing a Linear System with Finite-State HybridOutput Feedback

Daniel Liberzon Yale Univ.

Summary: The purpose of this short note is to establishand explore a link between the problem of stabilizing a lin-ear system using finite-state hybrid output feedback and theproblem of finding a stabilizing switching sequence for aswitched linear system with unstable individual matrices,each of which separately has recently received attention inthe literature.

Keywords: Switched linear system, finite-state hybrid out-put feedback.

MA3 (I)Automotive Control and EnergyConversion Systems

MA3-1 1000

Identification of Manifold Two-Phase Fuel Flow Modelin a Spark Ignition Engine with Kalman Filter andLeast Square Methods

I. Arsie Univ. di SalernoC. Pianese Univ. di SalernoGianfranco Rizzo Univ. di Salerno

Summary: In this paper the identification of the two-phasefuel flow model in the intake manifold for a spark ignitionengine is approached. The dynamic model is part of an inte-grated system of models with hierarchical structure, rangingfrom phenomenological to neural network approaches, forthe analysis and the optimal design of engine control strate-gies in automotive engines, which is actually in use by a ma-jor automotive supplier. Two different techniques for the es-timation of model parameters are compared: (i) a classicalleast square method and (ii) the Kalman filtering approach.The former approach requires a set of off-line identificationsperformed through the generation of air-fuel ratio transientsfor each engine operating condition. Model parameters arethen identified via inverse modelling approach using nonlinear least square techniques and stored in a look-up tablein the ECU. The second technique consists in the design ofa non-linear observer based on an extended Kalman filter.This latter approach can be applicable in on-line operationsin order to estimate both states and parameters of the dy-

namical model. The study has been performed on a set of35 air-fuel ratio dynamic transients generated on a dynamictest bench for a spark ignition Alfa Romeo 1.4 litres with4 cylinders, equipped with a IAW multi-point ECS. A trainof square waves have been imposed on the nominal injec-tion time pulse width in order to generate the air-fuel ratiostrength excursions.Both techniques allow to predict the observed values withgood accuracy, consistently with the physical processes oc-curring in the region interested by the fuel injection. Theresults obtained from the two techniques are discussed andcompared, and the emerging advantages of the Kalman fil-tering approach are shown.

MA3-2 1020

Optimal Idle Speed Control with Induction-to-PowerFinite Delay for SI Engines

Luigi Glielmo Univ. di Napoli Federico IIStefania Santini Univ. di Napoli Federico IIGabriele Serra Magneti Marelli Engine Control Div.

Summary: We present an idle-speed controller designedthrough an optimal LQ technique taking into account dur-ing the design phase the presence of a finite time delay be-tween variations in the manifold pressure and in the pro-duced torque. Effectiveness of the scheme and its robustnessto underestimation of the delay are shown through com-puter simulations.

Keywords: Internal combustion engine control, idle speedcontrol, time delay systems, optimal control, LQ control.

MA3-3 1040

Estimator-Based Adaptive Fuzzy Logic ControlTechnique for a Wind Turbine-Induction GeneratorSystem

Andrea Dadone Politecnico di BariLorenzo Dambrosio Politecnico di Bari

Summary: The control of a wind power plant, operating asan isolated power source, is analyzed. The plant consists of awind turbine and a three-phase induction electric generator,connected by means of a gear box. The mathematical modelsof the wind turbine and of the electrical generator are indi-cated. The use of an Estimator-based Adaptive Fuzzy Logiccontrol technique to govern the system is proposed. The re-sults of a control test case are shown in order to demonstratethe reliability of the proposed control technique.

Keywords: Control, fuzzy, adaptive, wind, system.

MA3-4 1100

Active Suspension Control of Ground Vehicle Heaveand Pitch Motions

Javier Campos The Univ. of Texas at ArlingtonLeo Davis Davis Technologies Int., Inc.Frank L. Lewis The Univ. of Texas at ArlingtonScott Ikenaga The Univ. of Texas at ArlingtonMark Evans Davis Technologies Int., Inc.

Summary: Ride quality depends of a combination of verti-

39

cal displacement (heave) and angular displacement (pitch).Road irregularities are the main factor affecting ride com-fort. Suspension elements between the road wheels andthe vehicle body generate vertical forces which excite bothheave and pitch motions. An active controller design basedon time-scale separation and an “input decoupling transfor-mation” is given. It is shown to give better performance thatconventional passive suspension control.

Keywords: Active suspension control, heave and pitch con-trol, skyhook damping control, time-scale separation, inputdecoupling transformation.

MA3-5 1120

An Object-Oriented Modular Simulation Model forIntegrated Gasoline Engine and AutomaticTransmission Control

Keum-Shik Hong Pusan National Univ.Kyung-Jinn Yang Pusan National Univ.

Summary: In this paper a computer simulation model forcontrol system design of gasoline engines with an auto-matic transmission is presented. A modular programmingapproach has been pursued, and MATLAB/SIMULINKhas been utilized as a programming environment. En-gine/transmission systems are analyzed in the object-oriented fashion. Thus, easy construction of various com-puter models by assembling various objects is possible. Anobject in this paper represents a physical part, an equa-tion, or an algorithm. The top level in the powertrainmodel consists of three classes: an engine, a transmission,and a driveline. Each class is designed to perform by it-self. The construction procedure of a typical powertrainmodel together with supplementary explanation is demon-strated. It is expected that the whole program and individ-ual class constructed in this paper are useful for the automo-tive engineers who design a new engine/transmission sys-tem and/or modify an existing system.

Keywords: MATLAB/SIMULINK, gasoline engine, auto-matic transmission, object-oriented model.

MA3-6 1140

A Comprehensive Model for ICE Oriented to theElectronic Control of the Injection

Michele Anatone Univ. of L’AquilaRoberto Carapellucci Univ. of L’AquilaRoberto Cipollone Univ. of L’AquilaAntonio Sciarretta Univ. of L’Aquila

Summary: The electronic control of spark ignition port in-jected engines requires simulation tools able to predict on-line the relevant dynamics concurring to the mixture forma-tion, mainly during engine transients.A comprehensive mathematical model, specifically con-ceived for this application, is presented in this paper. Themodel is based on a time-dependent physically consistentdescription of the main processes. The most peculiar as-pect is the integration between the description of the air andexhaust gas dynamics inside the manifolds and the modelfor the fuel dynamics in liquid and vapour phases. The gasmodel describes the pressure wave propagation in the ducts

in a lumped-parameter way; the fuel model adopts a quasi-lagrangian two-dimensional approach for the spray and azero-dimensional representation for the fuel puddles. Theoverall model, which has a modular structure, also accountsfor the other relevant processes occurring in the engine, suchas combustion, heat transfer, pollutants formation, shaft dy-namics, etc.The model has been applied on a one-cylinder, electronicallyinjected, research engine (AVL 540), that is under testing bythe authors. The results obtained for the air and exhaust dy-namics point out the accuracy of the model when comparedwith the more complex and resource-consuming method ofcharacteristics. The model has been then applied to build thesteady air maps of the engine and to characterize the param-eters of an universally adopted fuel dynamics model (X-τ) atdifferent operating conditions.

Keywords: Internal combustion engines, electronic control,modeling.

MA4Linear Systems 1

MA4-1 1000

Stable Inversion of MIMO Linear Discrete TimeNon-Minimum Phase Systems

Koshy George Delft Univ. of TechnologyMichel Verhaegen Delft Univ. of TechnologyJacquelien M.A. Scherpen Delft Univ. of Technology

Summary: A novel technique to achieve output trackingvia stable inversion of non-minimum phase linear systems ispresented wherein the desired signal is obtained from fieldmeasurements, and hence corrupted by noise. The earlierapproach to stable inversion does not take into account thenoise in the system. The unknown input decoupled ob-server approach is applicable only to minimum phase sys-tems. Moreover, the unobservable states are inadequatelyconstructed resulting in inferior output tracking in the pres-ence of noise. In this paper we extend this procedure to non-minimum phase systems. We present the novel Stable Dy-namic model Inversion (SDI) approach which is applicableto non-minimum phase systems, and takes into account thepresence of noise in target time histories.

Keywords: Discrete time systems, model inversion, outputtracking.

MA4-2 1020

State Space and Internal Models in Discrete-time LQRegulator Design

Ryszard Gessing Politechnika Slaska

Summary: Discrete-time state space model, denoted by I,is proposed for direct implementation of the discrete-timelinear-quadratic regulator (DLQR) in the case when not allthe state components but only the output of the plant isavailable, or using other words, when not a state feedbackbut output feedback is implemented. It is shown that forthe proposed state, the DLQR problem solution determinesthe dynamic regulator with output feedback. In connection

40

with this the order of the closed loop system is increasedwith respect to the case when the DLQR with state feed-back is applied. It is shown that the CL system with dy-namic DLQR and output feedback implements solution ofthe optimal DLQ problem with state feedback for the aug-mented state space model, denoted by II. It is also shownthat the DLQR problem solution for the model II and ap-propriately chosen performance index gives partially pre-scribed pole placement of the closed-loop system. The caseof non zero both the set point and disturbance is consideredby using an appropriate internal model corrector. Includingthis corrector to the augmented plant the considered case istransformed to usual DLQR stabilisation problem with zeroset point. Using the proposed state determination and inter-nal model approach, the modified DLQR design technique,giving a partially prescribed pole placement, is described.Finally, the method is illustrated in an example.

Keywords: Linear-quadratic regulator, discrete-time sys-tems, state space models, internal models, system design.

MA4-3 1040

Modified Internal Model Control for Unstable Systems

Kou Yamada Yamagata Univ.

Summary: In the present paper, we propose a modified In-ternal Model Control systems that is implemented for un-stable plants. This modification is simple and natural. Somecharacteristics of modified Internal Model Control such asstability, robust stability and so on are clarified.

MA4-4 1100

The Wiener-Hopf Standard Control Problem: A StableFractional Approach

Li Xie Univ. of Posts and Telecom.Dingyu Xue Northeastern Univ.

Summary: In this paper, we present the solution to the stan-dard Wiener-Hopf control problem with the quadratic cost.The solution presented here is based on the stable, rationaland proper fractional representation theory and spectral fac-torizations, and in particular the controller class is proper.Three cases of the external signal are considered. Under aset of assumptions, the minimum and finite costs are given.Meanwhile controllers are also parameterized in terms ofan arbitrary stable, real rational, and strictly proper matrixZ(s).

Keywords: The standard control problem, Wiener-Hopf op-timization design, stable fractional representation.

MA4-5 1120

Reduction of Singular 2D Models to EquivalentStandard Models

Tadeusz Kaczorek Warsaw Univ. of Technology

Summary: A new extended Roesser type model is intro-duced. It is shown that:

1. Any singular 2D general model (1) with can be reducedto the model (6) (or (6’));

2. Regular singular 2D model (9) can be reduced to stan-

dard extended Roesser type model (11),

Sufficient conditions are established under which a singular2D general model (1) can be reduced to standard models ofthe form (28) or (35).

Keywords: Extended Roesser model, singular 2D model,reduction, sufficient conditions.

MA4-6 1140

Some New Results in Theory of Controllability

Agamirza Bashirov Eastern Mediterranean Univ.Nazim Mahmudov Eastern Mediterranean Univ.

Summary: The new necessary and sufficient conditions, for-mulated in terms of convergence of a certain sequence ofoperators involving the resolvent of the negative of the con-trollability operator, are found for deterministic linear sta-tionary control systems to be completely and approximatelycontrollable, respectively. These conditions are applied tostudy the ST -controllability (that is a property of attainingfor the time T an arbitrarily small neighborhood of eachpoint in the state space with a probability arbitrarily near toone) and the CT -controllability (that is the ST -controllabilityfortified with some uniformity) of stochastic systems. It isshown that a partially observable linear stationary controlsystem with an additive Gaussian white noise disturbanceis ST -controllable (CT -controllable) for each T > 0 if andonly if its deterministic part is approximately (completely)controllable for each time T > 0.

Keywords: Controllability, stochastic controllability, linearsystems.

MA5Adaptive Control 1

MA5-1 1000

Iterative Adaptive (Unfalsified) Control

Robert Kosut SC Solutions Inc.

Summary: Uncertainty model unfalsification is reviewed.Based on the concept of unfalsification, an iterative direct(unfalsified) adaptive control scheme is proposed whichmay alleviate some of the difficulties of iterative adaptation,e.g., convergence.

MA5-2 1020

On the Design of Direct Adaptive Controllers

Felipe M. Pait Univ. de Sao Paulo

Summary: Direct adaptive control systems without the fa-miliar reference models is considered. A framework for de-sign using quadratic cost functions is presented, and cor-responding error equations are derived using ideas fromlinear-quadratic optimal control.

Keywords: Adaptive control, direct control, Linear-Quadratic optimal control.

41

MA5-3 1040

Tuning via Measurements of the Squared Error

Felipe M. Pait Univ. de Sao Paulo

Summary: Given data x, we wish to adjust a parameter vec-tor p(t) so as to minimize z(t) = xp − y as best we can insome norm sense. If y (or equivalently z) were available, wemight choose as our cost function the integral of z2 and min-imize it using standard least-squares algorithms. We con-sider the case when neither y nor z are available; rather, ateach instant we are able to choose p(t) and measure z2(t),that is to say, z’s magnitude but not its sign.

Keywords: Adaptive control, parameter estimation, directcontrol.

MA5-4 1100

Adaptive Generalized Predictive Control Subject toInput Constraints

Andrzej Krolikowski Technical Univ. of Poznan

Summary: Generalized predictive control (GPC) problemof ARIMAX/ARMAX system in the pres-ence of input con-straints and parametric uncertainty is considered. An adap-tive controller is implemented in an indirect way, and theconsidered constraints imposed on the control signal are ofthe rate, amplitude and energy type. A simulation compara-tive study of the adap-tive control system behavior is givenwith respect to the design parameters and constraints. Ad-ditionally, two one-step controllers are compared by meansof simulations.

Keywords: Generalized predictive control, adaptive con-trol, input constraints.

MA5-5 1120

Decentralized Adaptive Controller with Zero ResidualTracking Errors

Boris M. Mirkin Academy of Sc. of Kyrgyz Republic

Summary: In this paper we develop a unified approachto the solution of the adaptive decentralized tracking prob-lem. First we propose a decentralized information setup ofcontrol with reference model coordination, which allows touse coordinating information about reference signals of theother subsystems in all local control laws. This setup guar-antees zero residual tracking errors for unmodeled intercon-nections and the local dynamics. We proposed a modifiedlocal adaptive control scheme with an additional delayedsignal, which improves the transient performance. We useour new setup for the decentralized adaptive control of hy-brid systems in which the control parameters are updated atdiscrete instants.

Keywords: Adaptive decentralized control, large-scale sys-tems, model coordination.

MA5-6 1140

Advanced Adaptive Control for Complex NonlinearProcesses

Nicolae Constantin Univ. Politehnica of BucharestIon Dumitrache Univ. Politehnica of Bucharest

Summary: Adaptive techniques based on neural-networksare investigated in an application to the identification andsubsequent on-line control of a process exhibiting nonlinear-ities and typical disturbances. The method proposed con-sists of a novel identification technique based on extendedmemory adaptation (EMA) and an efficient implementationof the predictive control based on a nonlinear programmingmethod. A forced circulation evaporator was chosen as a re-alistic nonlinear case study for the techniques discussed inthe paper.

Keywords: Adaptive control, nonlinear models, on-linecontrol, neural networks.

MM1Optimization Methods

MM1-1 1430

PI Controller Tuning via Multiobjective Optimization

I. K. Kookos Imperial CollegeK. G. Arvanitis National Tech. Univ. of AthensG. Kalogeropoulos Univ. of Athens

Summary: A great number of PID controller tuning meth-ods is now available to the designer of process control sys-tems. Most of them are based on the satisfaction of singledesign objectives, such as the decay ratio, phase and gainmargins, resonant peak and frequency, overshoot and cer-tain error integral criteria. However, these methods haveseveral shortcomings that stem from the fact that all degreesof freedom, namely the controller adjustable parameters, areconsumed in order to satisfy a single objective. It is widelyrecognized that the solutions of numerous design problems,in various branches of engineering, are incomplete becausethey fail to take into account all the important characteristicsof the particular problem. Controller design problems areamong them. The problem that the controller designer facesis the simultaneous satisfaction of several criteria that areposed either on the time or on the frequency domain. In thisrespect, in this paper, a new method for tuning PI and PIDcontrollers for models commonly used in process control, ispresented. The proposed method is based on the satisfactionof more than one control design objectives. The design ob-jectives are the satisfaction of certain phase and gain margin,the maximization of the resonant frequency and the mini-mization of a weighted integral of squared error. In solvingthe multiobjective optimization problem obtained, a simpli-fied goal attainment formulation is proposed. The useful-ness of the proposed method is demonstrated through sim-ulation examples and a comparison with well known tuningformulas is also provided. The proposed method gives satis-factory results for models such as integrator plus delay timeand first order plus delay time models. Furthermore, themethod is shown to give acceptable tunings even in the ex-

42

treme cases where the delay time is the dominant feature ofthe system under study. Thus, the proposed method is ap-plicable in a wide range of controller design problems com-monly encountered in process control giving satisfactory re-sults.

Keywords: PID controllers, process control, tuning rules,multiobjective optimization, goal attainment method.

MM1-2 1450

Decomposition-Coordinated Optimization ofLarge-Scale Discrete Systems withParallel-Sequential Coordinated Scheme

Nataly M. Lychenko Academy of Sc. of Kyrgyz Republic

Summary: Hierarchical algorithms are developed for opti-mal control of interconnected discrete dynamic large-scalesystems with control and state constraints. Synthesis of al-gorithms based on goal function adaptation in a speciallyformulated intermediate equivalent optimization problemin three (or two) levels. New algorithms are used iterativeparallel-sequential coordination scheme which take in to ac-count information about subsystem states in the calculationcoordinated parameters. One feature of this is that fixingstate and control prediction trajectories are not common forall subsystems but update for them. This algorithms haveshown computational benefits.

Keywords: Optimal control, large-scale interconnected sys-tems, decomposition-coordinated methods.

MM1-3 1510

Generalized PID Controller

Ilan Rusnak RAFAEL — ADA

Summary: The PI, PD and PID controllers are widely usedand successfully applied controllers to many applications.Their successful application, good performance, easiness oftuning are sufficient rational for their use, although theirstructure is justified by heuristics. In this paper by the useof optimal control theory we formulate a tracking problemand show those cases when their solution gives the PI, PDand PID controllers, thus avoiding heuristics and giving asystematic approach to explanation for their excellent per-formance. It is shown that the PI controller is optimal fora first order system, the PID controller is optimal for a 2ndorder systems with no zero. The reference trajectory is gen-erated by a system identical to the plant. Then the same ap-proach, that led to the PI and PID controller, is applied to ageneral linear, strictly proper system and a generalized PIDcontroller is derived. Such controller is called here PIDn-1controller. As an example, a generalized PID controller for aDC motor with one flexible mode is presented.

Keywords: PID controller, optimal control

MM1-4 1530

Lagrange Problem for Non-Standard NonlinearSingularly Perturbed Systems

Emilia Fridman Tel-Aviv Univ.

Summary: We study the infinite horizon optimal control

problem for an affine singularly perturbed system, whichis nonlinear not only in the slow variable (as in the stan-dard case), but also in the fast variable. We construct anε-independent composite controller by solving a slow par-tial differential equation, that corresponds to the reducedHamiltonian system, and by solving a fast partial differen-tial equation. The composite controller solves the local La-grange problem for all small enough ε. It solves also the cor-responding problem for the descriptor system. We obtain anasymptotic approximation of the optimal controller, optimaltrajectory and of the solution to Hamilton-Jacobi equation.

Keywords: Singular perturbations, nonlinear optimal con-trol, descriptor systems.

MM1-5 1550

Optimization for Part Nesting and Layout Using aDistributed SPMD Architecture

Joseph P. Wetstein Drexel Univ.Allon Guez Drexel Univ.

Summary: There is a need to perform highly complex real-time optimization during manufacturing to solve the prob-lem of maximizing the yield of production while minimiz-ing the cost of materials. Although traditional linear andnon-linear programming approaches have dealt with theseproblems, some problems are too complex and result in acombinatorial explosion for those methods. We are inter-ested in developing a solution to the leather nesting task,an NP-hard problem. We feel the best solution to a largeand complex problem such as this is to attempt to model thebehavior that is already evident in nature. There are manyexamples in nature providing an optimal packing or nest-ing of items. Our algorithm will offer a better solution toproblems of this class by presenting a controlled trade-offbetween computation time and regional optimality of thesolution. Our goal is to develop an algorithmic scheme thatwill yield the best possible solution to this combinatorial op-timization problem using a hierarchical approach to objectdetermination and placing, an intelligent controller, and par-allel executed simple local decision rules based on nearestneighbor parameters which achieve results that are compet-itive with or better than the results of a human nester, andto aid in manufacturing. In our SPMD model, each object tobe placed will be a separate process in a parallel computingarchitecture, and will consider the parameters (location, ori-entation) of its nearest neighbors (other parts in the vicinity)to make translocation and re-orientation decisions. We in-tend to show that the fusion of existing algorithms and theestablishment of uniform and ’simple’ local decision ruleswill more quickly yield to a better optimum for the entiresystem. The algorithms developed in this study are applica-ble to a large class of optimization problems.

MM1-6 1610

Computing Resources Dynamic Optimization ofDigital Multichannel Control Systems

Arkadi I. Frid Ufa Aviation Technical Univ.Adel K. Enikeev Ufa Aviation Technical Univ.Boris A. Novikov Ufa Aviation Technical Univ.

Summary: The new view on digital multichannel control

43

systems design is considered. The suggested method pro-vides an optimal reallocation of calculating resources ac-cording to quality criterion including its sensitivity to sam-pling period variations and the risk degree of inspected co-ordinate approaching to critical value. Reallocation of cal-culating resources is based on the results of pre-calculatedresponse surface which represents the solution of the sys-tem of non-linear equations by the Lagrange undeterminedmultipliers method.

Keywords: Dynamic optimisation, calculating resources,reallocation.

MM2Inteligent Control and Neural Networks

MM2-1 1430

The Basic Ideas of Neural Predictive Control

Leizer Schnitman Aeronautics Inst. of TechnologyAdhemar de B. Fontes Bahia Federal Univ.

Summary: This paper presents the application of predictivecontrol techniques using Artificial Neural Nets (ANN). Theidea is illustrate the structure of the predictive controller andthe optimization functions that is usually used to update thecontrol action, then apply the ANN technique. The ANNequations and its gradient equations are developed. Basedon the ANN capacity to predict, on a optimization functionand on a rule to update the control action, NPC (Neural Pre-dictive Control) algorithms are developed and applied tocontrol the selected plant. The paper also proposes an in-telligent adaptability to ponder control action as function ofdominant pole displacement.

Keywords: Predictive control, neural nets, nonlinear sys-tems.

MM2-2 1450

A Rule-Based Neuro-Optimal Controller for NonlinearMIMO Systems

Serhat Tuncay Middle East Technical Univ.Kemal Leblebicioglu Middle East Technical Univ.Canan Ozgen Middle East Technical Univ.Ugur Halici Middle East Technical Univ.

Summary: In this study, we propose a new method to con-trol multi-input multi-output (MIMO) systems optimally.The method is based on a rule-base derived optimally, whichis then interpolated by neural networks. The idea is orig-inally based on the knowledge-based artificial neural net-works (KBANN) which perform interpolation in the rulespace of an expert system.

Keywords: Optimal control, neural networks, rules basedsystems, interpolation.

MM2-3 1510

Neural Network Based Softsensor for a TubularReactor

Marius Anghelea Univ. of GentFilip Declercq Univ. of GentRobin De Keyser Univ. of GentMartin Decoster EXXON Chemical Comp.

Summary: The paper deals with the mathematical model-ing of a high pressure jacketed tubular reactor and the de-velopment of a neural network based softsensor for estimat-ing the polymer (low density polyethylene) quality at theexit of a tubular reactor. This is of a great practical impor-tance as the measurement of the polymer quality (given bythe weight average degree of polymerization) is essentiallyneeded for designing any control algorithm that regulatesthe end-use properties of the polymer. The reactor modelconsists of differential equations written as function of thedimensionless reactor length. The rates of formation of var-ious species in the reactor are described by the kinetics offree radical polymerization. The method of moments is usedfor describing the polymer molecular properties. The chaintransfer to solvent process and the coolant flow throughthe reactor jacket are taken into account. The reactor mix-ture speed is assumed constant, the inlet pressure and thepulse valve effects are not included in the model. The reac-tor model is simulated, compared with results obtained byother authors and resembles the behavior of a healthy tubu-lar reactor. The general property of the reactors in the chem-ical industry is that they are highly nonlinear. Moreover,important variables (such as the weight average degree ofpolymerization) can not be measured directly and on-line.An alternative is to use a softsensor measurement. The soft-sensor is an artificial intelligence instrument and belongs tothe class of inferential measurement techniques. The paperdevelops the general concept of the softsensor and appliesit to the problem of estimating the polymer quality. A tubu-lar reactor with one injection of initiator, with a preheatersection and coolant flow is considered. The process infor-mation easily measurable and directly available for buildingthe softsensor consists of the temperature profile, the coolantflow temperature and the solvent concentration. As softsen-sor structure a feedforward neural network with one hid-den layer is employed. Neural networks have strong nonlin-ear function approximation capabilities and are widely usedfor modeling of nonlinear processes. The data for the neu-ral networks training/validation is obtained by integratingthe reactor model for different initial conditions. The neu-ral networks inputs consist of directly measurable processvariables and process characteristics (such as the area underthe temperature profile peak, the peak location). The sol-vent concentration proves to be an important input for thesoftsensor and improves significantly the softsensor approx-imation property.

Keywords: Mathematical modeling, tubular reactor, poly-merization, neural network, softsensor.

44

MM2-4 1530

An Expert-Aided Implementation Interface forIndustrial Process Control Systems

James H. Taylor Univ. of New BrunswickCheney Chan Univ. of New Brunswick

Summary: We present the design and implementation of anew expert-system “front end” or Design Advisor for Imple-menting Systems (DAIS) for use in conjunction with a com-mercial digital control system environment, e.g., the ElsagBailey INFI 90 System. The objective of DAIS is to make itsubstantially easier for applications engineers to make ef-fective use of the broad spectrum of capabilities of this andsimilar hardware and software systems for industrial con-trols implementation. This concept is of quite general appli-cability for industrial controls environments.

Keywords: Computer-aided control engineering, expert-aided controls implementation, controls design advisor.

MM2-5 1550

A Self-Organizing Neurocontroller for VibrationSuppression

Dimitrios Moshou K.U.LeuvenJan Anthonis K.U.LeuvenPal Jancsok K.U.LeuvenHerman Ramon K.U.Leuven

Summary: The Self-Organizing Map Neural Network isused in a supervised way to represent a sensor-actuatormapping. The learning of the controller assumes no priorinformation, but only reward/failure signals that are pro-duced by an evaluation criterion. The evaluation criterionused is based on the low-pass filtering of the gradient of areward function and the local storing of the filtered gradientvalue. The control method is tested in vibration isolation ofa flexible spray boom used in agriculture for pesticide ap-plication. The Neural Network learns to stabilise the boomon-line without any prior information and with a very highperformance.

Keywords: Neural networks, self-organizing systems, ac-tive vehicle suspension, agriculture.

MM3Control Applications

MM3-1 1430

A New Modeling of the Macpherson SuspensionSystem and its Optimal Pole-Placement Control

Keum-Shik Hong Pusan National Univ.Dong-Sub Jeon Pusan National Univ.Hyun-Chull Sohn Pusan National Univ.

Summary: In this paper a new model and an optimal pole-placement control for the Macpherson suspension systemare investigated. The focus in this new modeling is the ro-tational motion of the unsprung mass. The two general-ized coordinates selected in this new model are the verti-

cal displacement of the sprung mass and the angular dis-placement of the control arm. The vertical acceleration ofthe sprung mass is measured, while the angular displace-ment of the control arm is estimated. It is shown that theconventional model is a special case of this new model sincethe transfer function of this new model coincides with thatof the conventional one if the lower support point of thedamper is located at the mass center of the unsprung mass.It is also shown that the resonance frequencies of this newmodel agree better with the experimental results. Therefore,this new model is more general in the sense that it providesan extra degree of freedom in determining a plant modelfor control system design. An optimal pole-placement con-trol which combines the LQ control and the pole-placementtechnique is investigated using this new model. The controllaw derived for an active suspension system is applied tothe system with a semi-active damper, and the performancedegradation with a semi-active actuator is evaluated. Simu-lations are provided.

Keywords: Suspension, control arm, frequency response,optimal control, pole-placement.

MM3-2 1450

A Numerical Algorithm for the Design of aDecentralized Controller for Open-Channel Networks

Carla Seatzu Univ. of Cagliari

Summary: In this paper we propose the design of a decen-tralized constant-volume control law for open-channel net-works. The decentralized controller enables us to maintainthe stored volumes in the different reaches practically con-stant, even with variations in users withdrawals, by actingonly on the upstream gate of the reach whose volume varia-tion is detected.Control law is designed by solving a linear least squaresproblem in the frequency domain. The numerical algo-rithm adopted allows us to impose the desired structure tothe feedback gain matrix by means of the optimization ofthe controller parameters. It makes the closed-loop trans-fer function approach a target function as closely as possibleover a specified frequency range.

Keywords: Open-channels, decentralized control, struc-tured feedback control law, linear least squares problem.

MM3-3 1510

Flight Control Design for a Missile: An ApproximateFeedback Linearization Approach

Antonios Tsourdos Cranfield Univ.Anna L. Blumel Cranfield Univ.Brian A. White Cranfield Univ.

Summary: Input-output approximate linearisation of anon-linear sixth order system has been studied. A methodfor controlling the non-linear system that is i/o linearisableis examined that retains the order and the relative degreeof the system in the linearisation process, hence producinga linearised system with no internal or zero dynamics. De-sired tracking performance for lateral accelerations and rollrate of the missile is achieved by using a non-linear controllaw that has been derived by selecting the lateral velocitiesand roll rate as the linearisation outputs. Simulation results

45

are shown that exercise the final design and show that thelinearisation and controller design are satisfactory.

Keywords: Non-linear multivariable control, feedback lin-earisation, trajectory control, missile system.

MM3-4 1530

Robust Quasi NID Current and Flux Control of anInduction Motor for Position Control

Marc van Duijnhoven Eindhoven Technical Univ.Marian J. Błachuta Silesian Technical Univ.

Summary: In the paper, a new control design method calledDynamic Contraction method is applied to the flux andquadrature current robust control of an induction motor op-erated using the field orientation principle. The resultinginput-output decoupled and linearized drive is then usedfor time-optimal position control. Two control structuresproviding a practically time-optimal control are presentedand compared.

Keywords: Dynamic Contraction Method, field orientedcontrol, time optimal control.

MM3-5 1550

Contact Elimination in Mechanical Face Seals UsingActive Control

Joshua Dayan Technion — IITMin Zou SeagateItzhak Green Georgia Tech

Summary: Wear and failure of mechanical seals may be crit-ical in certain application and should be avoided. Largerelative misalignment between the seal faces may cause in-termittent contact and the increased friction eventually canbring failure. Adjustment of seal clearance is probably themost readily implemented method, of reducing the relativemisalignment in order to eliminate seal face contact, duringseal operation. This method is demonstrated with the aidof a noncontacting flexibly mounted rotor (FMR) mechan-ical face seal test rig employing a cascade control scheme.Eddy current proximity probes measure the seal clearancedirectly. The inner loop controls the clearance, maintaininga desired gap through adjusting the air pressure in the ro-tor chamber of the seal. When contact is detected the outerloop adjusts the desired clearance according to variance dif-ferences in the probes signals. These differences in variancehave been found to be a reliable quantitative indication forsuch contacts. They are complimentary to other more quali-tative phenomenological indications, and provide the con-trolled variable data for the outer loop. Experiments areconducted to test and verify the control scheme and strat-egy. Analysis and results both show that, for the seal underinvestigation and contrary to intuition, reducing seal clear-ance can eliminate contact.

Keywords: Mechanical face seal, contact elimination, activecontrol.

MM3-6 1610

Design, Simulation & Control of a SegmentedReflector Test-bed

Mauricio Morales California State Univ., LAMajdedin Mirmirani California State Univ., LAHelen Boussalis California State Univ., LA

Summary: Segmented reflectors are one of the best prac-tical choices for future astrophysical missions. Because itlacks the dimensional stability provided by a monolithic pri-mary mirror, a segmented reflector requires an active controlsystem in order to make the reflecting surface have com-parable optical performance. This paper describes a con-trol oriented test-bed developed at the Control and Struc-tures Research Laboratory (CSRL) at California State Uni-versity, Los Angeles (CSULA). The CSRL test-bed is a 2.4 mfocal length Cassegrain configuration telescope consisting ofa 2.66 m actively controlled segmented primary and an ac-tive secondary. The primary consists of six hexagonal panelssurrounding a fixed central panel and supported by a light-weight flexible truss structure. The project has been fundedby NASA to study the complex dynamic behavior of largesegmented optical systems.

Keywords: Control, structures, segmented reflector, test-bed.

MM4Linear Systems 2

MM4-1 1430

Generalized Versions of Bode’s Theorem

Amos E. Gera Elta

Summary: The fundamental theorem of Bode states that theintegral over all frequencies of the natural log of the mag-nitude of the sensitivity function vanishes. This was seento be true for an open-loop stable function with the differ-ence between the degrees of the numerator and denomina-tor at least 2. Extensions of the theorem have been carriedout (1) to unstable open-loop systems (2) to eliminate thelimitation that the difference between numerator and denu-merator must be greater than 1. A generalized version ofBode’s theorem to include weighted sensitivity integrals ispresented. The theorem is also extended to include transferfunctions with time delay. Some examples are provided toshow its utility.

Keywords: Bode’s theorem, sensitivity integral, weightedsensitivity, time-delay systems.

MM4-2 1450

On a Conjecture and the Internal Model

Pedro M. G. Ferreira PUC-Rio

Summary: In the robust tracking problem with two-outputplants it is shown that if the plant is unstable, the mumera-tors corresponding to the two outputs cannot be unrelated.However, if the plant is stable, the two parts of the plant

46

can be unrelated and, in fact, the compensator which solvesthe problem incorporates an inverse internal model of theexogenous signal.

Keywords: Tracking, two-output plants, inverse internalmodel, robustness, stability.

MM4-3 1510

Reliable Computation of the Input-State-OutputRelations in Autoregressive Representations ofMultivariable Systems

Ferdinand Kraffer Inst. of Inf. Theory and Automation

Summary: Input-state-output analysis of systems with ex-ternal variables on equal footing is pursued through a nu-merical algorithm for processing a set of linear differentialequations in the form of an autoregressive representation.Instead of resorting to the computation of elementary poly-nomial operations, numerically robust routines from numer-ical linear algebra are used to compute an implicit state-space realization in the form of a minimal driving variablerepresentation. The representation is used to detect candi-date inputs among the external variables. The algorithmis based on polynomial matrix to state space conversionsleading to application of well-proven methods of numer-ical linear algebra such as Gram-Schmidt orthonormaliza-tion, Householder transformations, and the singular valuedecomposition.

Keywords: Subspace methods, numerical methods, linearsystems.

MM4-4 1530

The Suboptimal Tracking Problem in Linear Systems

Petr Dostal Technical Univ. BrnoVladimir Bobal Technical Univ. Brno

Summary: The contribution is focused on control sys-tem design for the purposes of suboptimal LQ tracking incontinuous-time SISO linear systems. The proposed methodis based on the polynomial approach. The presented proce-dures are proposed for a class of references frequently usedin practice. The resulting controller is obtained via the so-lution of a polynomial Diophantine equation with the rightside given by spectral factorization. The theoretical resultsare tested on an illustrative example.

Keywords: LQ tracking, polynomial approach, spectral fac-torization, Diophantine equation.

MM4-5 1550

Margins and Bandwidth Limitations of NMP SISOFeedback Systems

Marcel Sidi Center of Technolog. Edu., HolonOded Yaniv Tel-Aviv Univ.

Summary: Equations and graphs in order to evaluate thelimitations and tradeoff between extreme cross over fre-quencies and gain and phase margins of an important classof open loop non minimum-phase transfer functions, as afunction of the right half plane zeros or poles, are given.

Keywords: Feedback, non minimum phase, bandwidth

limitations, margins, SISO.

MM4-6 1610

Reachability and Controllability of Positive LinearSystems with State Feedbacks

Tadeusz Kaczorek Warsaw Univ. of Technology

Summary: It is shown that the reachability and controlla-bility of positive linear systems is not invariant under thestate-feedbacks.

Keywords: Controllability, invariance, positive linear sys-tem, reachability, state-feedback.

MM5 (I)Nonlinear Systems Identification inPractice

MM5-1 1430

Adaptive Hybrid Physical/Neural Network Modelingand its Application to Greenhouse ClimateOptimization

Raphael Linker Technion — IITIdo Seginer Technion — IITPer-Olof Gutman Technion — IIT

Summary: This paper focuses on adaptive modeling ofnon-linear systems which operate in slowly changing envi-ronments. Due to an inability to control the environment,a large amount of data spanning the whole feasible inputspace can not be collected over a reasonably short period oftime. As a result, modeling such systems with neural net-works, which usually have poor extrapolation properties,might lead to poor results. In order to avoid these poorextrapolations, hybrid physical/neural network models areused. Such models are formed by the combination, in paral-lel, of a physical model (approximately valid over the wholespace), and a radial basis function (RBF) network, whichprovides localized predictions only where training data areavailable.In this paper, the hybrid modeling approach is extended byadapting the RBF network on-line, so that the region overwhich the network is valid grows over time. In order to sim-plify the RBF training, the centers are located on a fixed rect-angular grid: if necessary, a new center is added at the gridpoint closest to the new datapoint. This approach also al-lows for keeping the RBF width constant and equal for allthe centers.The problem of discarding some of the data, so that thedatabase does not become prohibitively large, is also ad-dressed. In order to avoid ‘forgetting’ previously modeledregions of the space, datapoints are not discarded basedsolely on their ‘age’. Rather, the number of datapoints neareach center is limited, and when this limit is reached, theoldest datapoint associated with that center is discarded.Greenhouse climate modeling, and its use for climate opti-mization, is presented as an illustration of the method.

Keywords: Hybrid modeling, adaptation, greenhouse, ra-dial basis function.

47

MM5-2 1450

Initialization and Model Reduction for Wiener ModelIdentification

Anna Hagenblad Linkoping Univ.

Summary: The identification of nonlinear systems by theminimization of a prediction error criterion suffers from theproblem of local minima. To get a reliable estimate we needgood initial values for the parameters. In this paper we dis-cuss the class of nonlinear Wiener models, consisting of alinear dynamic system followed by a static nonlinearity. Byselecting a parameterization where the parameters enter lin-early in the error, we can obtain an initial estimate of themodel via linear regression. An example shows that this ap-proach may be preferential to trying to estimate the linearsystem directly form input-output data, if the input is notGaussian. We discuss some of the user’s choices and howthe linear regression initial estimate can be converted to adesired model structure to use in the prediction error crite-rion minimization. The method is also applied to experi-mental data.

Keywords: System identification, Wiener models, nonlinearsystems, local minima.

MM5-3 1510

Nonlinear Identification of Automobile VibrationDynamics

David T. Westwick Delft Univ. of TechnologyKoshy George Delft Univ. of TechnologyMichel Verhaegen Delft Univ. of Technology

Summary: The identification of nonlinear state-space sys-tems from input-output measurements is considered. Thesystem is separated into a linear state-space system with astatic nonlinearity, driven by the state and input, in feed-back. Initially, the contribution from the nonlinearity istreated as an unknown system input driving an otherwiselinear plant. A neural network is then used to model thefeedback nonlinearity. A realistic simulation of a nonlinearautomobile suspension is used to demonstrate the applica-tion of the identification algorithm.

Keywords: State-space models, subspace methods, stableinversion, nonlinear system identification, neural networks.

MM5-4 1530

Generalization: A Hidden Agenda in SystemIdentification

Jan Larsen Technical Univ. of DenmarkLars Kai Hansen Technical Univ. of Denmark

Summary: This paper concerns the importance of the gen-eralization concept in system identification.Typically system identification is carried out by adapting pa-rameters so as to minimize a criterion or cost function on alimited set of noisy training data samples obtained from theunderlying potentially nonlinear system. However, there isalways the hidden agenda that the model should performwell, not only on the training set, but also on the future sam-ples. Invoking the generalization concept provides a han-

dle on the model bias/variance trade off as well as modelpredictions. In this context, we define generalization perfor-mance as the expected performance on future data.We discuss how the generalization concept can be formu-lated for nonlinear stationary, as well as non-stationary sys-tems using flexible models such as neural networks. Fur-ther, various methods for assessing the generalization per-formance is mentioned, and finally, it is demonstrated howthe generalization performance actively can be used as a toolfor optimizing the model structure.The suggested framework is demonstrated on simple sys-tem identification problems.The paper can be downloaded via: http://eivind.imm.dtu.dk/staff/jlarsen/pubs or ftp://eivind.imm.dtu.dk/dist/1999/larsen.med99.ps.gz

Keywords: Generalization error, non-stationary systems,nonlinear system identification.

MM5-5 1550

Nonlinear Identification of the Position Sled Dynamicsof a CD Player

Jonas E. Sjoberg Chalmers Univ. of TechnologyPer-Olof Gutman Technion — IIT

Summary: This contribution concerns the identification ofthe dynamics of a sled carrying the optics housing of a CDplayer. The memory access time of the CD player depends,among other factors, on the settling time of the sled after astep change. This contribution focuses on the oscillations atthe end of a step response. Measured closed-loop data areused to identify different types of black-box models of thesled dynamic. First linear models are concerned, and thendifferent types of nonlinear models. The different types ofmodels are compared and discussed. Due to poor excita-tion of the plant, some conclusions are uncertain. However,it is clear that the nonlinear models give better simulationperformance on validation data than the linear ones. Also,the oscillations at the end of a step response seem to be con-troller induced. Therefore, it seems appropriate to use differ-ent models, and then also different controllers, at differentparts of a step response.

Keywords: Nonlinear system identification, nonlinear sys-tems, estimation, modeling.

MM5-6 1610

A Global Optimization Approach to Nonlinear SystemIdentification

A. Tiano Univ. of PaviaF. Pizzocchero Univ. of PaviaP. Venini Univ. of Pavia

Summary: This paper addresses the identification problemfor nonlinear multivariable dynamical systems. A novelidentification method is presented, which is based on a suit-able modification of Simulated Annealing Algorithm. Thismethod allows to formulate and solve numerically the re-lated minimization problem by using an efficient randomsearch minimization approach. The main features of theproposed identification method are illustrated through itsapplication to a case study, which consists of the simulated

48

hysteretic model of a vibrating civil structure under seismicexcitation. The results show that the proposed identificationmethod is quite efficient in comparison with other conven-tional identification methods.

MP1Flexible Structures

MP1-1 1650

Design of a Multivariable Pole-Placement Controllerfor the Primary Mirror of the 10m GrantecanTelescope

L. Acosta La Laguna Univ.M. Sigut La Laguna Univ.A. Hamilton La Laguna Univ.J. A. Mendez La Laguna Univ.G. N. Marichal La Laguna Univ.L. Moreno La Laguna Univ.

Summary: In the design of a telescope the most impor-tant specification is to obtain a quality in the images as highas possible. The Gran Telescopio Canarias (GTC) has gota 10m diameter primary mirror, which is segmented in 36hexagonal pieces. This paper presents a multivariable con-troller for the primary mirror based on a local-global strat-egy. This means that the command sent to the segments willhave a local contribution (using information of the own seg-ment) and a global contribution (using information of thewhole mirror). The goal of the control process is to keepthe 36 segments which form the primary mirror always ona paraboloidal surface. The controller design process hasbeen fundamental to know in detail the system dynamicsfeatures in the sense of symmetries and the coupling existingin the primary mirror of the GTC. The requirements aboutthe sampling frecuencies have also been studied. This workis the result of a collaboration between the GRANTECAN,S.A. Company and the Group of Computers and Control ofDepartment of Applied Physics of La Laguna University.

MP1-2 1710

Application of a Classical PD Regulator to the Controlof a Flexible Planar Closed Chain Linkage

Alessandro Gasparetto Univ. of UdineStefano Miani Univ. of Udine

Summary: This paper presents an approach to the controlof a flexible planar closed-chain linkage. A very accuratedynamic model of the system is briefly summarized. Sucha model is then employed to test a classical PD regulatorin simulation to control a flexible planar four-bar linkage.The chosen PD control is described, and the most significantresults of the simulation are presented and discussed.

Keywords: Flexible mechanism, four-bar linkage, PD con-trol, simulation.

MP1-3 1730

Control of Flexible Structures Using Models withDead Time

Natalya Raskin Technion — IITYoram Halevi Technion — IIT

Summary: A new method for modeling and control of non-collocated flexible structures is proposed. The first step ismodeling the controlled structure by a reduced order modelincluding a delay between actuator excitation and the non-collocated sensor measurement, caused by the finite wavepropagation velocity. A fixed order control scheme for com-pensating the response delay is then used. In addition,a new design methodology for the controller gain matri-ces, such that the residual dynamics is suppressed and thespillover effects are reduced, is suggested.

Keywords: Flexible structures, vibration control, order re-duction, dead time, spillover suppression.

MP1-4 1750

Balanced Realization of Flexible Structures withGeneral Damping: A Power Series Approach

Yoram Halevi Technion — IIT

Summary: A method of approximating the balanced real-ization for lightly damped flexible structures is presented.The damped system is treated as a perturbation from theundamped system, and the controllability and observabilitygramians, as well as the balancing transformation are givenas a power series in the perturbation scaling factor. The ap-proximation utilizes the special structure of the system i.e.the positive definitness of the inertia, damping and stiffnessmatrices, and the fact that the damping is small., to obtainclosed form expressions for the series coefficient matrices.These expressions lead to interesting structural properties,which are discussed and related to physical properties of vi-brating systems. The results can be obtained at any level ofaccuracy by appropriate truncation of the series.

Keywords: Order reduction, balanced realization, flexiblestructures.

MP1-5 1810

Computation in closed form of the equations ofmotion for a flexible beam with lumped masses androtational inertias

Laura Menini Univ. di Roma Tor VergataAntonio Tornambe Univ. di Roma TreLuca Zaccarian Univ. di Roma Tor Vergata

Summary: This paper is concerned with the problem ofmodelling flexible structures in which the effects of distrib-uted elasticity and of both distributed and lumped massesare to be taken into account. The eigenvalue-eigenfunctionproblem, which constitutes the exact model of the free vi-brations of the structure, is given for the general case of aflexible beam having lumped masses and rotational inertiaeplaced along its length. The proposed method is applied toa simple case study: a clamped beam with a rigid body at-

49

tached to its free end.

Keywords: Flexible structures, Lagrangian techniques.

MP2Fault Detection

MP2-1 1650

Improved Observer for Sensor Fault Diagnosis of aPower Plant

Silvio Simani Univ. di FerraraCesare Fantuzzi Univ. di FerraraSergio Beghelli Univ. di Ferrara

Summary: The paper focuses on the problem of the deriva-tion of a suitable mathematical description of a power plantfor diagnostic purpose, by using equation error models. Thediagnostic tool obtained has been tested on real data ac-quired from the 120MW power plant of Pont sur Sambre.

Keywords: Fault detection and diagnosis, analytical redun-dancy, model-based approach, unknown input observer,equation error model.

MP2-2 1710

Residual-Sensitive Fault Detection Filter

Robert H. Chen Univ. of California, LAJason L. Speyer Univ. of California, LA

Summary: A fault detection and identification algorithm,called the residual-sensitive fault detection filter, is pre-sented. The objective of the filter is to monitor certain faultscalled target faults and block other faults which are callednuisance faults. This filter is derived from solving a min-max problem which makes the residual sensitive to the tar-get fault, but insensitive to the nuisance faults. It is shownthat this filter approximates the properties of the classicalfault detection filter such that in the limit where the weight-ing on the nuisance faults is zero, the residual-sensitive faultdetection filter is equivalent to the unknown input observerand there exists a reduced-order filter. Fault detection filterdesigns can be obtained for both linear time-invariant andtime-varying systems.

Keywords: Fault detection and identification, worst casedesign, robust fault detection filter, analytical redundancy.

MP2-3 1730

Catastrophic Failure Evaluation

John M. Macdonald Los Alamos National Lab.Howard Nekimken Los Alamos National Lab.Rick Picard Los Alamos National Lab.Keith Olson Los Alamos National Lab.Adam Bates Los Alamos National Lab.Augustine Ortiz Los Alamos National Lab.

Summary: The random events of catastrophic failures im-pacting process control systems and networks are the topicof this paper. In an unpredictable catastrophic event suchas a lighting strike, power outage, or failures only affectingcertain portions of the overall system, how is the integrity

of a system determined? This investigation includes overallthe system impact on random catastrophic failures. Systemscenarios are evaluated with a method for determining im-pacts on these systems. Criticality Values and a compositeSystem Criticality Value are introduced. A system vulner-ability value is derived and investigated. A suggestion formapping this methodology onto a network is encouraged.

Keywords: System criticality value, criticality values, pro-cess control, catastrophic failures, redundancy.

MP3 (I)Target Tracking

MP3-1 1650

Nonlinear Filters with Virtual Measurements

Frederick E. Daum Raytheon Co.

Summary: The method of virtual measurements will be de-scribed for designing nonlinear filters. This filter theory isbased on the exponential family of probability densities, andit generalizes the Kalman filter and the Benes filter. In someimportant applications the performance of the new filter isvastly superior to the extended Kalman filter (EKF). Unlikethe EKF, the new theory does not use linearization. This the-ory can be used to design exact nonlinear filters as well asapproximate nonlinear filters. The new theory uses “virtualmeasurements” to avoid the problem of solving a system ofnonlinear partial differential equations, and replaces it withthe problem of solving two linear PDEs off-line, analogousto the Hopf-Cole transformation. The basic mathematicaltool is a relatively new theory called “finite sums decompo-sition,” developed by F. Neuman.

MP3-2 1710

System Level Performance of Radar Waveforms

Ruixin Niu Univ. of ConnecticutPeter Willett Univ. of ConnecticutYaakov Bar-Shalom Univ. of Connecticut

Summary: It is well known that different radar signal wave-forms produce very different resolution cells. These affectfine measurement error, gross measurement error throughspurious sidelobe “pop-ups,” and miss probability, andhence yield different tracking performance. In a recent pa-per, sonar waveform selection was explored via the hybridconditional averaging (HYCA) method — a technique forevaluating the dynamic interaction between tracking andimperfect detection. Extension to the radar case is the sub-ject of this paper, and this is more than evolutionary, since inorder that range-rate measurements be available from reso-lution cells it is necessary that coherent processing of multi-ple pulses be used.

Keywords: Target tracking, waveform, ambiguity function,resolution cell, radar.

50

MP3-3 1730

A Radar Power Multiplier Algorithm for Acquisition ofLow Observable Ballistic Missiles Using an ESARadar

Sivaloganathan Sivananthan ARCON Corp.Thiagalingam Kirubarajan Univ. of ConnecticutYaakov Bar-Shalom Univ. of Connecticut

Summary: The problem of acquiring an incoming theaterballistic missile (TBM) presents many complex challenges.The missile should be detected and its state estimated usingthe measurements available from a short window of timebecause the flight time is short. In this paper the acquisi-tion of an incoming tactical ballistic missile using the mea-surements from a surface based Electronically Scanned Ar-ray (ESA) radar is presented. In view of the emergence oflow radar cross section TBMs, it is important to be able toacquire low SNR targets at long range. Such targets are char-acterized by low detection probability and high false alarmrate. We present a batch Maximum Likelihood Estimator(MLE) to acquire the missile while it is exo-atmospheric. Theproposed estimator, which combines MLE with the Prob-abilistic Data Association (PDA) algorithm to handle falsealarms/clutter, also uses the amplitude information (signalstrength), in addition to range and angle measurements, toobtain accurate target state estimates. The use of the am-plitude information facilitates target acquisition under lowSNR conditions. Typically, ESA radars operate at around13dB, whereas the new estimator is shown to be effectiveeven at 4dB SNR, for a Swerling III type fluctuating target,which represents significant counter-stealth capability. Inother words, this algorithm acts as an effective “power mul-tiplier” for the radar by a factor of 8 (9dB). In addition tothe ML estimator, a track validation scheme, which is usedto confirm the presence of an incoming missile at the esti-mated location, is also presented. The Cramer-Rao LowerBound, which quantifies the state estimate accuracies attain-able for this low-observable estimation problem, is also pre-sented and shown to be achieved by the proposed estimator.It is also shown that the optimum detection threshold of theradar can be found by maximizing the information reduc-tion factor that accounts for the loss of information.

Keywords: Ballistic missile defense, anti-stealth capability,maximum likelihood estimation, electronically scanned ar-ray radar, Cramer-Rao lower bound.

MP3-4 1750

Trajectory and Launch Point Estimation for BallisticMissiles from Boost Phase LOS Measurements

Yicong Li Comverse Network SystemsThiagalingam Kirubarajan Univ. of ConnecticutYaakov Bar-Shalom Univ. of Connecticut

Summary: This paper addresses the problem of estimat-ing the trajectory and the launch point of a tactical ballis-tic missile using line of sight (LOS) measurements from oneor more passive sensors (typically satellite-borne). The ma-jor difficulties of this problem include the ill-conditioningof the estimation problem due to poor observability of thetarget motion via LOS measurements, the estimation of the

unknown launch time, and the incorporation of inaccuratetarget thrust profiles to model the target dynamics duringthe boost phase. We present a maximum likelihood (ML) es-timator based on the Levenberg-Marquardt algorithm thatprovides both the target state estimate and the associated er-ror covariance, taking into consideration the complicationsmentioned above. One important consideration in the de-fense against tactical ballistic missiles (TBM) is the determi-nation of the target position and error covariance at the ac-quisition range of a surveillance radar located in the vicinityof the impact point. We present a systematic procedure topropagate the target state and covariance to a nominal time,when it is within the detection range of a surveillance radarto obtain a cueing region. We also provide an estimate andthe error covariance of the (two dimensional) launch posi-tion, which can be used to search for the missile launch site.Monte Carlo simulation studies on typical single and mul-tiple sensor scenarios indicate that the proposed algorithmsare accurate in terms of the estimates and that the estimatorcalculated covariances are consistent with the errors.

Keywords: Tactical ballistic missile defense, boost phase es-timation, maximum likelihood estimation, angle-only track-ing, Cramer-Rao lower bound.

MP3-5 1810

Artificial Neural Network Embedded Kalman FilterBearing Only Passive Target Tracking

Alladi Surendra Rao Naval Sc. and Technological Lab.

Summary: Target tracking is an important issue in under-water surveillance systems. The tracking systems in SeaWarfare utilize Passive sonar to have bearing only infor-mation contaminated with noise, which is assumed here asadditive zero mean Gaussian noise. In underwater war-fare two dimensional target motion analysis is familiar. TheKalman Filter (KF) is used to obtain the target parameterswith the help of bearing data coming from sensor. The errorin target parameters of velocity, range, heading and bear-ing are estimated. For some of the scenarios the errors areunacceptable to real time combat systems. Hence alterna-tive methods are surveyed and Artificial Neural Network(ANN) is coupled with Kalman filter to reduce the creep-ing errors in the solution in spite of Kalman adaptive filtersexist. The network selected for this purpose is Backpropa-gation neural network. The network is pre- trained usingdifferent inputs to predict the said target parameters. Thesimulation results are presented and comparative studiesare conducted. The ANN provides the adaptive capabilitythe filter model needs.

Keywords: Gaussian noise, passive sonar, Kalman filter,neural network, backpropagation.

51

MP5Adaptive Control 2

MP5-1 1650

Adaptive Pole Placement Control of Linear SystemsUsing Periodic Multirate-Input Controllers

K. G. Arvanitis National Tech. Univ. of AthensG. Kalogeropoulos Univ. of Athens

Summary: A new indirect adaptive algorithm is derived forpole placement control of linear continuous-time systemswith unknown parameters. The control structure proposedin the paper, relies on a periodic controller, which suitablymodulates the sampled output and discrete reference sig-nals by a multirate periodically time-varying function. Sucha control strategy, allows us to assign the poles of the sam-pled closed-loop system to desired prespecified values anddoes not make assumptions on the plant other than con-trollability, observability and known order. The proposedindirect adaptive control scheme estimates the unknownplant parameters (and consequently the controller param-eters) on-line, from sequential data of the inputs and theoutputs of the plant, which are recursively updated withinthe time limit imposed by a fundamental sampling period. On the basis of the proposed algorithm, the adaptive poleplacement problem is reduced to a controller determinationbased on the well known Ackermanns’ formula. Knownindirect adaptive pole placement schemes usually resort tothe computation of dynamic controllers through the solu-tion of a polynomial Diophantine equation, thus introducinghigh order exogenous dynamics in the control loop. More-over, in many cases, the solution of the Diophantine equa-tion for a desired set of closed-loop eigenvalues might yieldan unstable controller, and the overall adaptive pole place-ment scheme is then unstable with unstable compensatorsbecause their outputs are unbounded. The proposed con-trol strategy avoids these problems, since here gain con-trollers are needed to be designed. The adaptive scheme,presented in the paper, is readily applicable to nostably in-vertible plants having arbitrary poles and zeros and relativedegree and to systems which do not possess the parity inter-lacing property (namely, they are not strongly stabilizable).Moreover, persistency of excitation and, therefore, param-eter convergence, of the continuous-time plant is providedwithout making any assumption either on the existence ofspecific convex sets in which the estimated parameters be-long or on the coprimeness of the polynomials describingthe ARMA model, or finally on the richeness of the refer-ence signals, as compared to known adaptive pole place-ment schemes.

Keywords: Adaptive control, parameter estimation, poleplacement, multirate controllers, periodically varying con-trollers.

MP5-2 1710

Development of a Self-Tuning PID Controller Basedon Neural Network for Nonlinear Systems

Woo-yong Han Jeonju Technical CollegeJin-wook Han Chonbuk National Univ.Chang-goo Lee Chonbuk National Univ.

Summary: The key point of this research is to develop afast tracker for time-varying nonlinear systems which pre-vious knowledge (i.e., dynamic equations) about the plantwere not known. In order to carry out this research goal,this paper suggests a novel error self-recurrent neural net-works, and develops a fast on-line learning algorithm by us-ing the recursive least squares method. The new neural net-works are considerably faster than the backpropagation al-gorithm and have advantages of being less affected by poorinitial weights and learning rate. Nonlinear adaptive PIDcontroller based on these neural networks has been derivedand tested for the fast tracking problem in a robot manipu-lator.

MP5-3 1730

Multi-Drug Infusion Control Using a Robust DirectAdaptive Controller for Plants with Time Delays

Selahattin Ozcelik Texas A&M Univ.–KingsvilleCesar C. Palerm Rensselaer Polytechnic Inst.Howard Kaufman Rensselaer Polytechnic Inst.

Summary: The control of hemodynamic variables, particu-larly mean arterial pressure (MAP) and cardiac output (CO),is a challenging problem. A good controller is difficult todesign, due to the complex, nonlinear behavior of the sys-tem. Adding to this are the significant changes in dynamicsfrom one patient to another, and even variations in the pa-tients response to the drugs as his condition evolves. A ro-bust direct model reference adaptive controller (DMRAC) isdeveloped for such plants with uncertainty in both the timedelay elements and in the transfer function coefficients. Inorder to satisfy the conditions for asymptotic model follow-ing, it is sufficient to satisfy certain passivity conditions forall possible values of the plant parameters. This is done bytransforming the plant variations and time delays into a fre-quency dependent plant perturbation in the plant transferfunction. Feedforward compensator design procedures arethen developed using an optimization based robust stabilityanalysis, so that the passivity conditions are satisfied.

Keywords: Control of hemodynamic variables, robustadaptive control, positive real.

MP5-4 1750

Indirect Adaptive Control of Drug Infusion for aCirculatory System Model

G. Achuthan Rensselaer Polytechnic Inst.Y. Alekseyenko Rensselaer Polytechnic Inst.A. Ishihara Rensselaer Polytechnic Inst.Howard Kaufman Rensselaer Polytechnic Inst.

Summary: An indirect adaptive control algorithm using re-cursive identification and linear quadratic regulation was

52

usedto compute the infusion of two drugs in order to con-trol blood pressure and cardiac output in a realistic physi-ological nonlinear multiple-input multiple-output represen-tation. Two types of recursive identification were consid-ered; namely, conventional recursive least squares (RLS) anda modified version (MRLS) that penalizes large parameterchanges. Results show that the adaptive procedures are ca-pable of controlling the responses to within their specifiedtolerances and that initial tuning of the adaptation parame-ters can be reasonably performed using a linearized systemmodel.

MP5-5 1810

Optimal Adaptive Control of Uncertain StochasticDiscrete Linear Systems

Ilan Rusnak RAFAEL — ADA

Summary: The problem of optimal control of stochastic dis-crete linear time-invariant uncertain systems on finite timeinterval is formulated and partially solved. This optimalsolution shows that previously published adaptive optimalcontrol schemes and indirect adaptive control schemes donot need heuristics for their rationalization. It is shown thatthese schemes are suboptimal causal approximations of theoptimal solution. The solution is achieved by the introduc-tion of the State and Parameters Observability form - SPOF.This representation of linear time-invariant systems enablesapplication of tools from the LQR-LQG theory of control andestimation of discrete linear time-varying systems. The op-timal solution is exact and non causal. It is composed of acausal optimal estimator of the augmented state composedof the state of the system and the parameters and of a non-causal controller. The solution shows that certainty equiva-lence principle applies for the state and parameters, but theseparation does not apply. A causal suboptimal controller,using certainty equivalence is proposed as an ad-hoc solu-tion. This controller needs only the knowledge of the orderof the system. The scheme is bibo stable for sufficiently lownoises. As an example, the proposed algorithm, is appliedto an unstable nonminimum phase model of a dynamic ve-hicle.

Keywords: Optimal control, adaptive control, uncertain lin-ear systems, stochastic systems.

TUESDAY, June 29

TA1Sampled-Data Systems

TA1-1 950

Improved Wiener-Hopf Method for H2-Design ofSampled-Data Systems

Bernhard P. Lampe Univ. of RostockYephim N. Rosenwasser St.Petersburg U. Ocean Techn.

Summary: The paper deals with the direct design of sam-pled-data systems in theH2-metric by applying the Wiener–Hopf method. The presented technique avoids basic con-trollers during the design procedure that are normally usedto parametrize the set of stabilizing controllers. The sug-gested method is independent from the pole situation of thetransfer matrices of the plant model and is also applicable ifpoles are placed on the imaginary axis.

Keywords: Sampled-data control, MIMO, parametric trans-fer matrix, Wiener-Hopf method, H2-optimization.

TA1-2 1010

H∞ Design of Generalized Sampling and HoldFunctions with Waveform Constraints

Allan C. Kahane Technion — IITLeonid Mirkin Technion — IITZalman J. Palmor Technion — IIT

Summary: This paper deals with the sampled-dataH∞ con-trol problem where both the discrete-time part of the con-troller and the A/D (sampler) and D/A (hold) convertersare design parameters. It is known that the optimal samplerand hold that solve this problem have continuous (exponen-tial) waveforms and thus are not readily implementable ondigital hardware. In this respect, in this paper the problem istreated subject to waveform constraints on hold and samplingfunctions. In partricular, the generalized hold is constrainedto be piecewise-constant and the generalized sampler is con-strained to have piecewise impulse waveform.The paper presents complete solution to this problem. Aseparation between the design of the sampler and the holdis established. Moreover, some interesting interpretationsof the resulting sampled-data controller are discussed. Inparticular, it is shown that the (sub) optimal hold attemptsto “reconstruct” the H∞ state-feedback control law of thesingle-rate sampled-data control system with faster sam-pling period.

Keywords: Sampled-data systems, generalized samplingand hold circuits, H∞ control, lifting technique.

TA1-3 1030

Sampling Zeros and Robust Sampled-Data ControlDesign

Steven Weller Univ. of Newcastle

Summary: In this paper, we investigate the implica-

53

tions for robust sampled-data feedback design of minimumphase sampling zeros appearing in the transfer function ofdiscrete-time plants. Such zeros may be obtained by zero-order hold (ZOH) sampling of continuous-time models hav-ing relative degree two or greater. In particular, we addressthe robustness of sampled-data control systems to multi-plicative uncertainty in the model of the continuous-timeplant. We argue that lightly damped controller poles, whichmay arise from attempting to cancel, or almost cancel, sam-pling zeros of the discretized plant are likely to introducepeaks into the fundamental complementary sensitivity func-tion near the Nyquist frequency. This in turn makes thesatisfaction of necessary conditions for robust stability dif-ficult for all but the most modest amounts of modeling un-certainty in the continuous-time plant. Some H2- and H∞-optimal discrete-time and sampled data designs may lead to(near-) cancellation, and we therefore argue that their suit-ability is restricted.

Keywords: Sampled-data systems, zeros, intersample, sam-pling zeros.

TA1-4 1050

Self-Tuning PID Controller Using δ-ModelIdentification

Vladimir Bobal Technical Univ. BrnoPetr Dostal Technical Univ. BrnoMartin Sysel Technical Univ. Brno

Summary: This contribution presents an application of aself-tuning digital PID controller for process control mod-elled by δ-models. The process is identified by the regres-sion (ARX) models using the recursive least square method(RLSM) with LD decomposition and applied directional for-getting. Controller synthesis is designed on the basis ofa modified Ziegler-Nichols criterion for digital PID controlloops. The ultimate (critical) proportional gain and period ofoscillations have been derived for the second-order δ-model.Control results using digital PID controller on the basic δ-models and z-models are compared.

Keywords: δ-model, recursive identification, PID con-troller, Ziegler-Nichols criterion, self-tuning control.

TA2 (I)Recent Innovations in Process Control

TA2-1 950

Multiple Model Control of a Pilot Distillation Column

Julio A. Rodriguez Sydney Univ.Graham C. Goodwin Univ. of NewcastleJose A. (“Cacho”) Romagnoli Sydney Univ.

Summary: This paper discusses a complete controller de-sign strategy required for implementing Multiple ModelControl (MMC), applied to nonlinear systems. It is shownthat the multiple model design can be recast into a super-visory arrangement, where a global supervisor is utilisedto select the appropriate controller from a fixed family set.Unlike current techniques where Fuzzy Validity Functionsor Bayesian Estimators are utilised in the selection mecha-

nism, the approach of a Multiple Model Observer (MMO)is employed for the selection architecture within the super-visor. This notion, is a natural extension of the MMC de-sign. Switching between the individual controllers is ac-complished bumplessly by using a Multiple Model Bump-less Transfer Mechanism, thus producing a smooth and con-tinuous control signal as the plant passes through differentoperating regions. The above notion is applied to a Pilot Bi-nary Distillation Column, which is nonlinear in nature. Theprincipal nonlinearity of the process is strongly related toits operating point. This paper illustrates that, as the dis-tillation column moves from one operating point to another,the MMC self-regulates according to the operative trajectory,consequently ensuring that global stability and performanceis maintained at an optimal point.

Keywords: Multiple Model Control, Multiple Model Bump-less Transfer Mechanism, LPV.

TA2-2 1010

Online Outlier Detection and Removal

Patrick H. Menold ETH ZurichRonald K. Pearson ETH ZurichFrank Allgower ETH Zurich

Summary: Outliers occur regularly enough in real-worldmeasurement data to constitute a significant practical prob-lem that is not adequately addressed by traditional smooth-ing filters designed to reduce the effects of high-frequencynoise. To address this problem, this paper describes a simpledata cleaning filter for outlier detection and removal whichis based on a causal moving data window that is appropri-ate to real–time applications like closed loop control. Thisfilter is an extension of the well–known median filter: theobserved data point yk is compared to the median y†k ofpresent and past data points. If the distance between thesepoints is large relative to a specified threshold, yk is declaredan outlier and replaced with a more reasonable value y?k. Inthe most favorable circumstances alters the above describeddata cleaning filter only outliers (e.g., shot noise) and doesnot modify nominal data points. Simple implementations ofthis filter require few tuning parameters and no explicit pro-cess model is required for filter tuning. This paper presentssome useful tuning guidelines based on simple characteriza-tions of the nominal variation seen in outlier-free portions ofthe data. To illustrate the utility of this filter, applications arepresented for both real data examples and a simulation ex-ample where the exact results are known and performancecan be assessed more precisely. It is also demonstrated thatthe data cleaning filter described here can be combined withtraditional linear smoothing filters to achieve both protec-tion against outliers and effective noise reduction, but theoutlier filter should preceed the noise filter to achieve theseresults.

Keywords: Data cleaning, decision–based filtering, MADscale estimate, nonlinear digital filter, outliers.

54

TA2-3 1030

Identification for Control Purposes by RelayTechniques: Achievable Performance versusComplexity

G. Marchetti Univ. di PisaClaudio Scali Univ. di Pisa

Summary: Relay techniques are very appealing to performidentification of chemical processes for control purposes; be-ing fast and easy to use, they can be frequently repeated inorder to perform an autotuning of PI/PID controllers. In thepaper two relay techniques are compared in terms of ease ofapplication. duration of experimental tests and achievableperformance.The first one is the Two Channel Relay (Friman and Waller,1997) and makes use of two relays in parallel (one aug-mented by an integrator); it allows to identify a point in thethird quadrant of the Nyquist plane having a specified phaseangle and therefore to guarantee that the resulting PI/PIDcontroller will have a desired gain and phase margin. Thesecond one is the ATV+ technique (recently introduced byScali et al (1999), as a modification of the ATV technique byLi et al (1991)); it allows to identify some more points in thethird quadrant for a completely unknown process and thento build a process model for the design of a PI/PID or modelbased controller.The two techniques are briefly recalled, putting into evi-dence their main features. Different indexes are introducedfor a quantitative evaluation of the duration of tests and ofachievable performance.From the comparison of results for 45 sample processes, itcan be concluded that the TCR is very effective, as it re-quires shorter times for the experimental tests and it allowsto achieve reasonable performance for many cases. On theother side, the ATV+ requires about twice longer experimen-tal times, but, being coupled with a an appropriate design ofthe controllers, allows to achieve better performance for allthe cases with PI/PID controllers and further improvementwith model based controllers.

Keywords: Relay feedback, identification, autotuning, PIDcontrol.

TA2-4 1050

Algorithmic Internal Model Control of UnstableSystems

Ridvan Berber Univ. of AnkaraColeman Brosilow Case Western Reserve Univ.

Summary: An internally stable Algorithmic Internal ModelControl (AIMC) strategy that uses linear or nonlinear modelstate feedback is proposed for unstable systems. The closedloop responses are those that would be obtained from a twodegree of freedom IMC control system, if it were internallystable. Results of several simulations demonstrate the valid-ity of the approach.

Keywords: Internal stability, unstable, linear, nonlinear,model state feedback.

TA2-5 1110

Robust Stability Analysis of Nonlinear ProcessesUsing Empirical State Affine Models and LMI’s

Hector Budman Univ. of WaterlooTimothy Knapp Univ. of Waterloo

Summary: A novel methodology is proposed for the anal-ysis of robust stability of a nonlinear process under linearcontrol. The analysis is based on state-affine empirical mod-els regressed from input-output data. The state model isrepresented by a set of polynomial matrices nonlinear withrespect to the manipulated variables. This model in com-bination with a linear PI controller results in a closed loopmodel that can be shown to lie in a polytope of matrices.This allows for the formulation of a Lyapunov stability testin terms of a simple set of LMI’s (Linear Matrix Inequalities).This set of inequalities can be also expanded to account forinput saturation. The stability analysis produces regions ofstability, in terms of the PI controller parameters, that aresignificantly larger than the regions previously calculatedby a Structured Singular Value test. The conservativeness ofthe analysis is assessed by comparison to closed loop simu-lations of a highly nonlinear CSTR (Continuous Stirred TankReactor) under PI control.

Keywords: Robust stability, Linear Matrix Inequalities,CSTR reactor.

TA2-6 1130

Model Predictive Control of a Continuous GranulationProcess

Anthony A. Adetayo DuPont Central R&DMartin Pottmann DuPont DacronBabatunde A. Ogunnaike DuPont Central R&DBrian J. Ennis E&G Associates

Summary: Granulation, the process by which granules aremade fron powdered, slurried, solution, or molten feed ma-terial is an important process in many industries. This pa-per discusses the development and application of a controlsystem for a specific granulation process. The discussion isused to illustrate the unique characteristics of granulationprocesses in general, and how, as a result, these processespresent unique control problems that prevent the direct ap-plication of many traditional control techniques.In practice, the objective in the granulation process it to pro-duce granules with consistent product quality, as indicatedby various industry standard variables that can be relatedto two fundamental process quantities: particle size distri-bution and bulk density. While it is customary to specifya desired setpoint value for bulk density, the specificationon particle size distribution takes the form of an upper limit(dU), and a lower limit (dL), determined by the screen sizesused in product classification. The paper discusses the pe-culiar control problems arising from such a combination ofproduct quality specifications and then develops a modelpredictive scheme that systematically addresses the prob-lems. Results that illustrate the control system implementa-tion and performance for various situations of practical im-portance will be presented.

Keywords: Chemical process control, Model Predictive

55

Control, granulation process, particle size distribution.

TA3Stability and Stabilization

TA3-1 950

Analytic Conditions for Stabilizability

L. H. Keel Tennessee State Univ.S. P. Bhattacharyya Texas A&M Univ.

Summary: The Nyquist criterion gives a graphical condi-tion for closed loop stability stated in terms of the Nyquistplot of the open loop system transfer function G(s). In thispaper we develop an equivalent, but new, analytical crite-rion for closed loop stability, based on analysis of the be-haviour of a real polynomial functionX(u) constructed fromG(s). It is shown that the real negative zeros ui of X(u) andthe signs of X(u)|u=ui determine the range of stabilizinggains K completely, and in closed form. Besides providing anongraphical and computationally simpler alternative to theNyquist criterion and root locus techniques, this solution isa first step towards investigating stabilizability by higher or-der controllers. Some illustrative examples are given.

Keywords: Constant gain stabilizability, Nyquist criterion,stabilizing gain, Root Locus, real axis.

TA3-2 1010

Stability of Dynamical Systems with ParameterPerturbations

Mark R. Liberzon Moscow State Aviation Techn. Univ.

Summary: New approach for the absolute stability analy-sis of nonstationary control systems is used for investigationof one special kind of dynamical systems. This approach isbased on results from the inner theory, stability theory, op-timal control theory, variational methods. Inner approachallows to obtain sufficient algebraic conditions of absolutestability for different kinds of dynamical systems. The wayto obtain necessary and sufficient conditions for absolutestability of systems with parameter perturbations is shownin this paper. Inner approach is combined with use of thePoutriagin’s Maximum Principle and solving of the Cauchyproblem. This method leads in some cases to algebraic nec-essary and sufficient conditions of absolute stability, but insome cases the question about necessary and sufficient con-ditions of absolute stability obtained by use of developedmethod is open.

TA3-3 1030

BIBO Stability of NARX Models

Andrzej Dzielinski Warsaw Univ. of Technology

Summary: The paper explains an approach to BIBO stabil-ity investigation of NARX control systems. The approach isbased on difference inequalities and assumes the availabilityof an approximate NARX model and the system order. Suf-ficient conditions for modelling error are derived ensuringthe boundedness of the error between model’s and plant’soutputs for the same inputs. For this class of bounded in-

puts sufficient conditions for BIBO stability are given andshown practicable. They also allow designing a controllerusing the model, leading to BIBO stable closed-loop system.

Keywords: BIBO stability, nonlinear systems, difference in-equalities, NARX models.

TA3-4 1050

Improving Efficiency in the Computation of PiecewiseQuadratic Lyapunov Functions

Mikael Johansson Lund Inst. of TechnologyAndrey Ghulchak Lund Inst. of TechnologyAnders Rantzer Lund Inst. of Technology

Summary: In a series of papers, the authors have developeda method for analysis of piecewise linear systems. The ideais to use Lyapunov functions that are piecewise quadratic.Such Lyapunov functions can be computed via convex op-timization in terms of linear matrix inequalities. This pa-per presents two approaches for improving the efficiency ofthese computations. It is shown that by splitting the analysiscomputations into two distinct steps, one can decrease thecomputations with roughly 50% essentially without intro-ducing conservatism. By using ellipsoidal boundings ratherthan polyhedral descriptions of the operating regimes, it ispossible to reduce the computations even further. Com-bined, the two approaches allow the computation times tobe reduced with an order of magnitude compared to previ-ous formulations. However, it is shown that the use of ellip-soidal cell boundings in the S-procedure introduces conser-vatism in comparison with analysis based on polytopic re-gion descriptions. An explicit formula for the minimal vol-ume ellipsoid containing a simplex is also given, togetherwith a complete proof.

Keywords: Piecewise linear systems, Lyapunov stability,convex optimization, convex polytopes, minimum volumeellipsoids.

TA3-5 1110

Practical Stability of Synchronized Chaotic Attractorsand its Control

Tomasz Kapitaniak Technical Univ. of LodzKrzysztof Czolczynski Technical Univ. of LodzJohn Brindley Univ. of Leeds

Summary: In the paper we discussed the application ofthe concept of practical stability to chaotic synchronized at-tractors located at the invariant subspaces, which should bevery useful in the study of chaos synchronization problems.In such problems, having the practical device, we can esti-mate the bounds of the short time perturbations and definetheir sets ω (limits of the ucertainties in initial conditionsand short time perturbations) and Ω (to which evovle theperturbed trajectories). If the synchronized chaotic state ispractically stable in relation to the considered perturbation,we can be sure, that the evolution of the system will notleave the attractor further than allowed by the boundaries ofthe set Ω. Additionally, we present the controlling methodwhich allows enlarging the practical stability regions.

Keywords: Chaos, attractors, practical stability.

56

TA3-6 1130

Remarks on Open-Loop Stabilizability of LinearInfinite-Dimensional Time-Varying Discrete-TimeSystems

K. Maciej Przyłuski Polish Academy of Sc.

Summary: We study open-loop stabilizability of linearinfinite-dimensional time-varying discrete-time systems de-scribed by a linear difference equation of the form xk+1 =Akxk+Bkuk. In particular, we introduce the concept of ‘sta-bilizability modulus’ and present a theorem saying that for abroad class of systems open-loop stabilizability is preservedunder small perturbations of system parameters.

Keywords: Stabilizability, time-varying systems, infinite-dimensional systems, linear systems, discrete-time systems.

TA4Nonlinear Systems 1

TA4-1 950

On Oscillations in Resonant Equations with ComplexNonlinearities

Alexander M. Krasnosel’skii Russian Academy of Sc.

Summary: In the paper the analysis is presented of forcedperiodic oscillations in systems described by the second or-der ODE with resonant linear part and complex nonlinear-ities: with hysteresis and with delay. For such equationswe give conditions of the existence of at least one periodicsolution and conditions of the existence of unbounded se-quences of such solutions. Analogous results are formulatedfor forced periodic oscillations in resonant control systems.

Keywords: Forced periodic oscillations, resonance, delay,hysteresis.

TA4-2 1010

Feedback Resonance in 1-DOF and 2-DOF NonlinearOscillators

Alexander L. Fradkov Russian Academy of Sc.Boris R. Andrievsky Russian Academy of Sc.

Summary: The possibilities of studying nonlinear physicalsystems by small feedback action are discussed. Analyti-cal bounds of possible system energy change by feedbackare established. It is shown that for 1-DOF nonlinear oscil-lator the change of energy by feedback can reach the limitachievable for linear oscillator by harmonic (nonfeedback)action which corresponds to the resonance phenomenon.The feedback resonance phenomenon is demonstrated alsofor 2-DOF system consisting of two coupled pendulums andillustrated by computer simulation results.

Keywords: Nonlinear control, control of oscillations.

TA4-3 1030

Input-Output Models for a Class of NonlinearSystems: Questions and Answers

Ulle Kotta Tallinn Technical Univ.

Summary: In this paper we investigate the possibility ofhaving an input-output model, having a specific structure,for observable multi-input multi-output systems with vec-tor relative degree. The interest in this input-output formarises from the fact that the model has been extensively usedin control design, including sliding mode control. Sincethe subclass of systems having this specific structure is ex-tremely restrictive, we suggest an alternative approach.

Keywords: Nonlinear system, input-output model, vectorrelative degree.

TA4-4 1050

Aspects of Traction Control

Bernard Friedland New Jersey Inst. of Technology

Summary: Propulsion by traction raises several issues, in-cluding modeling of the friction force that produces trac-tion and the design of appropriate control laws. The tra-dional “adhesion” model and several other static and dy-namic friction models are described. Control laws that ac-count in some manner for the severe nonlinearity of tractionare investigated by simulation. It is shown that ignoring thenonlinear effects can result in an unstable system, but thatthe instability can be avoided by an appropriate control lawdesign including an observer that accounts for the nonlinearfriction model.

Keywords: Traction, friction, nonlinear systems.

TA4-5 1110

Energy Control of Hamiltonian Systems underDisturbances

Ilya G. Polushin Russian Academy of Sc.Alexander L. Fradkov Russian Academy of Sc.

Summary: The problem of the energy level stabilization forHamiltonian systems in presence of disturbances is consid-ered. First, it is shown that for 1-DOF systems under suf-ficiently small uniformly bounded force disturbances thespeed-gradient control law ensures ultimate boundednessof energy error. As an auxiliary result the new sufficientconditions for ultimate boundedness of Lyapunov functionalong the trajectories of nonlinear nonstationary dynami-cal system are obtained. Second, for n-DOF systems withdissipation-like disturbances the bounds for achievable en-ergy level are given.

Keywords: Nonlinear control, control of Hamiltonian sys-tems.

57

TA4-6 1130

Nonlinear Systems Admitting Hybrid FeedbackControl Stabilization

Elena Litsyn The College of Judea and SamariaYurii V. Nepomnyashchikh Perm State Univ.Arcady Ponosov Inst. for Matematiske Fag

Summary: A hybrid feedback controller appears as an in-teraction of two counterparts: one is continuous and theother is discrete. In other words, it is a continuous controllaw with an incorporated instrument to make logical deci-sions. Examples can be found in manufacturing systems, in-telligent vehicle highway systems, various chemical plants,population dynamics.It can be shown that in some cases a linear continuous plant(a system of differential equations) which cannot be stabi-lized by an ordinary feedback controller, admits at the sametime a stabilizing feedback controller.We consider a nonlinear differential equation in the planeR2

x = Ax+ Bu+ f(x),

u = u(y), y = Cx,

where f(x) = o(|x|), (A,B) is controllable, (A,C) is observ-able. The following generalization of the main result of Art-stein (1996) is proved:Theorem 1. Assume that the matrix A has no real eigenvalues.Then for any λ > 0 there exist positive constants M, ε and a hy-brid feedback stabilizer u = u(y(·)) with a finite number of loca-tions such that any solution of the controlled system above satisfies

|x(t)| ≤M exp(−λ t)|x(0)|

for all |x(0)| ≤ ε.The proof exploits differential inequalities and a specialtechnique recently developed by the authors.

TA5Filtering

TA5-1 950

Robust H∞ Filtering of Stationary Discrete-TimeLinear Systems with Stochastic Uncertainties

E. Gershon Tel-Aviv Univ.Uri Shaked Tel-Aviv Univ.I. Yaesh Taas Israel Industries

Summary: The problem of H∞ filtering of stationary dis-crete-time linear systems with stochastic uncertainties in thestate space matrices is addressed, where the uncertaintiesare modeled as white noise. The relevant cost function isthe expected value of the standard H∞ performance indexwith respect to the uncertain parameters. A previously de-veloped stochastic bounded real lemma is applied which re-sults in a modified Riccati inequality. This inequality is ex-pressed in a linear matrix inequality form whose solutionprovides the filter parameters. The method proposed is ap-plied also to the case where, in addition to the stochastic un-certainty, other deterministic parameters of the system are

not perfectly known and are assumed to lie in a given poly-tope. The problem of mixed H2/H∞ filtering for the abovesystem is also treated. The theory developed is demon-strated by a simple tracking example.

Keywords: Stochastic H∞ filtering, polytopic uncertainty,mixed H2/H∞ filtering.

TA5-2 1010

The J-Spectral Interactor Matrix in the Discrete-TimeSingular H∞ Filtering Problem

Patrizio Colaneri Politecnico di MilanoMassimo Maroni Politecnico di Milano

Summary: This paper introduces the concept of J-spectralinteractor matrix (JSIM) which is intimately associated withthe singular filtering problem in H∞. An algorithm for thecomputation of a JSIM is proposed and the role of the sys-tem delays in the solution of the singular filtering problemis eventually clarified.

Keywords: Singular H∞ filtering, Interactor matrix, Riccatiequations.

TA5-3 1030

Kalman Bucy Filtering for Singular Output-NoiseCovariance

Francesco Carravetta CNR-IASIAlfredo Germani L’Aquila Univ.Costanzo Manes L’Aquila Univ.

Summary: For a linear Gaussian stochastic system, the fil-tering problem is considered, when the covariance matrix ofthe observation noise is not invertible. A method that al-lows to build up the optimal filter in a number of cases ispresented.

Keywords: Kalman-Bucy filtering, ε-optimal filter, singularproblems.

TA5-4 1050

On the Feasibility and Convergence of H∞ MultistepPredictors

Massimo Maroni Politecnico di MilanoPaolo Bolzern Politecnico di Milano

Summary: An H∞ multistep predictor is designed so as toguarantee a prescribed level of energy attenuation from thedisturbances to the prediction error. It is shown that, fora given value of the attenuation level, an admissible pre-dictor exists over a finite horizon if and only if the solu-tion of a suitable difference Riccati equation lies uniformlyabove a computable lower threshold, which depends on theprediction look-ahead horizon (feasibility condition). More-over, sufficient conditions on the initial state uncertaintyare worked out, which ensure the existence of the predictorover an arbitrarily long time interval and its convergence tosteady-state.

Keywords: H∞ estimation, multistep prediction, Riccatiequations.

58

TA5-5 1110

Nonlinear Observers for a Class of Differential DelaySystems

Woihida Aggoune Univ. Henri Poincare, INRIA CONGEMohamed Darouach Univ. Henri Poincare

Summary: This paper focuses on the design of observersfor a class of nonlinear systems with time-varying delay.Sufficient convergence conditions are established from theLyapunov-Krasovskii theory. These conditions are linked tothe existence of a positive definite matrix satisfying a certainRiccati equation. Using an H∞ theory result, we proposesufficient conditions to guarantee such an existence.

Keywords: Observers, nonlinear systems, time-varying de-lay, Riccati-type equation, Lyapunov-Krasovskii theory,H∞theory.

TM1 (I)Control of Distributed Parameter Systems

TM1-1 1310

Boundary Control of the Korteweg–de Vries–BurgersEquation: Further Results on Stabilization andNumerical Demonstration

Andras Balogh Univ. of California, San DiegoMiroslav Krstic Univ. of California, San Diego

Summary: We consider the Korteweg–de Vries–Burgersequation on the interval [0, 1]. Motivated by poor decayrates of a recently proposed control law by Liu and Krsticwhich keeps some of the boundary conditions as homoge-neous, we propose a strengthened set of feedback boundaryconditions. We establish stability properties of the closed–loop system and illustrate the performance improvement bya simulation example.

Keywords: Korteweg–de Vries–Burgers equation, nonlin-ear boundary feedback control, global stabilization.

TM1-2 1330

Finite Horizon H∞ Control of Systems with StateDelays

Emilia Fridman Tel-Aviv Univ.Uri Shaked Tel-Aviv Univ.

Summary: The finite horizon H∞ control of time-invariantlinear systems with a finite number of point and distributedtime-delays is considered. For controllers coupled Riccatitype partial differential equations are derived. The solutionsto these equations are related to the solutions of the associ-ated Hamiltonian systems. For small time delays the solu-tions and the resulting controllers are approximated by se-ries expansions in powers of the largest delay. Unlike the in-finite horizon case, these approximations possess both reg-ular and boundary layer terms. It is shown that the con-troller obtained by high-order approximations improves theperformance of the system. The performance of the closed-loop system under the memoryless zero-approximation con-

troller is analyzed.

Keywords: Delay systems, H∞-state-feedback control, asy-mptotic approximations, continuous-time systems, smalldelays.

TM1-3 1350

Numerical Criterion for Stabilizing Steady StateSolutions of the Navier-Stokes Equations

Edriss S. Titi Univ. of California, IrvineChongsheng Cao Univ. of California, IrvineYannis Kevrekidis Univ. of California, Irvine

Summary: In this talk we show that by stabilizing a steadystate solution to the Galerkin approximation of the Navier-Stokes equations, using certain linear feedback control, onein fact is stabilizing a near by steady state solution to thefully three dimensional Navier-Stokes equations. Similar re-sults also hold in the context of Nonlinear Galerkin method.It is worth mentioning that all our conditions are explicitand verified by the computed approximate Galerkin solu-tion and that no a priori assumptions are made on the un-known exact solution of the Navier-Stokes equations.

TM1-4 1410

Lax-Phillips Scattering and Well-Posed LinearSystems

Olof J. Staffans Abo Akademi Univ.

Summary: We discuss the connection between Lax–Phillipsscattering theory and the theory of well-posed linear sys-tems, and show that the latter theory is a natural extensionof the former. As a consequence of this, there is a close con-nection between the Lax–Phillips generator and the genera-tors of the corresponding well-posed linear system. All theessential information about these two systems is containedin the system operator S = [A BN ], where A is the genera-tor of the (central) semigroup, B is the control operator, andN is the combined observation/feedthrough operator. If thesystem is compatible in the sense of Helton or regular in thesense of Weiss, then this system operator can be written inthe more familiar form S = [A B

C D ], where C is the observa-tion operator and D is the (generalized) feedthrough oper-ator. We show that S is closed and densely defined. In thereflexive case the adjoint of S is the system operator of thedual system. We give formulas for the Lax–Phillips gener-ator and resolvent in terms of the system operator. By ap-plying the Hille–Yoshida theorem to the Lax–Phillips semi-group we get necessary and sufficient conditions for the Lp-admissibility or joint Lp-admissibility of a control operatorB and an observation operator C.For more details see http://www.abo.fi/˜staffans/ .

Keywords: Lax–Phillips scattering theory, Lp-well-posedlinear system, Lp-admissible control operator, Lp-admissibleobservation operator, Lp-admissible transfer function.

59

TM1-5 1430

Identification and Adaptive Control of SomeStochastic Distributed Parameter Systems

Bozenna Pasik-Duncan Univ. of Kansas

Summary: An important class of controlled linear distrib-uted parameter control systems are those with boundary orpoint control. A survey of some existing adaptive controlproblems with their solutions for the boundary or the pointcontrol of a linear stochastic distributed parameter systemsis presented. The distributed parameter system is modeledby an evolution equation with an infinitesimal generator foran analytic semigroup. Since there is boundary or pointcontrol, the linear transformation for the control in the stateequation is also an unbounded operator. The unknown pa-rameters in the model appear affinely in both the infinites-imal generator of the semigroup and the linear transforma-tion of the control. Strong consistency is verified for a fam-ily of least squares estimates of the unknown parameters.For a quadratic cost functional of the state and the control,the certainty equivalence control is self-optimizing, that isthe family of average costs converges to the optimal ergodiccost. Another control problem considered here is when thecontrol occurs on the boundary. The “highest-order” opera-tor is assumed to be known but the “lower-order” operatorscontain unknown parameters. Furthermore, the linear op-erators of the state and the control on the boundary containunknown parameters. The noise in the system is a cylindri-cal white Gaussian noise. The performance measure is an er-godic, quadratic cost functional. This time for the identifica-tion of the unknown parameters a diminishing excitation isused that has no effect on the ergodic cost functional but en-sures sufficient excitation for strong consistency. The adap-tive control is the certainty equivalence control for the er-godic, quadratic cost functional with switchings to the zerocontrol.

TM2 (I)Integration of Process Design and ProcessControl

TM2-1 1310

Interaction of Design and Control

Daniel R. Lewin Technion — IIT

Summary: Traditionally, plant controllability and operabil-ity has been considered rather late in the design process,or largely ignored, often leading to poorly performing sys-tems. The indisputable fact that design decisions invariablyimpact on the controllability and resiliency of processes isdriving modern design methods to handle flowsheet con-trollability implicitly in an integrated fashion. This paperdescribes the current state of the art in integration of pro-cess design and process control. A survey of the literaturewould suggest that two alternative approaches could be har-nessed to ensure the controllability and resiliency of chem-ical plants. Controllability and resiliency analysis methodsare used as screening methods relatively early on in the de-sign process. Furthermore, the integrated design and control

paradigms can be applied to fully optimize and integrate thedesign of the process and its operation. It is the objective ofthis presentation is to make a case for the necessary combi-nation of these two approaches.

Keywords: Process design, process control, controllabilityand resiliency assessment, integrated design and control.

TM2-2 1330

Simultaneous Process Design and Process Control:Application to Complex Separation Systems

Roderick Ross Imperial CollegeVikrant Bansal Imperial CollegeJohn D. Perkins Imperial CollegeEfstratios N. Pistikopoulos Imperial College

Summary: The design and control of a literature-baseddouble-effect and an industrial azeotropic distillation sys-tem are considered. Rigorous dynamic modelling is usedto capture the key operability characteristics of each pro-cess. The economic and operational benefits of consideringthe process design and process control tasks simultaneouslyare explored with the aid of advanced dynamic optimizationtechniques. The inclusion of structural decisions into the op-timization is a very challenging area of research. In this re-gard, algorithmic developments are presented which showpotential for the efficient solution of the resulting large-scalemixed-integer dynamic optimization problems.

Keywords: Dynamic modelling, design, process control,dynamic optimization.

TM2-3 1350

Process Design with Complex Nonlinearities

Warren D. Seider Univ. of Pennsylvania

Summary: As process designs move into or closer to re-gimes of complex nonlinearity, they become more sensitiveto model uncertainties and disturbances. This paper exam-ines process designs that operate closer to or within theseregimes often to achieve greater profitability and improvedperformance. Emphasis is placed on developments over thepast decade. Considerations for process controllers in theoperation of such processes are elucidated.

Keywords: Chemical reactors, distillation, chaotic mixing,process design, nonlinear control.

TM2-4 1410

Towards Integration of Controllability into Plant Design

S. Bay Jørgensen Technical Univ. of DenmarkR. Gani Technical Univ. of DenmarkT. R. Andersen Technical Univ. of Denmark

Summary: In process design practice the plant piping andinstrumentation diagram evolves iteratively using mainlyexperience and process reasoning to address questions re-lated to plant controllability. It would be desirable to be ableto address such questions more quantitatively at differentabstraction levels during process design such that controlla-bility evaluation can be integrated into the design process.Only few attempts have been reported towards integratingcontrollability investigations into early stages of plant de-

60

sign. This paper reviews literature on controllability fromthe perspective of controllability assessment with the aim ofidentifying tests, which may be used at different stages ofplant design. First definitions of terms used within processflexibility design and controllability assessment for controlstructure design are given. Methods for controllability eval-uation are reviewed for some modes of process operation.Basically two types of evaluation and design methods pre-vail. One type is based on linear model analysis, whereasanother type is based on physical chemical insight and thusprovides nonlinear information.Control structure development for controllability is illus-trated on an energy integrated distillation plant by usinga heuristic process knowledge based method to develop abasic control structure and subsequently using an optimi-sation based approach for selecting a product purity con-trol structure. Controllability properties of a more energyefficient process design alternative is discussed to illustratethe potential trade off by choice of the most energy efficientdesign that however has the lowest controllability. Basedon the review and the examples a procedure for integratingcontrollability assessment for control structure developmentinto plant design is proposed.

TM2-5 1430

Controllability and Resiliency Analysis for aHeat-Integrated C3-Splitter

Boris M. Solovyev Technion — IITDaniel R. Lewin Technion — IIT

Summary: Controllability and resiliency (C&R) diagnosis iscarried out on an industrial heat-integrated propane/propy-lene distillation column (C3-splitter). The analysis is basedon short-cut dynamic models, which are obtained directlyfrom the steady-state material and energy balances solvedusing a commercial process simulator. The results indicatethat the designed operating point is open loop unstable. Sys-tematic C&R screen-ing of all of the alternative decentral-ized control configurations suggests that the preferable con-trol pairings are in line with current efforts to stabilize theprocess. However, the severe bandwidth limitations due todynamic interactions for the best possible decentralized con-figuration imply that multivariable control is required foradequate performance.

Keywords: Controllability and resiliency assessment, pro-cess design, heat-integrated processes, distillation.

TM2-6 1450

On the Generation of the Most Promising ControlStructure for Large Dimensional Systems

I. K. Kookos Imperial CollegeK. G. Arvanitis National Tech. Univ. of AthensG. Kalogeropoulos Univ. of Athens

Summary: Multivariable controllers have several advan-tages over single loop controllers for multivariable plants.However, in process control applications, decentralized con-trol systems are far more common than any multivariablecontroller. This is due to several characteristics of the de-centralized controller, such as flexibility in operation, fail-ure tolerance and simplified design and tuning, that are par-

ticularly desirable in process control applications. In de-signing decentralized control systems, the first problem thathas to be solved is the problem of control structure selec-tion or input-output variable pairing problem, i.e. whichof the available manipulated variables is to be used in or-der to control each of the controlled variables. In this re-spect, an algorithmic method for variable pairing selectionof large dimensional systems is proposed in this paper. Theproposed method relies on the main properties of the Rela-tive Gain Array (RGA) and of the Relative Interaction Array(RIA) matrices and the concepts of interaction, integrity andstability. It is shown that the minimization of the overallinteraction in multi-input, multi-output large scale systemsunder several stability and structural constraints can be for-mulated either as a Mixed Integer Nonlinear Programmingproblem or as a Mixed Integer Linear Programming problemwhen the RGA or the RIA matrices, respectively, are used asinteraction measures. The proposed method can be readilyapplied to systems with arbitrarily large dimensions, pro-viding a simple and quick solution to the problem of thegeneration of feasible and promising control structures. Inorder to demonstrate the usefulness of the proposed algo-rithm as a rigorous and systematic solution to the input-output variable pairing problem, two large scale industrialproblems, namely the hydrodealkylation of toluene (HAD)process and the Tennessee Eastman problem, are consideredin the paper.

Keywords: Process control, control structure selection, vari-able pairing problem, Relative Gain Array, Relative Interac-tion Array.

TM3 (I)Advances to Meet the Missile GuidanceChallenge at the Verge of the NewMillennium

TM3-1 1310

Integrated Design of Agile Missile Guidance andControl Systems

P. K. Menon Optimal Synthesis Inc.E. J. Ohlmeyer Naval Surface Warfare Center

Summary: Recent threat assessments by the Navy have in-dicated the need for improving the accuracy of defensivemissiles. This objective can only be achieved by enhancingthe performance of the missile subsystems and by findingmethods to exploit the synergism existing between subsys-tems. Traditional approach for missile guidance and controlsystems has been to design these subsystems separately andthen to integrate them together before verifying their perfor-mance. Such an approach does not exploit any synergisticrelationships between these and other subsystems. As a firststep towards the development of integrated design method-ologies, this paper develops a technique for integrated de-sign of missile guidance and control systems.The application of the state dependent Riccati Equation(SDRE) method for integrated guidance/control system de-sign is discussed in this paper. Satisfaction of terminalaspect angle constraints in the guidance/control problemis also discussed. Numerical results using a six degree-

61

of-freedom missile simulation are given. Integrated guid-ance/control systems are expected to result in significantimprovements in missile performance, leading to lowerweight and enhanced lethality. Both of these factors willlead to a more effective, lower-cost weapon system. Inte-grated system design methods developed under the presentresearch effort also have extensive applications in high per-formance aircraft control and guidance systems.

Keywords: Integrated, guidance, autopilot, missile.

TM3-2 1330

Optimal Guidance with Time Delay for ContinuousTime Systems

R. Gitizadeh Taas Israel IndustriesI. Yaesh Taas Israel IndustriesJosef Z. Ben-Asher Technion — IIT

Summary: The guidance problem with continuous time-de-layed control is considered. The interception conflict we aredealing with involves a pursuer continuously measuring hisrelative position and velocity, and who is able to control hisown acceleration subject to a given pure time delay, and anevader who can apply constant maneuvers. The cost func-tion for this optimal control problem includes the control ef-fort and a quadratically weighted version of the miss dis-tance and final relative velocity. Within this setup, severalguidance problems are formulated and analytically solved:proportional navigation (PN), augmented PN (APN), andaugmented optimal rendezvous (AOR). The solution is ob-tained by applying results previously published by the au-thors to the corresponding discrete-time guidance problemsand by using an alternative derivation based on the theoryof continuous-time linear quadratic optimal control with aninput delay. The resulting new guidance laws are comparedto the numerically classical guidance laws. The examplesdemonstrate the advantage of the optimal guidance lawswhich take the delay into account over the classical ones.

Keywords: Optimal, guidance, delay, APN, AOR.

TM3-3 1350

Optimal Guidance Laws with Uncertain Time-of-Flight

Ilan Rusnak RAFAEL — ADA

Summary: The existing optimal guidance laws assume thattime-to-go is known exactly. The time-to-go is usually esti-mated and thus is a random variable. This paper deals withthe issue of optimal guidance with uncertain time-to-go. Aproblem of control of linear discrete systems with unknowntime-to-go is formulated and solved. The solution is appliedto derive guidance laws. The solution depends on the prob-ability density function of the time-of-flight. This guidancelaw has the structure of a rendezvous guidance law wherethe guidance gains are time-dependent and depend on thedistribution of the time-to-go. Examples that demonstratethese dependencies are presented.

Keywords: Guidance law, optimal guidance law, time-to-go, uncertain time-to-go.

TM3-4 1410

Design of Non-Saturating Guidance Systems

Pini Gurfil RAFAEL — ADAMario Jodorkovsky RAFAEL — ADAMoshe Guelman Technion — IIT

Summary: Design of non-saturating guidance systems isconsidered. Assuming linearized kinematics, a proportionalnavigation guidance model is introduced. The missile guid-ance loop discussed contains nonlinearities such as limitedmissile maneuverability, limited acceleration command andconstrained measured line-of-sight angular rate. A novelapproach, based on input-output stability, renders designguidelines that assure operation in the non-saturating re-gion, given the missile-target maneuver ratio. These guide-lines yield a proportional navigation based guidance lawthat assures zero miss distance for any bounded target ma-neuver. It is shown that if the total dynamics of the guidanceloop is designed to be positive real, and the effective propor-tional navigation constant is chosen to be a simple functionof the maneuver ratio, no saturation shall occur. The illustra-tive examples validate the analysis, and show that the newguidance law is robust enough to guarantee a significantperformance improvement even if the design guidelines aresomewhat loosened.

Keywords: Missile guidance, proportional navigation, L∞stability, zero miss distance.

TM3-5 1430

Robust Missile Guidance Law against HighlyManeuvering Targets

Josef Shinar Technion — IITTal Shima Technion — IIT

Summary: Simulation studies of future anti-missile defensescenarios clearly indicated that currently available guidancelaws and estimation techniques are unable to guarantee ahit-to-kill accuracy in the interception of the anticipatedhighly maneuvering targets.In this paper the future interception scenarios of highly ma-neuvering anti-surface missiles are formulated as zero-sumpursuit-evasion games with imperfect information. The so-lution of the perfect information version of the game indi-cates that, if the actual target maneuver is known, a robusthit-to-kill homing accuracy can be guaranteed even withmodest maneuverability and agility advantages. However,in a realistic environment with noise corrupted measure-ments the estimated target maneuver changes are observedwith a delay, leading to a devastating affect on the guaran-teed homing performance.This paper describes the development of a new guidancelaw that explicitly takes into account the estimation delayand compensates for it. Applying this new guidance lawleads to a significant reduction of the guaranteed miss dis-tance and restores the robustness with respect to the actualtarget maneuver. The homing performance of the new guid-ance law was tested by a set of linearized Monte Carlo sim-ulations, showing very promising results.

Keywords: Missile, guidance, estimation, delay, differential

62

game.

TM4Nonlinear Systems 2

TM4-1 1310

From Physical Realizations to Nonlinear Stability,Passivity and Optimality

Michael Margaliot Tel-Aviv Univ.Gideon Langholz Tel-Aviv Univ.

Summary: In this paper we consider the following prob-lem: Given a dynamical system, described by a differentialequation, determine an appropriate measure for the energystored in the system. This problem is of great importancebecause it is related directly to fundamental concepts suchas stability, passivity, and optimality.First, we solve the problem for linear dynamical systems.We realize the linear differential equation using a linear elec-trical circuit, so that the energy stored in the system is justthe energy stored in the circuit’s components. This leads toan intuitive proof of the Routh-Hurwitz stability criterion.In addition, we use the energy balance in the circuit to de-rive passivity and optimality relations.Then, we extend the results to a class of nonlinear systemsby replacing the linear components in the circuit with moregeneral nonlinear ones. We derive explicit storage functionsfor passivity analysis of these nonlinear systems and for theformulation and explicit solution of a novel nonlinear opti-mal control problem.

Keywords: Nonlinear systems, Stability, Passivity, Optimalcontrol.

TM4-2 1330

Nonlinear State Estimation for Rigid Body Motion withLow-Pass Sensors

Henrik Rehbinder KTH, SwedenXiaoming Hu KTH, Sweden

Summary: In this paper we consider the state estimationproblem for the nonlinear kinematic equations of a rigidbody observed under low pass sensors. On the way tosolve that problem, the convergence of a state estimatorfor a generic stable time-varying linear system is shown.The problem is motivated from a walking robot applicationwhere inclinometers and gyros are the sensors used. Weshow that a non local high gain observer exists for the non-linear rigid body kinematic equations and that it under asmall angle assumption is possible to use one inclinometeronly to estimate two angles.

Keywords: Nonlinear state estimation, rigid body motion,linear time-varying systems, exponential observers, incli-nometers.

TM4-3 1350

Control Systems with Actuator Saturation andBifurcations at Infinity

Enrique Ponce Univ. SevillaJavier Aracil Univ. SevillaDaniel Pagano Univ. Federal de Santa Catarina

Summary: It is known that unstable open-loop plants canbe stabilized under constrained controls only locally. To un-derstand this fact, it is shown how bifurcations at infinityare always involved in the stabilizing process. These bifur-cations are easily detected by studying the Nyquist plots.The approach is illustrated with a concrete example of a anti-windup scheme taken from the literature.

Keywords: Nonlinear systems, local and global stability,anti-windup, bifurcations.

TM4-4 1410

An Antiwindup Control Using µ-Synthesis

E. Lu Univ. Bordeaux IB. Bergeon Univ. Bordeaux IS. Ygorra Univ. Bordeaux I

Summary: This paper deals with a systematic approachto design a controller which, on one hand satisfies objec-tives of performance and on the other hand, prevents fromnoxious influence of saturation. These different constraintsleads to an augmented plant on which the optimisation usesµ-synthesis and its D-K iteration procedure, judiciously ini-tialised by a pre-scaling.

Keywords: Input saturation, anti-windup, µ-synthesis, de-scribing function, decoupling.

TM4-5 1430

A High Gain Observer for Robust State FeedbackController

Ahmet Ucar Firat Univ.

Summary: Nonlinear control strategies; high gain feedbackcontrol, Lyapunov Min-Max control, and Variable StructureSystem (VSS) with Sliding Mode Control (SMC), assumethat the system states are available for feedback. Howeversome system states are often contaminated by high levels ofnoise and constraining the system performance such as thevelocity measurement in electromechanical systems. There-fore it is important to design those control strategies by uti-lizing only output feedback rather than full state feedback.A high gain observer is introduced in this paper to recon-struct unmeasurable states for nonlinear feedback controlstrategies. This observer developed based on the teory ofsingular perturbation and variable structure systems. Thevalidity its performance shown through numerical examplein the presence of uncertainty in the system in order to im-plement such designs to the electromechanical systems.

Keywords: High-gain observer, output feedback control,sliding mode control, robot manipulators.

63

TM4-6 1450

On the Role of Invariance in the Theory of Systemsand Control — An Intelligible Introduction for theBeginners

Masasuke Shima Hokkaido Univ.

Summary: A framework of the introductory course to thecontrol system theory is presented from the viewpoint ofnonlinear theory. In the first part, the description is based onthe input-output relations of the input affine systems. Thenotions appearing first are the relative order and the invari-ance. Geometric notions are gradually introduced for thebeginners. Feedback design methods are given for the non-linear systems and then for the linear systems. In the sec-ond part, the Hamiltonian formulation and the variationalmethod are used to derive conditions of the optimality andthe invariance. Both parts are linked via the notion of invari-ance. This style is partly realized in the author’s lectures.

Keywords: Introductory course, input-output expression,invariance, Hamiltonian formulation, optimality.

TM5Identification 1

TM5-1 1310

A Directional Forgetting Algorithm Based on theDecomposition of the Information Matrix

Liyu Cao Carleton Univ.Howard M. Schwartz Carleton Univ.

Summary: A novel directional forgetting algorithm is de-veloped based on a decomposition of the information ma-trix. This algorithm performs forgetting only to a specifiedpart of the information matrix, thus preventing the problemknown as estimator windup which is a characteristic of thestandard exponential forgetting algorithm. This algorithmis able to track fast parameter changes and is similar in com-plexity to the standard least square algorithm. The superiorperformance of the algorithm is verified via theoretical andsimulation studies.

Keywords: Parameter estimation, recursive algorithm, ex-ponential forgetting, directional forgetting.

TM5-2 1330

A Parameter Estimation Method for a Special Class ofSystems of Ordinary Differential Equations

Carla Seatzu Univ. of Cagliari

Summary: In this paper a special class of systems of ordi-nary differential equations is considered. This class is par-ticularly common both in biological and medical field and isdenoted as S-Systems.The problem we deal with is the estimation of unknownparameters in a system of equations, when a set of obser-vational data is available. The procedure we now proposearises from the requirement of overcoming the main diffi-culties typical of iterative gradient based methods. The main

idea of the method is that of approximating each state vari-able by a fitting process and then splitting the overall esti-mation problem into a set of simpler independent problems,thus lessening the difficulty concerning high parameter vec-tor dimensions. Each problem consists of the minimizationof a differential residual and in particular cases reduces tothe solution of an overdetermined linear algebraic system.It is also possible to take into account the parameter con-straints with a modest computational effort.The results of several numerical simulations are also pre-sented. The robustness of the method has also been testedaffecting data with different noise levels.

Keywords: Parameter estimation, S-Systems, B-splines,data fitting.

TM5-3 1350

An Algorithm for Control System Loop GainIdentification

Meir Pachter Air Force Inst. of Technology

Summary: The identification of a linear discrete-time con-trol system’s loop gain is addressed. The classical Kalmanfilter theory for linear control systems is extended and thecontrol system’s state and loop gain are jointly estimated.Explicit formulae for the loop gain’s unbiased estimate andestimation error covariance are derived.

Keywords: System identification, Kalman filtering, adap-tive control.

TM5-4 1410

Real-Time Identification Using a Classical NonlinearOptimization Algorithm and the Flatness Properties ofa System: Application to an Intensity/PressureConverter

Augustin Sanchez National Inst. Appl. Sc., ToulouseVincent Mahout National Inst. Appl. Sc., Toulouse

Summary: This article presents a modification of a classicalnonlinear identification algorithm, which permits then on-line identification (or real-time identification). This modifi-cation is particularly based on using nonlinear optimizationalgorithm, not with a classical model of the process, whichneeds a numerical integration algorithm to solve it, but withthe flatness properties of the model of the process (Fliess etal., 1995). The states of the system are then obtained withoutany integration, which follows a significant saving of cal-culation time. This modified method is applied to on-lineidentification of the parameters of a nonlinear model of anintensity/pressure converter (i/p converter), used to sup-ply air pressure inside an artificial pneumatic muscle usedlike actuator on the robots of the laboratory. To illustrate themethod, experimental results are given and discussed.

Keywords: Real-time identification, nonlinear program-ming, flatness, intensity/pressure converter.

64

TM5-5 1430

Stopping of Algorithms and Faults Detection inKalman Filter Application

Chingiz Hajiyev Istanbul Technical Univ.

Summary: An approach to the generation of stopping rulesin parametric identification problems is proposed on the ba-sis of the computation of a statistic of the difference betweentwo successive estimates. This statistic is also used for faultdetection in the Kalman filter.The developed decision rules are applied to a linear systemidentification problem. The domain of possible utilizationof the Kalman filter is partitioned into three zones accord-ing to presented rule. When the value of above mentionedstatistic lies between the confidence limits, the decision ismade to continue the estimation. When its values attain theconfidence limits, the decision is made to stop the estima-tion process and, accordingly, the kind of corrective actionsin the estimation process is decided.The stopping rule developed here has the advantage that itsapplication does not require the specification of an admis-sible error ellipsoid, whose construction represents an inde-pendent problem.Experimental results are presented to demonstrate the per-formance of the proposed algorithms.

Keywords: Kalman Filter, Stopping Rule, System Identifi-cation, Fault Detection, Decision rule.

TM5-6 1450

Estimation Variance is not Model StructureIndependent

Brett Ninness Univ. of NewcastleHakan Hjalmarsson Royal Inst. of TechnolgyFredrik Gustafsson Linkoping Univ.

Summary: This paper establishes that when using a leastsquares criterion to estimate an output error type modelstructure, then the measurement noise induced variability ofthe frequency response estimate depends on the estimated(and hence also on the true) pole positions. This dependenceon pole position is perhaps counter to prevailing wisdomthat for any ‘shift invariant’ model structure, the variabilitydepends only on model order, data length, and input andnoise spectral densities. That is, it is counter to the beliefthat variance error is model-structure independent.

Keywords: Output error identification, estimation algo-rithms, estimation theory, identification algorithms, fre-quency domains, modelling errors.

TP1 (I)Dynamical Models

TP1-1 1530

Stability, Euler Approximations of Dynamical Systemsand Fixed Point Iterations

Elza Farkhi Tel-Aviv Univ.

Summary: Consider an infinite time autonomous dynami-cal system:

x(t) ∈ F(x(t)) for a. e. t ∈ (0,∞), x(0) = xo,

where F is an upper semi-continuous set-valued mappingwith convex compact images, from R

n to itself. Lookingfor stationary (equilibrium) points of this system is equiv-alent to looking for fixed points of the Euler map Gh(x) =x + hF(x) for each scalar h > 0. Assume that F is boundedon bounded sets and satisfies the following one-sided Lips-chitz condition (OSL) (weak monotonicity condition) with aconstant L < 0:For each x, y ∈ X and each u ∈ F(x) there is v ∈ F(y) suchthat

〈x− y, v−w〉 ≤ L‖x− y‖2.

OSL condition may be formulated also in a Hilbert space ora Banach spaces with an uniformly convex conjugate space.Generally, OSL dynamical systems have more than one tra-jectory, which differs from the classical monotonicity case.The above assumptions imply, from one side, nonemptinessand uniform boundedness on the infinite time interval ofthe trajectories and the equilibrium points sets, and, fromthe other, set-valued asymptotic stability of the attainableset and of the trajectories set.We study the following explicit Euler iterative process, ap-proximating the continuous dynamical system:

yk+1 ∈ Ghk(yk) = yk + hkF(yk), y0 = yo,

k = 0, 1, . . . The iteration is stationary if hk = h0 for a fixedh0, or nonstationary with

∑∞k=0 hk =∞,

∑∞k=0 h

2k <∞.

It is shown that in our case the multifunction Gh is almostcontractive, while for F Lipschitz and negatively monotoneit is strictly contractive. On this base convergence estimatesare proved for stationary and nonstationary Euler iteration,which extend known results for classical monotone map-pings.

Keywords: Fixed points, monotone mapping, one sidedLipschitz, stability, differential inclusions, Euler method.

TP1-2 1550

Asymptotic Behavior of Infinite Products ofOrder-Preserving Mappings in Banach Space

Simeon Reich Technion — IITAlexander J. Zaslavski Technion — IIT

Summary: In this paper we present several results concern-ing the asymptotic behavior of (random) infinite productsof generic sequences of order-preserving mappings on inter-vals and cones in an ordered Banach space. Such operators

65

find application in many areas of mathematics and, in par-ticular, in dynamical models of economics and biology. Inaddition to weak ergodic theorems we also obtain conver-gence to a unique common fixed point (for self-mappings ofan interval) and to an operator of the form f(·)η, where f isa functional and η is a common fixed point. More precisely,we show that in appropriate complete metric spaces of se-quences of operators there exists a subset which is a count-able intersection of open everywhere dense sets such thatfor each sequence belonging to this subset the correspond-ing infinite products converge. Thus, instead of consider-ing a certain convergence property for a single sequenceof operators, we investigate it for a space of all such se-quences equipped with some natural metric, and show thatthis property holds for most of these sequences. This allowsus to establish convergence without restrictive assumptionson the space and on the operators themselves.

Keywords: Fixed point, generic property, ordered Banachspace, order-preserving mapping, random infinite product.

TP1-3 1610

Exponential Stabilization of Vibrating Systems byCollocated Feedback

George Weiss Imperial CollegeRuth F. Curtain Univ. of Groningen

Summary: We consider regular linear systems described byx = Ax + Bu, y = B∗Λx, where A generates a strongly con-tinuous semigroup on the Hilbert space X and A is essen-tially skew-adjoint and dissipative. This means that the do-mains of A∗ and A are equal and A∗ + A = −Q, whereQ is a bounded nonnegative operator. The control opera-tor B is possibly unbounded, but admissible and B∗Λ is theΛ-extension of B∗. Such a description fits many wave andbeam equations and it has been shown for many particularcases that the feedback u = −κy, with κ > 0, stabilizes thesystem, strongly or even exponentially. We show, by meansof a counterexample, that if B is sufficiently unbounded,then such a feedback may be unsuitable: the closed-loopsemigroup may even grow exponentially. However, if κ issufficiently small, and if the original system is exactly con-trollable and observable, then the closed-loop system is ex-ponentially stable. The above assumptions may be relaxedin various directions, for example, regularity may be re-placed by well-posedness, exact controllability may be re-placed by optimizability etc.

Keywords: Well-posed linear systems, positive transferfunctions, exact controllability and observability, skew-adjoint operators, collocated sensors and actuators.

TP1-4 1630

A Composite Semigroup for the Infinite-DimensionalDifferential Sylvester Equation

Zbigniew Emirsajlow Technical Univ. of Szczecin

Summary: This paper presents a certain approach to thestudy of the operator differential Sylvester equation whicharises in various control problems on finite time horizon. Acrucial role in this approach is played by the so-called com-posite semigroup. It is a strong-operator continuous semi-group defined on a Banach space of linear bounded oper-

ators obtained as a composition of two ‘classical’ stronglycontinuous semigroups defined on a Hilbert space. We in-vestigate basic properties of the solution to this equation inthe case when the operators occuring in in the equation areunbounded.

Keywords: Sylvester differential equation, composite semi-group.

TP1-5 1650

Input-Output Stability of Systems Governed byNonlinear Second Order Evolution Equations inHilbert Spaces

Michael I. Gil’ Ben-Gurion Univ.

Summary: We consider systems governed by nonlinear sec-ond order evolution equations in a Hilbert space and estab-lish explicit conditions for the input-output stability.

Keywords: Infinite dimensional systems, second order evo-lution equations, input-output stability.

TP2Process Control

TP2-1 1530

Wood Chip Refiner Control

Ahmed A. Ismail Univ. of British ColumbiaGuy A. Dumont Univ. of British Columbia

Summary: On a chip refiner the gain of the transfer func-tion between the refiner motor load and the plate gap is bothnonlinear and time-varying, with reversal in the sign of thegain indicating the onset of pulp pad collapse towards lowervalues of the plate gap. The control objective is to regulatethe motor load while avoiding pad collapse. The problem isprincipally stochastic in nature, since the gap at which gainreversal occurs can wander unpredictably. An active subop-timal dual controller is designed to control the motor load bymanipulating the plate gap. It uses an adaptive Kalman fil-ter to track both slow drifts and sudden sign changes in thegain. The controller minimizes a myopic nonlinear perfor-mance index designed especially to reflect the peculiaritiesof the process. Thus, no heuristic logic is needed. Simula-tions show the superior performance offered by this strat-egy.

Keywords: Dual control, adaptive control, time-varyingsystems, adaptive Kalman filtering, pulp industry.

TP2-2 1550

Automatic Tuning of the Window Size in the Box CarBackslope Data Compression Algorithm

Jens Pettersson Royal Inst. of TechnologyPer-Olof Gutman Technion — IIT

Summary: In process industries, as well as in other con-trolled processes, there is a need to record and store forposterity measurement data from a large number of sensorsin such a way that the exact time and size of a significantchange can be retrieved and analyzed. Even with today’s

66

computer memories, a short sampling interval will yield anunwieldy amount of data. Therefore various data compres-sion algorithms are used, one of which is the Box Car Back-slope (BCBS) filtering algorithm which does not store datawithin a preset window around the current prediction.The window size should be chosen such that normalmeasurement noise is ignored but significant changes arerecorded. Until now the window size of each data chan-nel had to be tuned manually. In this paper the BCBS algo-rithm is explained and a novel algorithm for the automaticon-line tuning of the window size is suggested, based on theminimization of a criterion weighing the data reduction rate,and the variance of the error between filtered and measureddata.The algorithm was found to work very well in a paper man-ufacturing plant.

Keywords: Data compression, process control, time series.

TP2-3 1610

Experimental Tests of Digital Filters for Control of aPilot-Scale Batch Distillation Column

Ronia M. Oisiovici Univ. Estadual de CampinasSandra L. Cruz Univ. Estadual de CampinasJoao A. F. R. Pereira Univ. Estadual de Campinas

Summary: Measurement signals that a computer uses totake control actions are usually contaminated by noise. Thepresence of noise is undesirable because it may be detrimen-tal to the operational control. In order to reduce the noiselevel in a batch distillation column control loop, two digi-tal filters were experimentally tested: the double exponen-tial filter and the moving average filter. The moving aver-age was less effective than the double exponential filter. Theprofile of the controlled variable was smoother but the con-trol actions were more delayed when the moving averagewas used. The runs have shown how important the choiceof the digital filter is to achieve a good control performance.For control tasks, this choice must be a compromise betweendata smoothing and the ability to respond rapidly to realchanges in the process.

Keywords: Batch distillation, digital filtering, computercontrol.

TP2-4 1630

A Linear Time-Varying State-Space Model of BatchDistillation Columns for Control Applications

Ronia M. Oisiovici Univ. Estadual de CampinasSandra L. Cruz Univ. Estadual de Campinas

Summary: Batch distillation is widely used in the produc-tion of fine chemicals, which must be manufactured accord-ing to high and well-defined standards of purity. However,due to the strongly nonlinear and time-varying behaviourof batch distillation columns, the composition control is notmerely a task but a real challenge. One of the first stepstowards control is developing a mathematical model of theprocess of interest. A rigorous model is not always appro-priated for on-line control tasks, especially for batch sys-tems, which are characterised by frequent changes in pro-cess conditions. In this work, a linear time-varying state-

space model for batch distillation columns was developedand tested. The model is suitable for on-line implementationand to predict the system behaviour from measurable andeasily available information. Comparing the model predic-tions with rigorous simulation results, the state-space modelwas able to predict the batch distillation column behaviouraccurately, even for the nonideal mixture ethanol-water.

Keywords: State-space models, batch distillation, computercontrol.

TP2-5 1650

Neuro-Fuzzy Modeling in Petrochemical Industry

Maide Bucolo Univ. di CataniaSalvatore Graziani Univ. di CataniaLuigi Fortuna Univ. di CataniaMario Sinatra ERG Petroli Siracusa

Summary: In the last few years, problems concerning withboth air pollution and quality of products have gained a par-ticular attention in industrial companies. A great interestin new technologies for the process of manufacturing opti-mization and quality control has raised. Mathematical mod-els for quality control are highly nonlinear and need veryexpensive and sophisticated instruments. Soft-Computing,an innovative approach for constructing computationallyintelligent systems, has just come into the limelight. Thequintessence of designing intelligent systems of this kind isNeuro-Fuzzy computing. In this paper a Neuro-Fuzzy pre-diction model for the quality control of benzene is proposed.

Keywords: Quality control, petrochemical industry, model-ing, soft-computing.

TP3Aerospace Control

TP3-1 1530

An Integrated Algorithm for Path Planning and FlightController Scheduling for Autonomous Helicopters

Magnus Egerstedt Royal Inst. of TechnologyT. K. John Koo Univ. of California, BerkeleyFrank Hoffmann Univ. of California, BerkeleyShankar Sastry Univ. of California, Berkeley

Summary: This article investigates the problem of generat-ing optimal flight trajectories for an autonomous helicopter.We propose a planning strategy that partitions the optimiza-tion problem into isolated segments. Given a set of nom-inal waypoints we then generate trajectories that interpo-late close to these points. This path generation is done fortwo different cases, corresponding to the two flight con-trollers that either govern position or velocity of the heli-copter. Based on a given cost functional, the planner selectsthe optimal one among these multiple paths. This approachthus provides a systematic way for generating not only theflight path, but also a suitable switching strategy, i.e. whento switch between the different controllers.

Keywords: Optimal control, path planning, unmannedaerial vehicles.

67

TP3-2 1550

Actuator Design for Aircraft Robustness VersusCategory II PIO

Francesco Amato Univ. of NaplesRaffaele Iervolino Univ. of NaplesStefano Scala Italian Aerospace Research CentreLeopoldo Verde Italian Aerospace Research Centre

Summary: In this paper we deal with the analysis of Pilot inthe Loop Oscillations (PIO) of Category II (with rate and po-sition limiting), a phenomenon usually due to a misadapta-tion between the pilot and the aircraft response during sometasks in which tight closed loop control of the aircraft is re-quired from the pilot, with the aircraft not responding to pi-lot commands as expected by the pilot himself. We proposean approach, based on robust stability analysis, which as-sumes that PIO are characterized by a limit cycle behaviour.In this approach the nonlinear elements are substituted byfictitious linear parameters, which can be considered time-invariant or time-varying; in this way we obtain two criteriafor robustness versus Category II PIO. If, using the proposedcriteria, the aircraft under consideration is shown to be Cat-egory II PIO prone, since limit cycles occurrence is due to abad design of the nonlinear actuators, we propose an algo-rithm which, taking into account the trade-off between real-ization costs and performances, provides the guidelines forthe design of actuators which should guarantee robustnessversus Category II PIO. Finally, to demonstrate the use ofthe new proposed method, we apply our technique to a casestudy, namely the X-15 aircraft PIO, occurred on June 8, 1959during a landing flare.

Keywords: Robust stability, nonlinear systems, pilot in theloop oscillations (PI0), actuator design.

TP3-3 1610

On Algorithms for Attitude Estimation Using GPS

Itzhack Y. Bar-Itzhack Technion — IITAssaf Nadler Technion — IIT

Summary: This paper discusses algorithms for attitude de-termination using GPS differential phase measurements, as-suming that the cycle integer ambiguities are known. Theproblem of attitude determination is posed as a parame-ter optimization problem. One proposed set of optimal so-lutions, which includes solutions of Wahba’s problem, isbased on least squares fit of some attitude parameters to aset of vector measurements. The use of these algorithmsrequires the conversion of the basic GPS scalar phase mea-surements into unit vectors. Another possible approach isbased on a least squares fit of the attitude quaternion to theGPS phase measurements themselves. The cost function ofthe fit is given in the literature in the most straightforwardformulation as a function of the attitude matrix. The pa-per presents the conversion of the matrix-based cost func-tion to a quaternion-based cost function, which correspondsto the cost function minimized by QUEST. However, unlikethe QUEST cost function, the converted cost function is not asimple quadratic form, therefore the simple QUEST solutionis not applicable in this case. Three iterative solutions forfinding the optimal quaternion are derived. The first algo-

rithm is a linearly convergent one whose convergence rateis slow. The other two converge very fast. The algorithmspresented in this paper can handle cases of planar antennaarrays and thus cover a deficiency in earlier algorithms. Theefficiency of the new algorithms is demonstrated throughnumerical examples.

TP4Time Delay Systems

TP4-1 1530

Every Stabilizing Dead-Time Controller has anObserver-Predictor-Based Structure

Leonid Mirkin Technion — IITNatalya Raskin Technion — IIT

Summary: This paper considers the stabilization problemfor systems with a single delay h in the feedback loop. Thestate-space parametrizations of all stabilizing regulators arederived. These parametrizations have simple structures andclear interpretations. In particular, it is shown that everystabilizing controller consists of a delayed state observer xo,an h time units ahead predictor xp, and a stabilizing statefeedback, i.e.:

xo(t) = Axo(t) + Bu(t− h) − L(y(t− h) − Cxo(t)

)xp(t) = e

Ahxo(t) +

∫tt−h

eA(t−τ)

Bu(τ)dτ

u(t) = Fxp(t) + υ(t),

where F and L are any matrices so that A + BF and A + LCare Hurwitz and υ = Qε, where ε(t) = y(t− h) −Cxo(t) isthe innovation andQ(s) ∈ H∞ but otherwise is arbitrary.Applications of the proposed parametrization to the H2 op-timal control and the robust stabilization of dead-time sys-tems are discussed.

Keywords: Delay systems, parametrization, H2 control, ro-bust stability.

TP4-2 1550

The Structure at Infinity of Linear Delay Systems andthe Row-by-Row Decoupling Problem

Rabah Rabah Inst. de Rech. en Cyb. de NantesMichel Malabre Inst. de Rech. en Cyb. de Nantes

Summary: For linear finite dimensional systems with trans-fer function matrix T(s), the behavior of T(s) at infinity playsan important role in several control problems. This behaviorcan also be interpreted as that of the system at time t = 0,and may be described by the so-called structure at infinityor the canonical form at infinity. As, in this case, the matrixT(s) is rational and strictly proper, the computation of thecanonical form is easy and this form is invariant under statefeedback. This allows to characterize the existence of solu-tions to the model matching, the disturbance rejection andthe row-by-row decoupling problems. For linear time de-lay systems with transfer function matrix T(s, e−s) we con-sider the row-by-row decoupling problem. We use the con-cept of weak structure at infinity (the variable s is real) andstrong structure at infinity (s is a complex variable). This al-

68

lows to design a decoupling precompensator which is weakbiproper. This precompensator is then realized by general-ized static state feedback which can include delayed deriva-tives of the new control, and thus requires some smoothnessof the new control. If the new control is not smooth enough,then the decoupling problem cannot be solvable by general-ized static state feedback.

Keywords: Linear time delay systems, structure at infinity,decoupling problem.

TP4-3 1610

Stabilization of Singularly Perturbed Linearly Systemswith Delay and Saturating Control

Achim Ionita National Inst. of Aerospace Res.Vasile Dragan Romanian Academy

Summary: The paper deals with the feedback control lawdesign methodology that applies to singularly perturbedlinearly systems with time-delay control under saturationconstraints. The results obtained by a scalar inequalities al-lows us to investigate a variety of control problems.

Keywords: Singular perturbed systems, input-delay, satu-ration, stabilizing feedback gain.

TP4-4 1630

Near Optimal PLL Design for Decision FeedbackCarrier and Timing Recovery

Oded Yaniv Tel-Aviv Univ.Dan Raphaeli Tel-Aviv Univ.

Summary: A new design method is presented for the de-sign of PLL loop filters for carrier recovery, bit timing orother synchronization loops given phase noise spectrumand noise level. Unlike the conventional designs, our de-sign incorporates a possible large decision delay and S-curveslope uncertainty. Large decision delays frequently exists inmodern receivers due to, for example, a convolutional de-coder or an equalizer. The new design also applies to coher-ent optical communications where delay in the loop limitsthe laser line width. We provide an easy to use completedesign procedure for second order loops. We also introducea design procedure for higher order loops for near-optimalperformance. We show that using the traditional second or-der loop is suboptimal when there is a delay in the loop,and also show large improvements, either in the amount ofallowed delay, or the phase error variance in the presence ofdelay.

Keywords: PLL, delay, margin, optimal, QFT, feedback car-rier.

TP5Identification 2

TP5-1 1530

Modelling and Identification of a High TemperatureShort Time Pasteurization Process Including Delays

Carlos F. Alastruey Public Univ. of NavarraManuel De la Sen Univ. of the Basque CountryMario Garcia-Sanz Public Univ. of Navarra

Summary: In this paper, an improved mathematical modelfor a High Temperature Short Time (HTST) pasteurizationplant is proposed. The main differences from previous mod-els are that the four interconnected blocks of the heat ex-changer model are assumed to be of third order; thereforefundamental physical properties of the plant are not ne-glected. In addition, time delays at the output of the heat ex-changers are considered in order to take into account the factby which the temperature sensor is not physically within theheat exchanger. Following the proposed model, a parameteridentification procedure is suggested, by using stable filter-ing for the input and output signals.

Keywords: Time-delay, interconnected systems, pasteuriza-tion process, parameter identification.

TP5-2 1550

Parameter Identification In Nonlineaer Systems UsingHopfield Neural Networks

Zhenning Hu Univ. of Missouri-RollaS. N. Balakrishnan Univ. of Missouri-Rolla

Summary: In this study, we use Hopfield neural network(HNN) for identifying parameters of a nonlinear system. Weuse a linearization process and develop the equations fora parameter identification algorithm. we use a scalar timevarying problem and a complex nine-state nonlinear prob-lem to demonstrate the potential of this method.

Keywords: Parameter estimation, nonlinear, neural net-works.

TP5-3 1610

Optimal Combination of Identification and Control forBounded-Noise ARX Systems

Andrzej Krolikowski Technical Univ. of Poznan

Summary: An optimal combination of sequential identifi-cation and control for linear bounded-input bounded-noisediscrete-time ARX system is considered. Various configura-tions of identyfy-ing/ controlling sequences are investigatedin order to find an optimal trade-off. The tracking control ofsecond-order system is simulated showing an optimal trade-off between identifica-tion and control periods.

Keywords: Identification, control, optimal trade-off, ARXsystem, bounded-input, bounded-noise.

69

TP5-4 1630

Closed-loop model-free subspace-based LQG-design

Wouter Favoreel Katholieke Univ. LeuvenBart De Moor Katholieke Univ. LeuvenMichel Gevers Univ. Catholique de LouvainPeter Van Overschee Katholieke Univ. Leuven

Summary: When only input/output data of a system areavailable the classical way to design a linear quadratic Gaus-sian controller consists of mainly three separate parts. Firsta system identification step is performed to find the systemparameters. With these parameters a Kalman filter is de-signed to find an estimate of the state of the system. Fi-nally, this state is then used in an LQ-controller. In theliterature these three steps are hardly ever considered asone joint identification/control problem. In a previous pa-per it was shown that, based on techniques from the fieldof subspace system identification, the three steps of theLQG-controller design can be replaced by a QR and a SV-decomposition. A drawback of the method is that the in-put and output data available for the LQG-design must beretrieved in open loop. In the present paper, a generaliza-tion of the results previously presented results to the casewhere the data is measured on a system working in closed-loop. It is shown that under mild conditions the closed-loopsubspace-based controller and the classical LQG-controllerare equivalent. The effectiveness of the method is illustratedby the hand of a simulation example. It is shown that theopen-loop subspace-based LQG-controller gives biased re-sults whereas the closed-loop version converges to the clas-sical LQG-controller when the length of the backward hori-zon increases.

Keywords: Subspace identification, identification for con-trol, LQG-control, Kalman filter, closed-loop.

TP5-5 1650

Measurement of Impedance Characteristics ofComputer Keyboard Keys

Mark Nagurka Marquette Univ.Richard Marklin Marquette Univ.

Summary: The aim of this project is to gain a more com-plete understanding of the tactile “feel” of computer key-board keys by quantifying their mechanical impedance. Toachieve this goal, a computer-controlled test rig that canmeasure computer key displacement, velocity, and contactforce has been designed, constructed and tested. This pa-per describes the hardware and software configuration, in-cluding the data acquisition method and motion control sys-tem. Preliminary results show that the key dissipates energyduring a depression-return stroke, indicating the presence ofdamping.

Keywords: Computer keyboard, impedance, damping, er-gonomics.

WEDNESDAY, June 30

WA1 (I)Optimal Control

WA1-1 1000

Structure of Optimal Solutions of Infinite DimensionalControl Problems

Alexander J. Zaslavski Technion — IIT

Summary: In this paper we present several results con-cerning the structure of optimal solutions for infinite-dimen-sional optimal control problems. The primary area of appli-cations of these problems concerns models of regional eco-nomic growth, cattle ranching models and systems with dis-tributed parameters and boundary controls arising in cer-tain engineering applications. We are concerned with theexistence of an overtaking optimal trajectory over an infi-nite horizon. The existence result that we obtain extends theresult of Carlson, Haurie and Jabrane (1987) to a situationwhere the trajectories are not necessary bounded. We showthat an optimal trajectory defined on an interval [0, T ] is con-tained in a small neighborhood of the optimal steady-state inthe weak topology for all t ∈ [0, T ] \ E where E ⊂ [0, T ] isa measurable set such that the Lebesgue measure of E doesnot exceed a constant which depends only on the neighbor-hood of the optimal steady-state and does not depend on T .Moreover, we show that the set E is a finite union of intervalsand their number does not exceed a constant which dependsonly on the neighbborhood.

WA1-2 1020

Lipschitz Stability of Solutions to Parametric OptimalControl for Parabolic Equations

Kazimierz Malanowski Polish Academy of Sc.Fredi Troltzsch Technische Univ. Chemnitz-Zwickau

Summary: Let H be a Banach space of parameters and G ⊂H an open set of feasible parameters. Moreover, let Ω ∈Rn be a bounded domain with regular boundary ∂Ω and

let T > 0 be a fixed time. Denote Q = Ω × (0, T) and Σ =∂Ω× (0, T). For each h ∈ G consider the following optimalcontrol problem for semilinear parabolic equation

(Ph) Find (yh, uh) ∈ C(Q)× L∞(Q) such that

Jh(yh, uh) = miny,u

Jh(y, u) :=

∫Q

ψ(x, t, y, u, h)dxdt

subject to

yt(x, t) +Ay(x, t) + a(x, t, y(x, t), u(x, t), h) = 0 inQ,∂νy(x, t) + b(x, t, y, h) = 0 in Σ,

y(x, 0) − χ(x) = 0 inΩ,

and

r(x, t) ≤ u(x, t) ≤ s(x, t) a. e. inQ,

where A is an elliptic operator, ν is the outword normal to∂Ω, ψ, a, b, r, s are given functions. All data functions areas smooth as it is needed.

70

We assume that for a given reference value h0 ∈ G of theparameter, problem (Ph0 ) has a solution (y0, u0) and we aregoing to characterize conditions under which this solutionis stable in the following sense:

(S) There exist neighborhoodsG0 and Z0 of h0 and (y0, u0), re-spectively, such that, for each h ∈ G0 there is a unique in Z0solution (yh, uh) of (Ph), which is a Lipschitz continuousfunction of h.

The main difficulty in solving (S) is connected with the non-smoothness introduced by the presence of inequality con-trol constraints. This nonsmoothness makes it impossibleto use the classical implicit function theorem. Instead ofthat, Robinson’s (1980) implicit function theorem for gener-alized equations (inclusions) is used. This theorem allows toreduce the stability analysis for the original problem (Oh),to such an analysis for a linear-quadratic accessory problem,subject to additive perturbations. It was shown by Troltzsch(1999) that the needed stability of the solutions to the acces-sory problem is satisfied if the so called strong coercivity con-dition holds with a certain margin of freedom. By Robinson’stheorem this condition is a sufficient condition of (S). Us-ing a generalization of Robinson’s theorem due to Dontchev(1995), we show, that this condition is also necessary, pro-vided that the dependence of data on the parameter is suffi-ciently strong.

WA1-3 1040

Optimal Control of Differential Inclusions InvolvingPartial Differential Operators

Alexander Ioffe Technion — IIT

Summary: We shall consider optimal control problem forsystems governed by operator inclusions of the form Lx ∈F(x), where F is a set valued mapping from Rn into itself(with possibly unbounded and non-convex values), L is adensely defined closed linear operator in a suitable spaceof mappings in Rn (typically one or another Sobolev space)with compact inverse and solutions of the inclusion are un-derstood in the weak sense. The main content of the talk isconnected with application of methods of non-smooth anal-ysis in the theory of necessary conditions (maximum princi-ples).

WA1-4 1100

On the Existence of Optimal Strategies for MultichainMarkov Decision Processes

Arie Leizarowitz Technion — IIT

Summary: In this work we address the question of whethera given finite state MDP has a Markov stationary optimalstrategy. The optimality criterion which we consider is thelong run average cost, and the action set is a compact metricspace. We consider general multichain MDPs and provide asharp answer to this question, namely conditions which arenecessary and sufficient for existence of optimal solutions.A consequence of this characterization is a procedure withthe following property. When applied to a given MDP theneither the procedure yields an optimal strategy, or else it in-dicates that such an optimal strategy does not exist. In caseoptimal strategies do exist this procedure yields a detailed

description of the structure of optimal strategies rather thanmerely establishing their existence.

WA1-5 1120

Feedback Control for Descriptor Systems

Galina A. Kurina Voronezh State Forestry Academy

Summary: The control in the feed-back form is obtained forfive linear-quadratic optimal control problems in a Hilbertspace with the state equation unresolved with respect to thederivative, namely, the problem with the quadratic or linear-quadratic performance index, the problem with fixed points,the periodic problem, the regulation problem on an infiniteinterval. For the first four problems the operator K(t) whichis the solution of the following differential operator Riccatiequation

ddt

(A ′K(t)) = −K ′(t)C(t) − C ′(t)K(t) −W(t)

+ (K ′(t)B(t) + S(t))R−1(t)(K ′(t)B(t)S(t)) ′

is used under different conditions. Here A is the opera-tor standing before the derivative in the state equation, theprime with a notation of an operator denotes the conjugateoperator. For the solving of the fifth problem the solution Kof the operator Riccati equation

K′C+ C ′K− (K ′B+ S)R−1(K ′B+ S) ′ +W = 0

satisfying the symmetry condition

A′K = K

′A

is used. In the present paper it is not necessary to selectfrom the state equation an equation resolved with respect tothe derivative as it was made in numerous works devotedto linear-quadratic control problems for descriptor systems.Besides, the form of relations, defining the control in thefeed-back form, is identical both for a singular operator,standing before the derivative, and for a nonsingular oper-ator, that is very convenient is a research of singularly per-turbed control problems. For the entry of the control in thefeed-back form the operator is used which is the solution ofthe operator Riccati equation and it acts in all state space,unlike in a subspace in the case of a singular operator beforethe derivative, as it was in previous works of other authors.In contrast to previous works of other authors the regularityof the pencil of the operators from the state equation is notrequired.

Keywords: Feedback control, descriptor systems.

WA2Manufacturing

WA2-1 1000

Optimal Design of Transfer Lines and MultipositionMachines

Alexandre Dolgui Univ. of Technology of TroyesNikolai N. Guschinsky Academy of Sc. of BelarusGenrikh M. Levin Academy of Sc. of Belarus

Summary: Automatic transfer lines and multi-positionequipment for machining large parts in mass production are

71

investigated. Mathematical models and methods for opti-mal cost design of these production systems are considered.The following designed system parameters are determined:number of workstations; assignment of operations to theworkstations; orientation of the parts, number of positionsand cutting modes for each workstation. For solving the ob-tained optimization problem, a special multilevel decompo-sition scheme is proposed. It uses the decomposition ap-proaches in combination with the methods of nonlinear anddiscrete programming.

Keywords: CAD/CAM, automatic transfer lines, optimiza-tion, parametric decomposition.

WA2-2 1020

Conrol Architecture of a Flexible Microrobot-BasedMicroassembly Station

Sergej Fatikow Univ. of KarlsruheJ. Seyfried Univ. of Karlsruhe

Summary: The assembly of complex microsystems con-sisting of several single components (i.e., hybrid microsys-tems) is a task which has to be solved to make mass pro-duction of microsystems possible. Therefore, it is necessaryto introduce flexible, highly precise and fast microassemblymethods. In this paper, the control system of a microrobot-based microassembly desktop station that has been de-ve-loped at the University of Karlsruhe, will be presented fromthe lower to the planning levels. This comprises vision-based closed-loop control, user interfaces, a re-configurablecomputer-array, execution planning and assembly planningalgorithms tailored to the needs of the microassembly sta-tion.

Keywords: Planning, control, microassembly, microrobo-tics, micromanipulation.

WA2-3 1040

The Relationship between Planning and ProductionActivities in Process Industries

Vladimir Jovan Jozef Stefan Inst.

Summary: In contrast to assembly industries, the CIM con-cept has not yet gained full acceptance in process industries.One of the reasons for this is that in process industries func-tional relations between the different management levels aresometimes difficult to define. This article deals with theproblem of the strict realisation of a short-term plan on theproduction line of a process-oriented factory. This problem,caused by the complexity and uncertainty of process indus-tries, can sometimes be overcomed. The case study in thesecond part of this article briefly describes the solution ofthis problem during the implementation of a brick produc-tion control system.

Keywords: Computer integrated manufacturing, manage-ment system, integration, factory automation, process con-trol system.

WA2-4 1100

Strategies for Integrating Preparation andRealisation — The Case of Product Models

Michael Holm Larsen The Technical Univ. of DenmarkPhillip Kirkby The Technical Univ. of DenmarkJohan Vesterager The Technical Univ. of Denmark

Summary: The purpose of this paper is to discuss the in-tegration of the Product Model (PM) and the Product StateModel (PSM). Focus is on information exchange from thePSM to the PM within the manufacturing of a single ship.The paper distinguishes between information and knowl-edge integration. The paper provides some overall strate-gies for integrating PM and PSM. The context of this dis-cussion is a development project at Odense Steel Shipyard(OSS).

Keywords: Manufacturing, realisation, product model,product state model, integration strategies, conditioning.

WA2-5 1120

Aided Decision and Authentication of LamellatedWood Frameworks

Claude Imberdis Inst. de productique, BesanconDominique Gendreau IUFM de Franche ComteMarc Dahan Inst. de productique, Besancon

Summary: The lameleted gluthed timber frame designneeds many actors: architect, carpenter, different reseach de-partements . . . each of them has their work worry. Becauseof the increasing difficulty of the buildings, it becomes nec-essary to develop somes logiciel tools to choose the assem-bly beetveen the beams, to respect the mechanical resistancesafety, to keep in conformity with the European standard.

Keywords: Lamellated wood frameworks, expert system,aided decision, assembly, structure.

WA3Robotics 1

WA3-1 1000

An Estimate to the Energy Function of a Rigid Robotwith a Stabilizing PD Controller

Amit Ailon Ben-Gurion Univ.Michael I. Gil’ Ben-Gurion Univ.

Summary: This study presents an explicit upper bound tothe energy function of an n-degree of freedom rigid robotwhile it is under the action of a PD controller. The result-ing upper bound is an exponential function that reflects theeffect of the controller gains on the form of the system re-sponse. A tuning-rule for setting the controller gains andadjusting the system rate of convergence towards the de-sired operating point in any given ball, centered at the sys-tem equilibrium point, has been demonstrated. As shown,the effect of the controller structure on the proposed upperbound is similar to the one resulted in the case of a second-

72

order linear system.

Keywords: Rigid robot, PD controller, solution estimates.

WA3-2 1020

Hierarchical Fuzzy Behavior-Based Control of aMulti-Agent Robotic System

Sigal Berman Ben-Gurion Univ.Marco A. A. de Oliveira The Univ. of New MexicoYael Edan Ben-Gurion Univ.Mohammad Jamshidi The Univ. of New Mexico

Summary: A hierarchical fuzzy behavior-based architecturefor the control of a multi-robot system is presented. The ar-bitration of distinct behaviors is achieved by weighing eachbehavior according to its applicability to the current con-trol cycle. This applicability is determined using global con-straints. Combining fuzzy logic and behavior-based controlincreases the systems adaptability and robustness. Simula-tion results of the proposed methodology are discussed anda future hardware implementation is outlined.

Keywords: Robot, multi-agent, behavior-based,control,fuzzy logic.

WA3-3 1040

Geometric and System Decomposition Techniques inApplication to Control of a Mobile Robot with Trailer

Hannah Michalska McGill Univ.

Summary: The trajectory interception approach in its origi-nal form primarily applies to systems whose controllabilityLie algebra is nilpotent and involves only Lie brackets of rel-atively low order. High order Lie brackets in the controlla-bility Lie algebra of the system lead to excessively complexformulations of the open-loop trajectory interception prob-lem which can no longer be solved analytically (in termsof the parameters which represent the values of a feedbackcontrol for an extended system). The purpose of this pa-per is to demonstrate that even in such difficult cases thetrajectory approach can still be made use of. The model ofa mobile robot with trailer used in this paper is not nilpo-tent and requires system motion in the directions of thirdorder Lie brackets. To compensate for the lack of nilpo-tency of the original model, a nilpotent approximation ofthe system is introduced. System decomposition is furtheremployed to obtain an analytically solvable trajectory inter-ception problem formulation. The example of the mobilerobot with trailer has the most complex algebraic structureof all the systems to which the trajectory interception prob-lem was ever applied.

Keywords: Vehicle control, nonholonomic systems, stabi-lization.

WA3-4 1100

Following a Path of Varying Curvature as an OutputRegulation Problem

Claudio Altafini Royal Inst. of Technology

Summary: Given a path of nonconstant curvature, localasymptotic stability can be proven for the general n-trailer

whenever the curvature can be considered as the output ofan exogenous dynamical system. It turns out that the con-trollers that provide convergence to zero of the tracking er-ror chosen for the path following problem are composed ofa prefeedback that input-output linearizes the system plus alinear part that can be chosen in an optimal way.

Keywords: Path following, nonholonomic vehicles, ouptutregulation, input-output linearization.

WA3-5 1120

On Enhancing GJK Algorithm for DistanceComputation Between Convex Polyhedra:Comparison of Improvements

Shen-Po Shiang National Taiwan Univ.Yu-ren Chien National Taiwan Univ.Jing-Sin Liu Academia Sinica

Summary: The computation of Euclidean distance betweentwo convex polyhedra is an important problem in robotics,computer graphics and animation. By geometric reason-ing, we present an improvement of the well-known dis-tance computation algorithm made by Gilbert, Johnson, andKeerthi (GJK). Some comparative simulations are shown toverify the algorithmic improvement in the process of dis-tance computation. In addition, our work provides a sim-ple and efficient algorithm for finding out the informationwhere the closest point of a convex polyhedron to a refer-ence point is on the face, the edge, or on one vertex of thepolyhedron.

Keywords: GJK, distance algorithm, convex polyhedra,comparison of improvements, TCSO.

WA4Robust Control

WA4-1 1000

Links Between Robust and Quadratic Stability ofUncertain Discrete-Time Polynomials

Didier Henrion LAAS-CNRS, ToulouseMichael Sebek Trnka Lab. & UTIAVladimır Kucera Trnka Lab. & UTIA

Summary: An uncertain polynomial is robustly stable, orstable in the sense of Kharitonov, if it is stable for any ad-missible value of the uncertainty, provided the uncertaintyis not varying. The same polynomial is quadratically stable,or stable in the sense of Lyapunov, if it is stable for any ad-missible value of the uncertainty, regardless of whether theuncertainty is varying or not. In this paper, relationshipsbetween robust and quadratic stability of discrete-time un-certain polynomials are studied.

Keywords: Robust stability, quadratic stability, Lyapunovtheorem, Kharitonov theorem, discrete-time uncertain poly-nomials.

73

WA4-2 1020

Development of the Modal Regulator Design Methodfor a Plant with Interval Parameters

Yelena Smagina IsraelIrina Brewer USA

Summary: We consider a problem of a modal P-regulatorsynthesis for a linear multivariable dynamical system withuncertain (interval) parameters: dx/dt = [A]x+[B]u, wherex = x(t) is a state vector, u = u(t) is an input vector, the ma-trices [A] and [B] are matrices with interval elements. Thedesigned feedback regulator u = Kx has to place all coef-ficients of the characteristic polynomial of the closed-loopsystem: dx/dt = ([A] + [B]K)x within assigned intervals.We develop the approach proposed in the previous worksof the authors and prove a direct correlation between sys-tem controllability and existence of a modal P-regulator.

Keywords: Robust control, multivariable system, intervalparameters, P-regulator.

WA4-3 1040

Robust Stability Condition for the System withFeedback Connected Uncertainty and UncertainNumber of Unstable Poles

Kou Yamada Yamagata Univ.

Summary: In this paper, we consider the robust stabi-lization problem for single-input/single-output continuoustime-invariant systems with feedback connected uncertaintysuch that the number of poles of the plant in the right halfplane is not necessarily equal to that of the nominalplant.First of all, we define a class of uncertainty to be considered.The necessary and sufficient robust stability condition forthe system with such class of uncertainty is presentedby us-ing a relation between the plant and the nominal plant.

WA4-4 1100

Real and Complex Stability Radii in AutomaticLoad-Frequency Control Systems via LQG/LTR andLMI

Marco H. Terra Univ. of Sao PauloGregoria M. T. Masca Univ. of Sao Paulo

Summary: In this work, two techniques of robust control(LQG/LTR and LMI), applied to a power electric system, areavailable via stability radii of the system. The structured un-certainties of the nominal model are considered in both de-signs. A set of models is generated considering the combi-nations of the parametric uncertainties. The structured sin-gular values of the both systems are analysed.

Keywords: Robust control, LQG/LTR, LMI, real and com-plex stability radii, µ-analysis and power systems.

WA4-5 1120

Robust Control for a Class of Linear InfiniteDimensional Systems with Multiplicative Disturbances

Alejandro Rodrıguez-Palacios CenidetGuillermo Fernandez-Anaya UIA

Summary: In this paper the problem of robust control fora class of linear infinite dimensional systems under mixeddisturbances of the multiplicative type is addressed. TheLyapunov function approach is used for proving that thereis a controller that stabilizes this class of systems under thepresence of uncertainties and perturbations, and guaranteessome tolerance level for the joint cost functional. A commentis added to the Riccati operator equation’s solution for thisproblem.

Keywords: Infinite-dimensional systems, robust control,mixed disturbances.

WA4-6 1140

About Some Interconnection Between LTR and RPIS

Philippe de Larminat Inst. de Rech. en Cyb. de NantesGuy Lebret Inst. de Rech. en Cyb. de NantesSophie Puren Ing. Pour Signaux et Systemes

Summary: In this paper, the Loop Transfer Recovery designprocedure is extended to non stabilizable systems. After abrief description of the systems considered in this paper, werevisit some results concerning the RPIS (Regulator Problemwith Internal Stability), and give the structure of the con-troller. Thereafter we consider the LTR dual approach andstress the particular configuration of the output sensitivityfunction of the closed-loop system. We show that it is suf-ficient to recover only a part of the sensitivity function toguarantee the stability robustness of the loop. Finally the ad-justment rules which lead to the desired result are described.

Keywords: Linear Control, robust control, LTR, RPIS.

WA5Signal and Image Processing

WA5-1 1000

A Novel Architecture for Digital Pulse Height Analysiswith Application to Radiation Spectroscopy

Itamar Elhanany Nuclear Research Center NegevShimshon Jacobi Nuclear Research Center NegevMichael Kahane Nuclear Research Center NegevEli Marcus Nuclear Research Center NegevDan Tirosh Nuclear Research Center NegevDov Barak Nuclear Research Center Negev

Summary: A novel digital approach to real-time, high-throughput, low-cost pulse height analysis (PHA) for ra-diation spectroscopy is presented. The analog nuclear sig-nal is sampled at a high rate using an analog-to-digital con-verter (ADC), and analyzed by a state-of-the-art field pro-grammable gate array (FPGA). A customized fixed-pointpolynomial fitting algorithm is utilized for pulse-height es-

74

timation. Other pulse parameters, such as width and asym-metry, are attainable for pulse shape analysis (PSA) pur-poses, such as particle identification. The mathematicalcomplexity of the algorithm is strongly reduced by applyingparallel arithmetic, resulting in complete elimination of pro-cessing dead-time. The proposed scheme is easily scalableby substituting the FPGA and ADC with more advanced in-tegrated circuits as they appear.

Keywords: Digital pulse height analysis, digital pulse pro-cessing, FPGA.

WA5-2 1020

Real-Time Adaptive Filtering for Nonstationary ImageRestoration Using Gaussian Input

Abdulriza Abilov Ankara Univ.Onder Tuzunalp Ankara Univ.Ziya Telatar Ankara Univ.

Summary: A new real-time adaptive filter algorithm is pre-sented for the restoration of the images which are degradedby the Atmospheric turbulence or imaging systems. Fil-ter model parameters of the proposed algorithm adaptivelyconverge degradation model parameter in a given time du-ration. Then, a restoration filter is constructed using men-tioned filter parameter. Considerable results have been ob-tained after the real-time restoration.

Keywords: Real-time adaptive filter, gaussian model, imagerestoration.

WA5-3 1040

The Edge Point Detection Problem in ImageSequences: Definition and Comparative Evaluation ofSome 3D Edge Detecting Schemes

L. Jetto Univ. of AnconaG. Orlando Univ. of AnconaA. Sanfilippo Univ. of Ancona

Summary: When dealing with image sequences it is im-portant to take into account the temporal correlation amongconsecutive frames to improve the performance of imageprocessing techniques. In this paper the edge detectionproblem is considered and some 3D edge detectors exploit-ing the information carried by the spatio-temporal correla-tion of the 3D signal are proposed. Their performance isquantitatively and qualitatively evaluated.

Keywords: Edge detection, three-dimensional signals, im-age sequence processing.

WM1 (I)Parameter Estimation

WM1-1 1430

Model-Based Detection Observer of ComponentFailures for Distributed Parameter Systems

Michael A. Demetriou Worcester Polytechnic Instit.

Summary: In this note, fault detection techniques basedon finite dimensional results are extended and applied to a

class of infinite dimensional dynamical systems. This spe-cial class of systems assumes linear plant dynamics havingan abrupt additive perturbation as the fault. This fault is as-sumed to be linear in the (unknown) constant (and possiblyfunctional) parameters. An observer-based model estimateis proposed which serves to monitor the system’s dynamicsand its well posedness is summarized. Using a Lyapunovsynthesis approach applied to infinite dimensional systems,a stable parameter learning scheme is developed. The result-ing parameter adaptation rule is able to “sense” the instanceof the fault occurrence. In addition, it identifies the fault pa-rameters using the additional assumption of persistence ofexcitation. Simulation studies are used to illustrate the ap-plicability of the theoretical results.

Keywords: Detection observers, failure diagnosis, distrib-uted parameter systems.

WM1-2 1450

Parameter Estimation Problem for a NonlinearParabolic Equation with a Singular Nonlocal DiffusionTerm

Azmy S. Ackleh Univ. of Southwestern Louisiana

Summary: We study a quasilinear reaction-diffusion prob-lem that models the dynamics of a population that is eagerto quickly get out of zones with low population densities.A least squares technique for identifying the singular diffu-sion coefficient is developed. Numerical results indicatingthe feasibility of this approach are presented.

WM1-3 1510

Parameter Identification in a NonautononousNonlinear Volterra Integral Equation

Azmy S. Ackleh Univ. Southwestern LouisianaSergiu Aizicovici Ohio Univ.Robert R. Ferdinand East Central Univ.Simeon Reich Technion — IIT

Summary: We propose a least squares technique for iden-tifying parameters in a nonautonomous nonlinear Volterraintegral equation. Numerical results indicating the feasibil-ity of this method are presented.

WM1-4 1530

Adaptive Control of a Time-Varying Parabolic System:Averaging Analysis

Keum-Shik Hong Pusan National Univ.Victor Solo Macquarie Univ.Joseph Bentsman Univ. of Illinois at Urbana

Summary: Related to the error dynamics of an adaptive sys-tem, averaging theorems are developed for coupled differ-ential equations which consist of ordinary differential equa-tions and a parabolic partial differential equation. The re-sults are then applied to the convergence analysis of the pa-rameter estimate errors to zero in the model reference adap-tive control of a nonautonomous parabolic partial differen-tial equation with slowly time varying parameters.

Keywords: Adaptive control, averaging method, conver-gence analysis, parabolic partial differential equation, slow

75

varying system.

WM1-5 1550

Approximation of High-Order Lumped Systems byusing Non-Integer Order Transfer Functions

Luigi Fortuna Univ. di CataniaSalvatore Graziani Univ. di CataniaGiovanni Muscato Univ. di CataniaGiuseppe Nunnari Univ. di CataniaDomenico Porto Univ. di Catania

Summary: Non-integer order systems have been studied byseveral authors to model particular physical systems (elec-trical, biological etc.). In particular it can be shown that anon integer order system is equivalent to an infinite orderLTI system. This feature can be useful considered for modelorder reduction purposes . The main aim of this paper isto show the mathematical background of this new approx-imation theory, the criteria for selecting the order of a non-integer order model which behaves as the original integerorder ones and the quality indexes that can be consideredfor assessing the goodness of the approximated model. Theintroduction of non integer order systems resulted to be anefficient way to compress frequency response informationusually intrinsic in an high number of poles and zeros. Thecomparison with a traditional method of model order reduc-tion proved that a reduced model with the same number ofparameters is not able to get the same good performance inthe frequency domain. To this aim, some examples and sim-ulations are reported.

Keywords: Model order reduction, non-integer order sys-tems, lumped systems, frequency domain methods.

WM2Fuzzy Logic Methods

WM2-1 1430

Closed-Loop Robust Controllers with Fuzzy GainScheduling for FNS Assisted Walking of Paraplegics

Mark Moulin Technion — IITGideon F. Inbar Technion — IIT

Summary: The FNS closed-loop control for assisted walk-ing of paraplegics is studied on the 5-link biped model. Thedesign objectives is that the tracking errors of the joint anglesreference trajectories must be reasonably bounded. The dis-turbances affecting the musculoskeletal model come mainlyfrom the uncertain nonlinear dynamics of the muscle actu-ator. Two robust control schemes are proposed: the SlidingMode control and the LQR control with fuzzy gain schedul-ing. The fuzzy scheduler output provides the relative de-gree (weights) of the system uncertainty according to thejoint angles tracking error. These fuzzy weights schedulesthe appropriate gain from the control gain vs. tracking errorinterpolated function. It turns out that the additional tun-ing of the control moments by the muscle inverse dynam-ics Neural Network is essential for the successful tracking.The extensive simulations show that the performance of theNeural Network static learning depends on the initial po-sition of the musculoskeletal system, and the Neural Net-

work weights initialization. The simulation results demon-strate that the desired uncertainty attenuation properties ofthe proposed control algorithms have been achieved. Theycan be used as a prototype of the real FNS control schemes.

Keywords: Rehabilitation engineering, functional neuro-stimulation, robust control, fuzzy logic, neural networks.

WM2-2 1450

Mathematical Formulation of Fuzzy Cognitive Maps

Chrysostomos D. Stylios Univ. of PatrasPeter P. Groumpos Univ. of Patras

Summary: This paper presents an overview in existing rep-resentations of Fuzzy Cognitive Maps (FCM) and a new ap-proach in the formulation of Fuzzy Cognitive Maps is exam-ined. The description and construction of Fuzzy CognitiveMaps (FCM) is briefly represented and some new ideas forthe modeling of Fuzzy Cognitive Maps are presented. Re-search in this area was mainly focalized on the representa-tion, construction and application of FCM, and now in thispaper different types and mathematical description of FuzzyCognitive Maps are examined and FCMs are mathematicallytransformed in forms that are analogous to Recurrent Neu-ral Networks. This similarity stimulates the investigationof Forward Accessibility for discrete-time FCM models. Fi-nally, an example of a process is presented and it is formu-lated in form that controllability aspects can be examined.

Keywords: Fuzzy Cognitive Maps, controllability.

WM2-3 1510

An Outline for a Universal Logic System: A LogicSystem in Eight Truth Values

Graeme Heald RMIT Univ.

Summary: A Universal Logic System establishes a truthvalue termed neutral between the contrary terms of truthand falsity. The Universal Logic System is composed of threeprimary logic sets: truth, falsity and neutrality together withthree secondary sets: not-true, not-false and not-neutral.Furthermore, there are the Universal and Null logical sets.The Universal set is the union of all primary logical states.Distinction is made between the True set and the Univer-sal set in the Universal logic system, unlike Boolean logicin which they are equated. This has fundamental implica-tions as a many valued logic system. Traffic light states at acontrolled intersection have been used as an illustration ofUniversal Logic.

Keywords: Universal, neutral, not-neutral, null, validity.

WM2-4 1530

Using Soft Computing Methodologies for MultistageSupervisory Control of Complex Systems

Chrysostomos D. Stylios Univ. of PatrasNikolinka Christova Univ. of PatrasPeter P. Groumpos Univ. of Patras

Summary: In this paper, a structure to supervise and con-trol complex systems is presented. The proposed multistagesupervisory control system is structured as a hierarchy ofthree levels - control, supervision and co-ordination levels.

76

While the controllers at control level exercise direct actionover the process, the supervision level generates decisions togovern the operations of the control algorithms at the lowerlevel. The co-ordination level governs the supervisory levelto assure proper overall system behavior. The system pos-sesses different control and supervisory strategies to accom-modate different operating conditions, adaptive behavior toreact under uncertain or unfamiliar situations and the ca-pability to coordinate distributed controllers to accomplishthe system task. The bottom control level is constituted ofconventional controllers or soft control technologies basedon Neural Networks, Fuzzy Logic and Genetic Algorithms.The supervisor is modeled as a Fuzzy Cognitive Map. Basedon the process status, the set of active control and supervi-sory algorithm is chosen.

Keywords: Supervisory control, complex systems, fuzzycognitive maps, fuzzy logic.

WM3Robotics 2

WM3-1 1430

Variable Structure Control with Varying Bounds ofRobot Manipulators

Pierre M. B. Nigrowsky Brunel Univ.Peter J. Turner Brunel Univ.

Summary: A controller for robot manipulators is proposedin this paper. The design is based on a Lyapunov ap-proach with V.S.C. and sliding mode technique. Fundamen-tal properties of the robot, as well as some engineering con-siderations are taken into account during the design proce-dure. Chattering is tackled by indexing the magnitude ofthe switching bound to the tracking error. Simulations re-veal a great reduction of chattering while maintaining goodcontroller performances.

Keywords: Robotics, V.S.C., sliding-mode, chattering, Lya-punov.

WM3-2 1450

An MRAC Output Feedback Controller for RobotManipulators

Howard M. Schwartz Carleton Univ.

Summary: An adaptive controller is proposed, for the track-ing control of robotic manipulators that does not require themeasurement of joint velocities. The controller belongs tothe class of model-reference adaptive controllers. An ob-server is used to generate an estimate of the joint velocitiesand an observer-based identifier with projection is used toupdate the parameter vector estimate. Simulation results aregiven to show the effectiveness of the control algorithm.

Keywords: Robot control, feedback linearization, outputfeedback, adaptive systems.

WM3-3 1510

Basic Fairing Principles of Fiberglass Pits andPatches

Glen C. Oliver Univ. of Texas at ArlingtonPanayiotis S. Shiakolas Univ. of Texas at ArlingtonTommy J. Lawley Univ. of Texas at Arlington

Summary: There are many reasons compelling us to au-tomating surface finishing of fiberglass composites. Harm-ful dusts, repetitive motion injury, and product quality arejust a few reasons for automation. We have developed amethodology for robotic surface finishing of fiberglass com-posite pits and patches. Methods for the filling of pits andpatches with a fill material, the subsequent forming of theuncured fill material, and the fairing of the various work-piece features are examined.An anthropomorphic manipulator is used with “around thearm” force control along with custom developed softwarecalled RobSurf. RobSurf provides for the reverse engineer-ing of the samples used, and creation of robot programsbased upon the reverse engineered surface. The necessaryfilling, forming, and fairing process parameters are exploredand the subsequent experimentally determined parametersare described. Factory implementation suggestions are pro-vided that utilize commercially available components forworkcell development.

Keywords: Robotic, surface, fairing, fiberglass, composite.

WM3-4 1530

RobSurf: A Near Real Time OLP System for RoboticSurface Finishing

Panayiotis S. Shiakolas Univ. of Texas at ArlingtonDragan Labalo Univ. of Texas at ArlingtonJ. Mick Fitzgerald Univ. of Texas at Arlington

Summary: In this paper we will discuss the develop-ment and use of a CAD based robot path and process-planning environment for surface finishing of conventional(metal) and non-conventional material (fiberglass compos-ites) workpieces. This system is based on a modular andparametric approach process modeling for experimentationin determining “optimum” process parameters. It can op-erate upon workpieces of various characteristics due to theintegrated soft setup methodologies.RobSurf is a surface modeling and path generation systemdeveloped specifically for experimentation and identifica-tion of process parameters for processes associated with sur-face finishing. Using a coordinate measuring device inter-faced with AutoCAD, the system is capable of generatinga CAD model of any surface through reverse engineeringtechniques and generating native robot control code withembedded process parameters for various routines such asfill, form, and fair as well as information of the process tool-ing employed. The generated robot control code is thentransferred to the robot controller via an RS-232C interfaceconnection.

Keywords: Robotic, OLP, soft-setup, reverse-engineering.

77

WM3-5 1550

Adaptive Nonlinear Visual Servoing UsingLyapunov-Based Design

Fabio Conticelli Scuola Superiore Sant’AnnaBenedetto Allotta Scuola Superiore Sant’Anna

Summary: In this paper, an adaptive nonlinear controlscheme is designed to solve the problem of controlling therelative pose between a robot camera and a rigid object. Theimage-based visual system of the camera-object interactionis expressed in terms of global coordinates fully defined inthe image plane, then a discrete time interaction model is de-rived, since the visual sampling time is not negligible at theactual state of technology. By exploiting nonlinear controlla-bility properties, a nonlinear control law is designed basedon Lyapunov’s direct method. Moreover, we propose a 3-D estimation procedure based on prediction errors to copewith the unknown depth of the object. Experimental resultswith a 6-DOF robot manipulator in eye-in-hand configura-tion validate the theoretical framework.

Keywords: Nonlinear visual model, controllability, discretetime visual servoing, asymptotic stability, experimental val-idation.

WM3-6 1610

On the Inclusion of Robot Dynamics in VisualServoing Systems

Fabio Conticelli Scuola Superiore Sant’AnnaBenedetto Allotta Scuola Superiore Sant’Anna

Summary: In this paper the problem of including the robotdynamics in the control loop of visual servoing systems isconsidered. After introducing the image-based visual in-teraction model between a robot camera and a rigid objectparametrized by a finite number of image features, the prob-lem of local feedback stabilization is considered, in both thecases of interconnection to a linear subsystem and to a non-linear open-chain manipulator. The proposed control sys-tem design uses backstepping approach to ensure local sta-blity of the equilibrium of the whole system, assuming thatthe linear subsystem is minimum phase of relative degreeone. In the case of inclusion of the nonlinear lagrangian dy-namics of the robot, the obtained control law is similar tothe well-known computed torque law. A case study is alsoreported to validate the developed control design, approxi-mating the robot and its local controller by a diagonal linearsubsystem.

Keywords: Image-based visual interaction, robot dynamics,backstepping approach, asymptotic stability.

WM4 (I)Sliding Mode Control

WM4-1 1430

Partial Lipschitz Nonlinear Sliding Mode Observers

Ali J. Koshkouei Univ. of SheffieldAlan S. I. Zinober Univ. of Sheffield

Summary: The stability of a nonlinear observer for sys-tems with uncertainties usually requires some sufficient con-ditions. The Lipschitz condition is a restrictive conditionwhich many classes of systems may not satisfy. In this paperwe consider a class of systems with two uncertain parts; onewhich satisfies the Lipschitz condition, whilst the other doesnot satisfy the Lipschitz condition but is a bounded uncer-tainty. Sliding mode theory is applied to yield feedforwardcompensation control to stabilize the error estimation sys-tem with non-Lipschitz uncertainty. New sufficient condi-tions for stability of the Thau observer are proposed. Theseconditions ensure the stability of the nonlinear observer byselecting a suitable observer gain matrix.

Keywords: Sliding observers, nonlinear observers, slidingmode.

WM4-2 1450

Dynamical Adaptive First and Second Order SlidingMode Control of Nonlinear Non-Triangular UncertainSystems

Alan S. I. Zinober Univ. of SheffieldJulie C. Scarratt Univ. of SheffieldAntonella Ferrara Univ. of PaviaLuisa Giacomini Aston Univ.Miguel Rios-Bolıvar Univ. de Los Andes

Summary: In this paper combined algorithms for the con-trol of non-triangular nonlinear systems with unmatcheduncertainties will be presented. The controllers consist ofa combination of Dynamical Adaptive Backstepping (DAB)and Sliding Mode Control (SMC) of first and second order.In order to solve a tracking problem, the DAB algorithm (ageneralization of the backstepping technique) makes use ofvirtual functions as well as tuning functions to construct atransformed system for which a regulation problem has tobe solved. The new state is extended by an (n − ρ)-th or-der subsystem in canonical form where n is the order of theoriginal system and ρ is the relative degree. The role of thesliding mode control is to replace the last step of the designof the control law to obtain more robustness towards distur-bances and unmodelled dynamics. The main advantages ofthe second order sliding mode algorithm are the preventionof chattering, higher accuracy and a significant simplifica-tion of the control law. A comparative study of these firstand second order sliding controllers will be presented.

Keywords: Backstepping, sliding mode control, compara-tive study.

78

WM4-3 1510

Adaptive Sliding Backstepping Control of NonlinearSemi-Strict Feedback Form Systems

Ali J. Koshkouei Univ. of SheffieldAlan S. I. Zinober Univ. of Sheffield

Summary: This paper considers the application of a com-bined adaptive backstepping sliding mode control (SMC)algorithm to a class of nonlinear continuous uncertain pro-cesses which can be converted to a semi-parametric strictform. The algorithm follows a systematic procedure for thedesign of dynamical adaptive SMC laws for the output reg-ulation of observable minimum phase nonlinear systems.

Keywords: Backstepping, sliding mode control, adaptivecontrol, output regulation, sliding surface.

WM4-4 1530

A Feedforward-Feedback Interpretation of a SlidingMode Control Law

Govert Monsees Delft Univ. of TechnologyKoshy George Delft Univ. of TechnologyJacquelien M.A. Scherpen Delft Univ. of TechnologyMichel Verhaegen Delft Univ. of Technology

Summary: In this paper we provide a feedforward-feed-back interpretation of a sliding mode control scheme. Givena desired trajectory, the feedforward signal is generated us-ing a stable inversion method, and the feedback signal in-cludes the switching term of the sliding mode control law.In this manner, we introduce robustness into the stable in-version technique. This approach is motivated by the needto replicate time signals typically in the automobile indus-try. The application of such an interpretation to a quartercar benchmark model yields encouraging results. Special at-tention will be given to non-minimum phase systems illus-trated by a simulation example of the lunar roving vehicle.

Keywords: Sliding mode, non-linear systems, Model Inver-sion.

WM4-5 1550

Nonminimum Phase Output Tracking via SlidingMode Control: Stable System Center Technique

Ilya A. Shkolnikov The Univ. of Alabama in HuntsvilleYuri B. Shtessel The Univ. of Alabama in Huntsville

Summary: Nonlinear output tracking in multi-input/multi-output (MIMO) nonminimum phase systems with matchednonlinearities as well as matched and unmatched distur-bances is considered in sliding modes. The output trackingproblem has been transformed to an equivalent state controlproblem. The nonminimum phase output tracking problemis solved using an extension of the method of system cen-ter for nonminimum phase systems and the dynamic slid-ing manifold technique. The asymptotic motion of the out-put tracking error with given eigenvalue placement for non-causal output tracking is provided in absence of unmatcheddisturbance. Linear bounded error dynamics with desiredeigenvalue placement forced by unmatched disturbance andan arbitrary reference output profile are provided for causal

output tracking in sliding mode. The theoretical results areillustrated on two numerical examples.

Keywords: Nonminimumphase systems, MIMO nonlineartracking, sliding mode control.

WM4-6 1610

2-Sliding Mode with Adaptation

Giorgio Bartolini Univ. of CagliaryArie Levant Inst. for Industrial MathematicsAlessandro Pisano Univ. of CagliaryElio Usai Univ. of Cagliary

Summary: Sliding mode is used in order to retain a dy-namic system accurately at a given constraint and is themain operation mode in variable structure systems. Suchmode is a motion on a discontinuity set of a dynamic systemand features theoretically-infinite-frequency switching. Thestandard sliding modes are known to feature finite-time con-vergence, precise keeping of the constraint and robustnesswith respect to internal and external disturbances. In real-ization their sliding precision is proportional to the time in-terval between measurements. Having generalized the no-tion of sliding mode, higher order sliding modes preserveor generalize its main properties and remove the chatteringeffect. With discrete measurements they may provide forup to the rth order of sliding precision with respect to themeasurement interval. The main implementation problemof these modes is the information demand growing with thesliding order. If the aim is to nullify some output variable σthen r-sliding mode realization generally requires measure-ments of the time derivatives of up to the (r − 2)th order ofσ to be available. A new approach demonstrated in the pa-per provides for 3-sliding accuracy realization while only σitself is available. That is the first controller of such kind.

Keywords: Nonlinear control, sliding mode, adaptation.

WM5Computer Networks and Queing Systems

WM5-1 1430

The Optimal Markov Strategy for Access in ISDNswith Reserves of Channels

Agassy Z. Melikov International American Univ.Dervis Z. Deniz Eastern Mediterranean Univ.

Summary: In this paper, the use of the Markov DecisionProcess (MDP) to find the optimal strategy for access in IS-DNs with reserves of channels is proposed. ISDN providesmultiple channels for telecommunications access. The prob-lem is formulated as a multi-resource queuing (MRQ) sys-tem where different types of customers require a randomnumber of channels simultaneously. The algorithm that re-alizes the optimal strategy for access when heterogeneouscustomers demand service in ISDNs is developed.

Keywords: Optimal Markov strategy, Markov DecisionProcess (MDP), access in ISDNs, reserves of channels.

79

WM5-2 1450

Tbit/sec Switching Scheme for ATM/WDMHigh-Speed Computer Networks

Itamar Elhanany Ben-Gurion Univ.Dan Sadot Ben-Gurion Univ.

Summary: A novel high throughput, reservation-basedswitch architecture for ATM/WDM networks is presented.The scheme is contention-free and highly flexible yield-ing a powerful solution for high-speed broadband packet-switched networks. Switching management and control isstudied for data rates of up to 10 Gbit/sec/port, providingaggregated throughput of over 1 Terabit/sec.

Keywords: Switching architectures, ATM/WDM, packet-switching computer networks, Tbps.

WM5-3 1510

Grid-based ATM Switch Architecture: A NewFault-Tolerant Space-Division Switch FabricArchitecture

Haralampos S. Laskaridis Aristotle Univ. of ThessalonikiAndreas A. Veglis Aristotle Univ. of ThessalonikiGeorgios I. Papadimitriou Aristotle Univ. of ThessalonikiAndreas S. Pomportsis Aristotle Univ. of Thessaloniki

Summary: Asynchronous Transfer Mode (ATM) has beenchosen to be employed in the implementation of B-ISDN be-cause of its superiority in fast packet switching. The deploy-ment of ATM in Wide Area Networks has revealed the ne-cessity of ATM switches with large number of input and out-put ports. Unfortunately, it has become obvious that ATMswitch fabrics form a bottleneck in Wide Area ATM Net-works. In this paper we present a new grid-based ATMswitch, which is fault-tolerant, self-routing and easily ex-pandable. The switch’s architecture is described in detail,along with the internal routing algorithm, and its simplic-ity in comparison to the Banyan networks is demonstrated.The analytical model of the performance as well as simula-tion results are presented. The characteristics of the switchregarding fault tolerance are also briefly discussed.

Keywords: ATM, switch fabric, grid-based, fault tolerance.

80

Index of Authors, Chairpersons, and Organizers

Key: ‘C’ = session Cair or Co-chair, ‘O’ = session Organizer; page numbers refer to abstracts

AAbilov, Abdulriza (WA5-2) . . . . . . . . . . . . . . . . . . . . . . . . 75Achuthan, G. (MP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Ackleh, Azmy S. (TP1-C)Ackleh, Azmy S. (WA1-C)Ackleh, Azmy S. (WM1-2) . . . . . . . . . . . . . . . . . . . . . . . . . 75Ackleh, Azmy S. (WM1-3) . . . . . . . . . . . . . . . . . . . . . . . . . 75Acosta, L. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Adetayo, Anthony A. (TA2-6) . . . . . . . . . . . . . . . . . . . . . . 55Aggoune, Woihida (TA5-5) . . . . . . . . . . . . . . . . . . . . . . . . 59Ailon, Amit (WA3-C)Ailon, Amit (WA3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Aizicovici, Sergiu (WM1-3) . . . . . . . . . . . . . . . . . . . . . . . . 75Alastruey, Carlos F. (TP5-1) . . . . . . . . . . . . . . . . . . . . . . . . 69Alekseyenko, Y. (MP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 52Allgower, Frank (MA1-6) . . . . . . . . . . . . . . . . . . . . . . . . . . 38Allgower, Frank (TA2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Allotta, Benedetto (WM3-5) . . . . . . . . . . . . . . . . . . . . . . . . 78Allotta, Benedetto (WM3-6) . . . . . . . . . . . . . . . . . . . . . . . . 78Altafini, Claudio (WA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . 73Amato, Francesco (TP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . 68Anatone, Michele (MA3-6) . . . . . . . . . . . . . . . . . . . . . . . . . 40Andersen, T. R. (TM2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Andrievsky, Boris R. (TA4-2) . . . . . . . . . . . . . . . . . . . . . . . 57Anghelea, Marius (MM2-C)Anghelea, Marius (MM2-3) . . . . . . . . . . . . . . . . . . . . . . . . 44Anthonis, Jan (MA1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Anthonis, Jan (MM2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Antsaklis, Panos J. (MA2-C)Antsaklis, Panos J. (MA2-4) . . . . . . . . . . . . . . . . . . . . . . . . 38Aracil, Javier (TM4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Arsie, I. (MA3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Arvanitis, K. G. (MP5-C)Arvanitis, K. G. (MM1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 42Arvanitis, K. G. (MP5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Arvanitis, K. G. (TM2-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

BBalakrishnan, S. N. (TP5-C)Balakrishnan, S. N. (TP5-2) . . . . . . . . . . . . . . . . . . . . . . . . 69Balduzzi, Fabio (MA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Balogh, Andras (TM1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Bansal, Vikrant (TM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Bar-Gil, Aharon (MA1-C)Bar-Itzhack, Itzhack Y. (TP3-3) . . . . . . . . . . . . . . . . . . . . . 68Bar-Shalom, Yaakov (MP3-O)Bar-Shalom, Yaakov (MP3-C)Bar-Shalom, Yaakov (MP3-2) . . . . . . . . . . . . . . . . . . . . . . 50Bar-Shalom, Yaakov (MP3-3) . . . . . . . . . . . . . . . . . . . . . . 51Bar-Shalom, Yaakov (MP3-4) . . . . . . . . . . . . . . . . . . . . . . 51Barak, Dov (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Baramov, Lubomır (MA1-3) . . . . . . . . . . . . . . . . . . . . . . . . 37Bartolini, Giorgio (WM4-6) . . . . . . . . . . . . . . . . . . . . . . . . 79Bashirov, Agamirza (MA4-6) . . . . . . . . . . . . . . . . . . . . . . . 41Bates, Adam (MP2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Beghelli, Sergio (MP2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Ben-Asher, Josef Z. (TM3-O)Ben-Asher, Josef Z. (TM3-2) . . . . . . . . . . . . . . . . . . . . . . . . 62

Ben-Asher, Josef Z. (TP3-C)Bentsman, Joseph (WM1-4) . . . . . . . . . . . . . . . . . . . . . . . . 75Berber, Ridvan (TA2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Bergeon, B. (TM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Berman, Nadav (WM3-C)Berman, Sigal (WA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Bhattacharyya, S. P. (TA3-1) . . . . . . . . . . . . . . . . . . . . . . . . 56Błachuta, Marian J. (MM3-4) . . . . . . . . . . . . . . . . . . . . . . . 46Blumel, Anna L. (MM3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 45Bobal, Vladimir (MM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 47Bobal, Vladimir (TA1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Bolzern, Paolo (TA5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Boussalis, Helen (MM3-6) . . . . . . . . . . . . . . . . . . . . . . . . . . 46Brewer, Irina (WA4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Brindley, John (TA3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Brosilow, Coleman (TA2-4) . . . . . . . . . . . . . . . . . . . . . . . . 55Bucolo, Maide (TP2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Budman, Hector (TA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

CCampos, Javier (MA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Canuto, Enrico (MA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Cao, Chongsheng (TM1-3) . . . . . . . . . . . . . . . . . . . . . . . . . 59Cao, Liyu (TM5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Carapellucci, Roberto (MA3-6) . . . . . . . . . . . . . . . . . . . . . 40Carravetta, Francesco (TA5-3) . . . . . . . . . . . . . . . . . . . . . . 58Chan, Cheney (MM2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Chen, Robert H. (MP2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 50Chien, Yu-ren (WA3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Christova, Nikolinka (WM2-4) . . . . . . . . . . . . . . . . . . . . . 76Cipollone, Roberto (MA3-6) . . . . . . . . . . . . . . . . . . . . . . . . 40Colaneri, Patrizio (TA5-C)Colaneri, Patrizio (TA5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Colantonio, M. C. (MA2-C)Colantonio, M. C. (MA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . 38Constantin, Nicolae (MA5-6) . . . . . . . . . . . . . . . . . . . . . . . 42Conticelli, Fabio (WM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . 78Conticelli, Fabio (WM3-6) . . . . . . . . . . . . . . . . . . . . . . . . . 78Cruz, Sandra L. (TP2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Cruz, Sandra L. (TP2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Curtain, Ruth F. (TP1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Czolczynski, Krzysztof (TA3-5) . . . . . . . . . . . . . . . . . . . . 56

DDadone, Andrea (MA3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 39Dahan, Marc (WA2-C)Dahan, Marc (WA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Dambrosio, Lorenzo (MA3-C)Dambrosio, Lorenzo (MA3-3) . . . . . . . . . . . . . . . . . . . . . . 39Darouach, Mohamed (TA5-5) . . . . . . . . . . . . . . . . . . . . . . 59Daum, Frederick E. (MP3-1) . . . . . . . . . . . . . . . . . . . . . . . 50Davidovitz, Avraham (TM3-C)Davis, Leo (MA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Dayan, Joshua (MM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Dayan, Yehoshua (MM3-C)De Keyser, Robin (MM2-3) . . . . . . . . . . . . . . . . . . . . . . . . . 44De la Sen, Manuel (TP5-1) . . . . . . . . . . . . . . . . . . . . . . . . . 69de Larminat, Philippe (WA4-6) . . . . . . . . . . . . . . . . . . . . 74

81

De Moor, Bart (TP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70de Oliveira, Marco A. A. (WA3-2) . . . . . . . . . . . . . . . . . . 73Declercq, Filip (MM2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Decoster, Martin (MM2-3) . . . . . . . . . . . . . . . . . . . . . . . . . 44Demetriou, Michael A. (TM1-O)Demetriou, Michael A. (TM1-C)Demetriou, Michael A. (TP1-O)Demetriou, Michael A. (WA1-O)Demetriou, Michael A. (WM1-O)Demetriou, Michael A. (WM1-C)Demetriou, Michael A. (WM1-1) . . . . . . . . . . . . . . . . . . . 75Deniz, Dervis Z. (WM5-1) . . . . . . . . . . . . . . . . . . . . . . . . . 79Desages, Alfredo (MA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . 37Dolgui, Alexandre (WA2-1) . . . . . . . . . . . . . . . . . . . . . . . . 71Dostal, Petr (MM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Dostal, Petr (TA1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Dragan, Vasile (TP4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Dumitrache, Ion (MA5-6) . . . . . . . . . . . . . . . . . . . . . . . . . . 42Dumont, Guy A. (TP2-C)Dumont, Guy A. (TP2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 66Dzielinski, Andrzej (TA3-3) . . . . . . . . . . . . . . . . . . . . . . . . 56

EEdan, Yael (WA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Egerstedt, Magnus (TP3-1) . . . . . . . . . . . . . . . . . . . . . . . . 67Elhanany, Itamar (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . 74Elhanany, Itamar (WM5-2) . . . . . . . . . . . . . . . . . . . . . . . . . 80Emirsajlow, Zbigniew (TP1-4) . . . . . . . . . . . . . . . . . . . . . 66Enikeev, Adel K. (MM1-6) . . . . . . . . . . . . . . . . . . . . . . . . . 43Ennis, Brian J. (TA2-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Evans, Mark (MA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

FFantuzzi, Cesare (MP2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 50Farkhi, Elza (TP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Fatikow, Sergej (WA2-C)Fatikow, Sergej (WA2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Favoreel, Wouter (TP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 70Ferdinand, Robert R. (WM1-3) . . . . . . . . . . . . . . . . . . . . . 75Fernandez-Anaya, Guillermo (WA4-5) . . . . . . . . . . . . . 74Ferrara, Antonella (WM4-2) . . . . . . . . . . . . . . . . . . . . . . . 78Ferreira, Pedro M. G. (MM4-C)Ferreira, Pedro M. G. (MM4-2) . . . . . . . . . . . . . . . . . . . . . 46Feuer, Arie (P-II-C)Feuer, Arie (MP5-C)Fitzgerald, J. Mick (WM3-4) . . . . . . . . . . . . . . . . . . . . . . . 77Florentino, Helenice O. (MA1-4) . . . . . . . . . . . . . . . . . . . 37Fontes, Adhemar de B. (MM2-1) . . . . . . . . . . . . . . . . . . . 44Fortuna, Luigi (TP2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Fortuna, Luigi (WM1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Fradkov, Alexander L. (TA4-C)Fradkov, Alexander L. (TA4-2) . . . . . . . . . . . . . . . . . . . . . 57Fradkov, Alexander L. (TA4-5) . . . . . . . . . . . . . . . . . . . . . 57Frid, Arkadi I. (MM1-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Fridman, Emilia (MM1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 43Fridman, Emilia (TM1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 59Fridman, Emilia (TP1-C)Friedland, Bernard (TA4-4) . . . . . . . . . . . . . . . . . . . . . . . . 57Friedland, Bernard (TM4-C)

GGani, R. (TM2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Garcia-Sanz, Mario (TP5-1) . . . . . . . . . . . . . . . . . . . . . . . . 69Gasparetto, Alessandro (MP1-2) . . . . . . . . . . . . . . . . . . . 49

Gendreau, Dominique (WA2-5) . . . . . . . . . . . . . . . . . . . . 72George, Koshy (MA4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40George, Koshy (MM5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 48George, Koshy (WM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 79Gera, Amos E. (MM4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Germani, Alfredo (TA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Gershon, E. (TA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Gessing, Ryszard (MA4-2) . . . . . . . . . . . . . . . . . . . . . . . . . 40Gevers, Michel (TP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Ghulchak, Andrey (TA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . 56Giacomini, Luisa (WM4-2) . . . . . . . . . . . . . . . . . . . . . . . . . 78Gil’, Michael I. (TP1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Gil’, Michael I. (WA3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Gitizadeh, R. (TM3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Glielmo, Luigi (MA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Goodwin, Graham C. (TA2-1) . . . . . . . . . . . . . . . . . . . . . . 54Graziani, Salvatore (TP2-5) . . . . . . . . . . . . . . . . . . . . . . . . 67Graziani, Salvatore (WM1-5) . . . . . . . . . . . . . . . . . . . . . . . 76Green, Itzhak (MM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Groumpos, Peter P. (WM2-2) . . . . . . . . . . . . . . . . . . . . . . . 76Groumpos, Peter P. (WM2-4) . . . . . . . . . . . . . . . . . . . . . . . 76Guelman, Moshe (TM3-4) . . . . . . . . . . . . . . . . . . . . . . . . . 62Guez, Allon (MM1-C)Guez, Allon (MM1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Gurfil, Pini (TM3-O)Gurfil, Pini (TM3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Guschinsky, Nikolai N. (WA2-1) . . . . . . . . . . . . . . . . . . . 71Gustafsson, Fredrik (TM5-6) . . . . . . . . . . . . . . . . . . . . . . . 65Gutman, Per-Olof (P-V-C)Gutman, Per-Olof (MM5-C)Gutman, Per-Olof (MM5-1) . . . . . . . . . . . . . . . . . . . . . . . . 47Gutman, Per-Olof (MM5-5) . . . . . . . . . . . . . . . . . . . . . . . . 48Gutman, Per-Olof (TP2-2) . . . . . . . . . . . . . . . . . . . . . . . . . 66

HHagenblad, Anna (MM5-2) . . . . . . . . . . . . . . . . . . . . . . . . 48Hajiyev, Chingiz (TM5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 65Halevi, Yoram (MP1-C)Halevi, Yoram (MP1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Halevi, Yoram (MP1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Halici, Ugur (MM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Hamilton, A. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Han, Jin-wook (MP5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Han, Woo-yong (MP5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Hansen, Lars Kai (MM5-4) . . . . . . . . . . . . . . . . . . . . . . . . . 48Heald, Graeme (WM2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Henrion, Didier (WA4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 73Hjalmarsson, Hakan (TM5-6) . . . . . . . . . . . . . . . . . . . . . . 65Hoffmann, Frank (TP3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 67Hong, Keum-Shik (MA3-5) . . . . . . . . . . . . . . . . . . . . . . . . 40Hong, Keum-Shik (MM3-C)Hong, Keum-Shik (MM3-1) . . . . . . . . . . . . . . . . . . . . . . . . 45Hong, Keum-Shik (WM1-4) . . . . . . . . . . . . . . . . . . . . . . . . 75Hu, Xiaoming (TM4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Hu, Zhenning (TP5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

IIdan, Moshe (TP3-C)Iervolino, Raffaele (TP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . 68Ikenaga, Scott (MA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Imberdis, Claude (WA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . 72Inbar, Gideon F. (WM2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 76Ioffe, Alexander (WA1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 71Ionita, Achim (TP4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

82

Ishihara, A. (MP5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Ismail, Ahmed A. (TP2-1) . . . . . . . . . . . . . . . . . . . . . . . . . 66

JJacobi, Shimshon (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . 74Jamshidi, Mohammad (WA3-2) . . . . . . . . . . . . . . . . . . . . 73Jamshidi, Mohammad (WM2-C)Jancsok, Pal (MM2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Jeon, Dong-Sub (MM3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 45Jetto, L. (WA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Jodorkovsky, Mario (TM3-4) . . . . . . . . . . . . . . . . . . . . . . . 62Johansson, Mikael (TA3-C)Johansson, Mikael (TA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . 56Jørgensen, S. Bay (TM2-4) . . . . . . . . . . . . . . . . . . . . . . . . . 60Jovan, Vladimir (WA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 72

KKaczorek, Tadeusz (MA4-C)Kaczorek, Tadeusz (MA4-5) . . . . . . . . . . . . . . . . . . . . . . . . 41Kaczorek, Tadeusz (MM4-6) . . . . . . . . . . . . . . . . . . . . . . . 47Kahane, Allan C. (TA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 53Kahane, Michael (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 74Kalogeropoulos, G. (MM1-1) . . . . . . . . . . . . . . . . . . . . . . . 42Kalogeropoulos, G. (MP5-1) . . . . . . . . . . . . . . . . . . . . . . . 52Kalogeropoulos, G. (TM2-6) . . . . . . . . . . . . . . . . . . . . . . . 61Kapitaniak, Tomasz (TA3-5) . . . . . . . . . . . . . . . . . . . . . . . 56Kaufman, Howard (MP5-3) . . . . . . . . . . . . . . . . . . . . . . . . 52Kaufman, Howard (MP5-4) . . . . . . . . . . . . . . . . . . . . . . . . 52Keel, L. H. (TA3-C)Keel, L. H. (TA3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Kevrekidis, Yannis (TM1-3) . . . . . . . . . . . . . . . . . . . . . . . . 59Kirkby, Phillip (WA2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Kirubarajan, Thiagalingam (MP3-3) . . . . . . . . . . . . . . . 51Kirubarajan, Thiagalingam (MP3-4) . . . . . . . . . . . . . . . 51Knapp, Timothy (TA2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Koo, T. K. John (TP3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Kookos, I. K. (MM1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Kookos, I. K. (TM2-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Koshkouei, Ali J. (WM4-1) . . . . . . . . . . . . . . . . . . . . . . . . . 78Koshkouei, Ali J. (WM4-3) . . . . . . . . . . . . . . . . . . . . . . . . . 79Kosut, Robert (MA5-C)Kosut, Robert (MA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Kotta, Ulle (TM4-C)Kotta, Ulle (TA4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Koutsoukos, Xenofon D. (MA2-4) . . . . . . . . . . . . . . . . . . 38Kraffer, Ferdinand (MM4-3) . . . . . . . . . . . . . . . . . . . . . . . . 47Krasnosel’skii, Alexander M. (TA4-C)Krasnosel’skii, Alexander M. (TA4-1) . . . . . . . . . . . . . . 57Krolikowski, Andrzej (MA5-4) . . . . . . . . . . . . . . . . . . . . . 42Krolikowski, Andrzej (TP5-3) . . . . . . . . . . . . . . . . . . . . . 69Krstic, Miroslav (TM1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Kucera, Vladimır (MM4-C)Kucera, Vladimır (WA4-1) . . . . . . . . . . . . . . . . . . . . . . . . . 73Kurina, Galina A. (WA1-5) . . . . . . . . . . . . . . . . . . . . . . . . . 71

LLabalo, Dragan (WM3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 77Lampe, Bernhard P. (TA1-1) . . . . . . . . . . . . . . . . . . . . . . . . 53Langholz, Gideon (TM4-1) . . . . . . . . . . . . . . . . . . . . . . . . . 63Larsen, Jan (MM5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Larsen, Michael Holm (WA2-4) . . . . . . . . . . . . . . . . . . . . 72Laskaridis, Haralampos S. (WM5-3) . . . . . . . . . . . . . . . 80Lawley, Tommy J. (WM3-3) . . . . . . . . . . . . . . . . . . . . . . . . 77Leblebicioglu, Kemal (MM2-2) . . . . . . . . . . . . . . . . . . . . . 44

Lebret, Guy (WA4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Lee, Chang-goo (MP5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Lefebvre, Dimitri (MA2-2) . . . . . . . . . . . . . . . . . . . . . . . . . 38Leizarowitz, Arie (WA1-4) . . . . . . . . . . . . . . . . . . . . . . . . . 71Levant, Arie (WM4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79Levin, Genrikh M. (WA2-1) . . . . . . . . . . . . . . . . . . . . . . . . 71Lewin, Daniel R. (TA2-O)Lewin, Daniel R. (TA2-C)Lewin, Daniel R. (TM2-O)Lewin, Daniel R. (TM2-C)Lewin, Daniel R. (TM2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 60Lewin, Daniel R. (TM2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 61Lewis, Frank L. (MA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Li, Yicong (MP3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51Liberzon, Daniel (MA2-6) . . . . . . . . . . . . . . . . . . . . . . . . . . 39Liberzon, Mark R. (TA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . 56Linker, Raphael (MM5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . 47Litsyn, Elena (TA4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Liu, Jing-Sin (WA3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Lu, E. (TM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Lychenko, Nataly M. (MM1-2) . . . . . . . . . . . . . . . . . . . . . 43

MMacdonald, John M. (MP2-3) . . . . . . . . . . . . . . . . . . . . . . 50Madan, Rabi (MP3-C)Mahmudov, Nazim (MA4-6) . . . . . . . . . . . . . . . . . . . . . . . 41Mahout, Vincent (TM5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 64Malabre, Michel (TP4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Malanowski, Kazimierz (WA1-C)Malanowski, Kazimierz (WA1-2) . . . . . . . . . . . . . . . . . . 70Manes, Costanzo (TA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Marchetti, G. (TA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Marcus, Eli (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Margaliot, Michael (TM4-1) . . . . . . . . . . . . . . . . . . . . . . . . 63Marichal, G. N. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Marklin, Richard (TP5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 70Maroni, Massimo (TA5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Maroni, Massimo (TA5-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Masca, Gregoria M. T. (WA4-4) . . . . . . . . . . . . . . . . . . . . 74Melikov, Agassy Z. (WM5-1) . . . . . . . . . . . . . . . . . . . . . . . 79Mendez, J. A. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Menini, Laura (MP1-C)Menini, Laura (MP1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Menold, Patrick H. (TA2-2) . . . . . . . . . . . . . . . . . . . . . . . . 54Menon, P. K. (TM3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Miani, Stefano (MP1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Michalska, Hannah (WA3-C)Michalska, Hannah (WA3-3) . . . . . . . . . . . . . . . . . . . . . . . 73Mirkin, Boris M. (MA5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 42Mirkin, Leonid (P-IV-C)Mirkin, Leonid (TA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Mirkin, Leonid (TP4-C)Mirkin, Leonid (TP4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Mirmirani, Majdedin (MM3-6) . . . . . . . . . . . . . . . . . . . . . 46Monsees, Govert (WM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . 79Morales, Mauricio (MM3-6) . . . . . . . . . . . . . . . . . . . . . . . . 46Moreno, L. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Moshou, Dimitrios (MM2-5) . . . . . . . . . . . . . . . . . . . . . . . 45Moulin, Mark (WM2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Muscato, Giovanni (WM1-5) . . . . . . . . . . . . . . . . . . . . . . . 76

NNadler, Assaf (TP3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Nagurka, Mark (TP5-C)

83

Nagurka, Mark (TP5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Nekimken, Howard (MP2-3) . . . . . . . . . . . . . . . . . . . . . . 50Nepomnyashchikh, Yurii V. (TA4-6) . . . . . . . . . . . . . . . . 58Nigrowsky, Pierre M. B. (WM3-1) . . . . . . . . . . . . . . . . . . 77Ninness, Brett (TM5-C)Ninness, Brett (TM5-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Niu, Ruixin (MP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Novikov, Boris A. (MM1-6) . . . . . . . . . . . . . . . . . . . . . . . . 43Nunnari, Giuseppe (WM1-5) . . . . . . . . . . . . . . . . . . . . . . 76

OOgunnaike, Babatunde A. (TA2-C)Ogunnaike, Babatunde A. (TA2-6) . . . . . . . . . . . . . . . . . 55Ogunnaike, Babatunde A. (TM2-C)Ohlmeyer, E. J. (TM3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Oisiovici, Ronia M. (TP2-C)Oisiovici, Ronia M. (TP2-3) . . . . . . . . . . . . . . . . . . . . . . . . 67Oisiovici, Ronia M. (TP2-4) . . . . . . . . . . . . . . . . . . . . . . . . 67Oliver, Glen C. (WM3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Olson, Keith (MP2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Orlando, G. (WA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Ortiz, Augustine (MP2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 50Ozcelik, Selahattin (MP5-3) . . . . . . . . . . . . . . . . . . . . . . . . 52Ozgen, Canan (MM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

PPachter, Meir (TM5-C)Pachter, Meir (TM5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Pagano, Daniel (TM4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Pait, Felipe M. (MA5-C)Pait, Felipe M. (MA5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Pait, Felipe M. (MA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Palerm, Cesar C. (MP5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . 52Palmor, Zalman J. (P-I-C)Palmor, Zalman J. (TA1-C)Palmor, Zalman J. (TA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . 53Papadimitriou, Georgios I. (WM5-3) . . . . . . . . . . . . . . . 80Papageorgiou, L. (MA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . 38Pasik-Duncan, Bozenna (TM1-5) . . . . . . . . . . . . . . . . . . . 60Pearson, Ronald K. (TA2-2) . . . . . . . . . . . . . . . . . . . . . . . . 54Pereira, Joao A. F. R. (TP2-3) . . . . . . . . . . . . . . . . . . . . . . . 67Perkins, John D. (TM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 60Pettersson, Jens (TP2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Pianese, C. (MA3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Picard, Rick (MP2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Pisano, Alessandro (WM4-6) . . . . . . . . . . . . . . . . . . . . . . . 79Pistikopoulos, Efstratios N. (TM2-2) . . . . . . . . . . . . . . . 60Pizzocchero, F. (MM5-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Polushin, Ilya G. (TA4-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Pomportsis, Andreas S. (WM5-3) . . . . . . . . . . . . . . . . . . 80Ponce, Enrique (TM4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Ponosov, Arcady (TA4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . 58Porto, Domenico (WM1-5) . . . . . . . . . . . . . . . . . . . . . . . . . 76Pottmann, Martin (TA2-6) . . . . . . . . . . . . . . . . . . . . . . . . . 55Przyłuski, K. Maciej (TA3-6) . . . . . . . . . . . . . . . . . . . . . . . 57Puren, Sophie (WA4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

RRabah, Rabah (TP4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Ramon, Herman (MA1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 37Ramon, Herman (MM2-5) . . . . . . . . . . . . . . . . . . . . . . . . . 45Rantzer, Anders (TA3-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 56Raphaeli, Dan (TP4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Raskin, Natalya (MP1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Raskin, Natalya (TP4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Rehbinder, Henrik (TM4-2) . . . . . . . . . . . . . . . . . . . . . . . . 63Rehm, Ansgar (MA1-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Reich, Simeon (TM1-O)Reich, Simeon (TM1-C)Reich, Simeon (TP1-O)Reich, Simeon (TP1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Reich, Simeon (WA1-O)Reich, Simeon (WM1-O)Reich, Simeon (WM1-C)Reich, Simeon (WM1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Rios-Bolıvar, Miguel (WM4-2) . . . . . . . . . . . . . . . . . . . . . 78Rizzo, Gianfranco (MA3-O)Rizzo, Gianfranco (MA3-C)Rizzo, Gianfranco (MA3-1) . . . . . . . . . . . . . . . . . . . . . . . . 39Rodriguez, Julio A. (TA2-1) . . . . . . . . . . . . . . . . . . . . . . . . 54Rodrıguez-Palacios, Alejandro (WA4-C)Rodrıguez-Palacios, Alejandro (WA4-5) . . . . . . . . . . . . 74Romagnoli, Jose A. (“Cacho”) (TA2-1) . . . . . . . . . . . . . 54Rosenwasser, Yephim N. (TA1-1) . . . . . . . . . . . . . . . . . . 53Ross, Roderick (TM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Rotstein, Hector (P-III-C)Rotstein, Hector (MA1-C)Rotstein, Hector (MA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 37Rusnak, Ilan (MM1-C)Rusnak, Ilan (MM1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Rusnak, Ilan (MP5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Rusnak, Ilan (TM3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

SSadot, Dan (WM5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80Sales, Roberto M. (MA1-4) . . . . . . . . . . . . . . . . . . . . . . . . . 37Sanchez, Augustin (TM5-4) . . . . . . . . . . . . . . . . . . . . . . . . 64Sanfilippo, A. (WA5-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Santini, Stefania (MA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 39Sastry, Shankar (TP3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Scala, Stefano (TP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Scali, Claudio (TA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Scarratt, Julie C. (WM4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 78Scherpen, Jacquelien M.A. (MA4-1) . . . . . . . . . . . . . . . . 40Scherpen, Jacquelien M.A. (WM4-4) . . . . . . . . . . . . . . . 79Schnitman, Leizer (MM2-1) . . . . . . . . . . . . . . . . . . . . . . . . 44Schwartz, Howard M. (TM5-1) . . . . . . . . . . . . . . . . . . . . 64Schwartz, Howard M. (WM3-2) . . . . . . . . . . . . . . . . . . . . 77Schwartz, Howard (WM3-C)Sciarretta, Antonio (MA3-6) . . . . . . . . . . . . . . . . . . . . . . . . 40Seatzu, Carla (MM3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Seatzu, Carla (TM5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Sebek, Michael (WA4-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Seginer, Ido (MM5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Seider, Warren D. (TM2-3) . . . . . . . . . . . . . . . . . . . . . . . . . 60Serra, Gabriele (MA3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Seyfried, J. (WA2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Shah, N. (MA2-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38Shaked, Uri (TA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Shaked, Uri (TM1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Shiakolas, Panayiotis S. (WM3-3) . . . . . . . . . . . . . . . . . . 77Shiakolas, Panayiotis S. (WM3-4) . . . . . . . . . . . . . . . . . . 77Shiang, Shen-Po (WA3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 73Shima, Masasuke (TM4-6) . . . . . . . . . . . . . . . . . . . . . . . . . 64Shima, Tal (TM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Shimkin, Nachum (WA5-C)Shimkin, Nachum (WM5-C)Shinar, Josef (TM3-O)

84

Shinar, Josef (TM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Shkolnikov, Ilya A. (WM4-5) . . . . . . . . . . . . . . . . . . . . . . . 79Shtessel, Yuri B. (WM4-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 79Sidi, Marcel (MA1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Sidi, Marcel (MM4-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Sigut, M. (MP1-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Simani, Silvio (MP2-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Sinatra, Mario (TP2-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Sivananthan, Sivaloganathan (MP3-3) . . . . . . . . . . . . . 51Sjoberg, Jonas E. (MM5-O)Sjoberg, Jonas E. (MM5-C)Sjoberg, Jonas E. (MM5-5) . . . . . . . . . . . . . . . . . . . . . . . . . . 48Smagina, Yelena (WA4-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 74Sohn, Hyun-Chull (MM3-1) . . . . . . . . . . . . . . . . . . . . . . . . 45Solo, Victor (WM1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Solovyev, Boris M. (TM2-5) . . . . . . . . . . . . . . . . . . . . . . . . 61Speyer, Jason L. (MP2-C)Speyer, Jason L. (MP2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Sreenivas, Ramavarapu S. (MA2-1) . . . . . . . . . . . . . . . . 38Staffans, Olof J. (TM1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Stylios, Chrysostomos D. (WM2-2) . . . . . . . . . . . . . . . . . 76Stylios, Chrysostomos D. (WM2-4) . . . . . . . . . . . . . . . . . 76Surendra Rao, Alladi (MP3-5) . . . . . . . . . . . . . . . . . . . . . 51Sysel, Martin (TA1-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

TTaylor, James H. (MM2-C)Taylor, James H. (MM2-4) . . . . . . . . . . . . . . . . . . . . . . . . . . 45Telatar, Ziya (WA5-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Terra, Marco H. (WA4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Tiano, A. (MM5-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Tirosh, Dan (WA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Titi, Edriss S. (TM1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Tornambe, Antonio (MP1-5) . . . . . . . . . . . . . . . . . . . . . . . 49Troltzsch, Fredi (WA1-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Tsourdos, Antonios (MM3-3) . . . . . . . . . . . . . . . . . . . . . . 45Tuncay, Serhat (MM2-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Turner, Peter J. (WM3-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Tuzunalp, Onder (WA5-2) . . . . . . . . . . . . . . . . . . . . . . . . . 75

UUcar, Ahmet (TM4-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Usai, Elio (WM4-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Vvan Duijnhoven, Marc (MM3-4) . . . . . . . . . . . . . . . . . . . . 46Van Overschee, Peter (TP5-4) . . . . . . . . . . . . . . . . . . . . . . 70Veglis, Andreas A. (WM5-3) . . . . . . . . . . . . . . . . . . . . . . . 80Venini, P. (MM5-6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Verde, Leopoldo (TP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . 68Verhaegen, Michel (MA4-1) . . . . . . . . . . . . . . . . . . . . . . . . 40Verhaegen, Michel (MM5-3) . . . . . . . . . . . . . . . . . . . . . . . . 48Verhaegen, Michel (WM4-4) . . . . . . . . . . . . . . . . . . . . . . . 79Vesterager, Johan (WA2-4) . . . . . . . . . . . . . . . . . . . . . . . . . 72

WWeiss, George (TP1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Weller, Steve (TA1-C)Weller, Steven (TA1-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Westwick, David T. (MM5-3) . . . . . . . . . . . . . . . . . . . . . . . 48Wetstein, Joseph P. (MM1-5) . . . . . . . . . . . . . . . . . . . . . . . 43White, Brian A. (MM3-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Willett, Peter (MP3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

XXie, Li (MA4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Xue, Dingyu (MA4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

YYaesh, I. (TA5-C)Yaesh, I. (TA5-1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Yaesh, I. (TM3-2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Yamada, Kou (MA4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Yamada, Kou (WA4-3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Yang, Kyung-Jinn (MA3-5) . . . . . . . . . . . . . . . . . . . . . . . . . 40Yaniv, Oded (MA4-C)Yaniv, Oded (MM4-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Yaniv, Oded (TP4-C)Yaniv, Oded (TP4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Ygorra, S. (TM4-4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

ZZaccarian, Luca (MP1-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Zaslavski, Alexander J. (TP1-2) . . . . . . . . . . . . . . . . . . . . 65Zaslavski, Alexander J. (WA1-1) . . . . . . . . . . . . . . . . . . . 70Zeheb, Ezra (WA4-C)Zinober, Alan S. I. (WM4-O)Zinober, Alan S. I. (WM4-C)Zinober, Alan S. I. (WM4-1) . . . . . . . . . . . . . . . . . . . . . . . . 78Zinober, Alan S. I. (WM4-2) . . . . . . . . . . . . . . . . . . . . . . . . 78Zinober, Alan S. I. (WM4-3) . . . . . . . . . . . . . . . . . . . . . . . . 79Zou, Min (MM3-5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

85

86

MED99

Cooperating Organizations

Flight control system example

Our integrated control design solution allowsyou to move smoothly through the stages of yourdesign process. Use MATLAB for data analysisand algorithm development (top), Simulink andStateflow for both dynamic and event-drivensimulation (middle), and Real-Time Workshopfor generating prototype code (bottom).

There are lots ofways to do controldesign, but only oneway to accelerate it.Speed control design projects to completionwith our fully integrated software solution

¯ÈÚ

ÔÂÙÏË

e-mail

„˜ÈÓ

Ò˜Ù

Ì˘‰˜ÏÁÓ

„ÒÂÓØÏÚÙÓ

ÔÈÈÚ†ÌÂÁ˙

˙·Â˙Î

Æ∞≥≠µ∂±†≤π∂≤†ÆÒÓ†Ò˜ÙφÂÈχ†ÁÏ˘Â†‰Ê†Û„†ÌψMATLAB†‡˘Â·†ÈÈÈÚ†¯ÓÂÁ†ÍφÁÏ˘†Â‡

Ó¢Ú·†®±π≤∑©†‡˙Ï„†Ô¯˜ÈÓ‡∞≥≠µ∂±≤π∂≤†∫Ò˜Ù†¨∞≥≠µ∂±µ±µ±†∫ßψ¨∂∑±≥≤†·È·‡≠Ï˙†¨±∞†Í·Èϯ˜†ßÁ¯

ee--mmaaiill:: iinnffoo@@oommiikkrroonn..ccoo..iill

Visit our Web site to see demos and examples andto order a free control design technical kit.

MATLAB ® Control Design Solutionswww.omikron.co.il/matlab.htmlwww.mathworks.com/controls

Move rapidly from initial designconcepts to prototype embeddedcode.With the new control designsolution from The MathWorks,you’ll spend less time coding anddebugging, and more timeexploring ideas and getting results.That’s because our solutionsupports the entire control designprocess with a complete, integratedset of powerful design, simulation,and code generation tools.

It’s based on MATLAB†® andSimulink®, the world’s leadingtechnical computing softwareNow you can explore real-worldsystem behavior while you’re

still in the devlopment stage, thenquickly fine-tune your design.Use MATLAB for data analysisand algorithm development,Simulink and StateflowTM forintuitive and realistic controlsystem simulation, and Real-Time Workshop® for generatingC code that runs on a variety ofhardware target platforms.

See how our control designsolution works for othersSee how leading automotive andaerospace companies areaccelerating development withour control design solution. Theirstories are included in our freetechnical kit. Get yours today.

89

90

91

92

93

94

FOR NOTES

MED99

MED99

FOR NOTES

MED99

MED99

FOR NOTES

MED99

MED99

Maja

Placed Image