Investigate the Crater Wear Monitoring of Single Point Cutting ...

10
Investigate the Crater Wear Monitoring of Single Point Cutting Tool Using Adaptive Neuro Fuzzy Inference System P. Kulandaivelu 1 * and P. Senthil Kumar 2 1 Department of Mechanical Engineering, K.S. Rangasamy College of Technology, Tiruchengode, India 2 Department of Mechanical Engineering, K.S.R. College of Engineering, Tiruchengode, India Abstract This paper deals with the crater wear characteristics of coated carbide tool inserts during dry turning of steel workpieces. Wear also occurs as a result of abrasion, as well as cracking and attrition, with the latter leading to the wearing through of the coating on the rake face under low speed conditions. When moderate speeds and feeds are used, the coating remains intact throughout the duration of testing. Wear mechanism maps linking the observed wear mechanisms to machining conditions are presented for the first time. These maps demonstrate clearly that transitions from one dominant wear mechanism to another may be related to variations in measured tool wear rates. Comparisons of the present wear maps with similar maps for uncoated carbide tools show that TiC coatings dramatically expand the range of machining conditions under which acceptable rates of tool wear might be experienced. However, the extent of improvement brought about by the coatings depends strongly on the cutting conditions, with the greatest benefits being seen at higher cutting speeds and feed rates. Among the available methods, tool condition monitoring using Acoustic Emission Techniques (AET) is an emerging one. Hence, the present work is carried out to study the stability, applicability and relative sensitivity of AET in tool condition monitoring in turning. Attempts were made using Adaptive Neuro Fuzzy Inference System (ANFIS) tool to predict the possibilities of establishing the correlation between the crater wear and the variation of acoustic emission parameters like average value, RMS value and area or energy of the signal. Key Words: Acoustic Emission Parameters, Crater Wear, Stress Wave, Adaptive Neuro Fuzzy Inference System 1. Introduction The Acoustic Emission Technique (AET) is rela- tively recent entry in the field of Non-Destructive Evalu- ation (NDE) which has particularly exhibited a very high potential for material characterization and damage as- sessment in conventional as well as non-conventional materials [1]. Due to its non-destructive nature, it is uti- lized for a wide range of applications. Acoustic Emission (AE) is defined as the class of phenomenon where tran- sient elastic waves are generated by the rapid release of energy from localized sources within a material [2]. In other words, AE refers to the stress waves gener- ated by dynamic processes in materials. Emission occurs as a release of a series of short impulsive energy packets [3]. The energy thus released travels as a spherical wave front and can be picked from the surface of a material using highly sensitive transducers, (usually electro me- chanical type). The picked energy is converted into elec- trical signal, which on suitable processing and analysis Journal of Applied Science and Engineering, Vol. 15, No. 3, pp. 265-274 (2012) 265 *Corresponding author. E-mail: [email protected]

Transcript of Investigate the Crater Wear Monitoring of Single Point Cutting ...

Investigate the Crater Wear Monitoring of

Single Point Cutting Tool Using Adaptive

Neuro Fuzzy Inference System

P. Kulandaivelu1* and P. Senthil Kumar2

1Department of Mechanical Engineering, K.S. Rangasamy College of Technology,

Tiruchengode, India2Department of Mechanical Engineering, K.S.R. College of Engineering,

Tiruchengode, India

Abstract

This paper deals with the crater wear characteristics of coated carbide tool inserts during dry

turning of steel workpieces. Wear also occurs as a result of abrasion, as well as cracking and attrition,

with the latter leading to the wearing through of the coating on the rake face under low speed

conditions. When moderate speeds and feeds are used, the coating remains intact throughout the

duration of testing. Wear mechanism maps linking the observed wear mechanisms to machining

conditions are presented for the first time. These maps demonstrate clearly that transitions from one

dominant wear mechanism to another may be related to variations in measured tool wear rates.

Comparisons of the present wear maps with similar maps for uncoated carbide tools show that TiC

coatings dramatically expand the range of machining conditions under which acceptable rates of tool

wear might be experienced. However, the extent of improvement brought about by the coatings

depends strongly on the cutting conditions, with the greatest benefits being seen at higher cutting

speeds and feed rates. Among the available methods, tool condition monitoring using Acoustic

Emission Techniques (AET) is an emerging one. Hence, the present work is carried out to study the

stability, applicability and relative sensitivity of AET in tool condition monitoring in turning. Attempts

were made using Adaptive Neuro Fuzzy Inference System (ANFIS) tool to predict the possibilities of

establishing the correlation between the crater wear and the variation of acoustic emission parameters

like average value, RMS value and area or energy of the signal.

Key Words: Acoustic Emission Parameters, Crater Wear, Stress Wave, Adaptive Neuro Fuzzy

Inference System

1. Introduction

The Acoustic Emission Technique (AET) is rela-

tively recent entry in the field of Non-Destructive Evalu-

ation (NDE) which has particularly exhibited a very high

potential for material characterization and damage as-

sessment in conventional as well as non-conventional

materials [1]. Due to its non-destructive nature, it is uti-

lized for a wide range of applications. Acoustic Emission

(AE) is defined as the class of phenomenon where tran-

sient elastic waves are generated by the rapid release of

energy from localized sources within a material [2].

In other words, AE refers to the stress waves gener-

ated by dynamic processes in materials. Emission occurs

as a release of a series of short impulsive energy packets

[3]. The energy thus released travels as a spherical wave

front and can be picked from the surface of a material

using highly sensitive transducers, (usually electro me-

chanical type). The picked energy is converted into elec-

trical signal, which on suitable processing and analysis

Journal of Applied Science and Engineering, Vol. 15, No. 3, pp. 265�274 (2012) 265

*Corresponding author. E-mail: [email protected]

can reveal valuable information about the source causing

the energy release [4]. The present global industrial sce-

nario is to produce quality products at competitive price.

This is possible with the increased productivity aimed at

zero error. To achieve this, industries are steering to-

wards ‘un manned factory’where human error is reduced

to a great extent. An essential part of a machining system

in the ‘un manned factory’ is the ability to change out

tools automatically due to wear or damage.

Hotton and Jiang [5,6] has represented that the tool

failure contributes on an average, up to 7% to the down-

time of machining centers. They concluded that most of

the tools fail either by fracturing or gradual wear. Inasaki

and Iwata [7,8] have stated that even though more me-

thods have been developed to monitor tool wear, none of

them has achieved significant use in industry.

A study by Jemielniak and Otman [9] showed that

the AE parameters did not exhibit a definite trend with

tool wear but rather a general random behavior with sud-

den variations related in the process deterioration phe-

nomena. Hence, the present work was carried out to

study the wear monitoring in single point cutting tool

using acoustic emission techniques.

2. Details of Experimental Set Up

Experimental set up includes lathe, work piece, cut-

ting tool, AE sensors, couplants, AE signal storing and

processing instrument among others. These were suit-

ably selected to suit the requirements.

2.1 Work Material

Frequently used material was selected to make the

research work as application oriented. Also, harder ma-

terial was preferred to have the faster rate of tool wear

which would reduce the number of observations re-

quired. To approach these points in mind C45 steel of

270 BHN was chosen and the sensors and preamplifiers

with filters were chosen as given below.

(a) AE sensor

(b) Pre-amplifier with filters

2.2 Cutting Tool

Coated carbide tool was selected based on its wider

application. To have faster flank wear, rough turning

grade of TK35 was chosen. Further the tool holder was

selected to provide zero clearance angles to accelerate

crater wear [10].

2.3 Cutting Conditions

The following cutting conditions were selected based

on the recommended cutting speed range of 110�300

m/minute for the selected work and cutting tool material.

The schematic experimental arrangement is shown

in Figure 1. The preferred work material was cleaned and

set on the lathe [11]. Also the selected cutting tool along

with the tool holder was fixed. The AE sensors for crater

wear were placed in the respective positions, as ex-

plained above, after cleaning the surface and applying

266 P. Kulandaivelu and P. Senthil Kumar

Make FAC 500

S.No 151618

Sensor Element Piezo � electric crystal

Operating Frequency Range 125 kHz�2 MHz

Make Physical Acoustic Corporation

Model 140 B; Gain � 40 dB

Operating Voltage +15 V

Filter 125 kHz � High Pass

Cutting velocity 160 m/min.

Feed 0.05�0.5 mm/rev.

Depth of cut 1.5 mm/cut.

Type of coolant No coolant [Dry condition]

Figure 1. Schematic AET diagram for crater wear.

the couplants. The sensors were fixed by using epoxy

resin couplant. These sensors output signals were fed to

the digitals to storage oscilloscope via pre-amplifier, fil-

ter and power supply unit. The stored signals were pro-

cessed off-line through a computer using ‘AUTODASP’

software.

3. Experimental Procedure

The details of the procedure for the experimental

work carried out are presented below:

i. The machine was set to the selected cutting condi-

tions.

ii. The oscilloscopes were set in ‘Auto Arm’ mode to

receive and store 15 frames of signals automatically.

As the machining interval was decided as 30 sec-

onds duration, each signal frame was of 2 seconds

duration.

iii. The machine was started and simultaneously the os-

cilloscopes were armed to capture the AE signals

generated due to crater wear [12].

iv. The machine was stopped at the end of 30 seconds

and the oscilloscopes were capturing and storing AE

signals generated due to crater wear in 15 frames of

2 seconds during this 30 seconds interval of metal

cutting.

v. Tool was taken out from the tool holder and cleaned

with carbon tetrachloride.

vi. AE signals stored in the oscilloscopes due to crater

wear are transferred to a computer through RS 423

serial interface for analysis at a later stage.

vii. The tool was again fixed to the tool holder with the

same cutting edge in cutting position.

viii. The step from 1 to 7 was repeated for next observa-

tion.

Like this, 40 observations were noted, which means

the experiment was carried out for 20 minutes and 800

AE waveforms of each 30 seconds duration were cap-

tured for the crater wear.

The captured signals were processed using ‘AUTO

DASP’ software. The AE signals in the oscilloscope

screen and the same in the monitor of computer when

transferred respectively. Inasaki [7] have stated in their

research investigations that a wide variety of measure-

ments can be carried out according to the type of AE sig-

nal, the experimental instrumentation used and the per-

sonal preferences of the individual research workers.

One of the most significant methods to analyze AE is

the measurement of the energy content of the AE signals.

The rate with which the energy is transmitted by the sig-

nal can be directly correlated with the rate of energy

generation by the original AE source [9]. A simple way

to measure such energy is the evaluation of the RMS

voltage of the signal.

Hence, in this work, AE parameters like the average

value of the signal (mV), RMS value of the signal (mV)

and area or energy of the signal (mV) were found using

‘AUTO DASP’ software. In this experimental work, AE

released due to wear, observed at each intervals of dura-

tion of 30 seconds was stored in 40 frames and so, 40 set

of values of AE parameters were found [13,14]. There-

fore, the means of these values were observed and from

these, the cumulative mean AE parameters are calcu-

lated. Table 1 gives the values for crater wear along with

cumulative mean AE parameters.

4. Verification of AE Experimental Work

on TCM

The AE parameters against crater wear were experi-

mentally studied. The observations have been analyzed

for correlations between the tool wear and the AE para-

meters and the results are presented in the following sec-

tions.

4.1 Crater Wear vs. AE Signal

Figure 2 shows the variation in crater wear with

time. It clearly indicates that the three stages of wear viz.,

stage-I where the rate of wear is low, stage-II where the

rate of wear is moderate and stage-III where the rate of

wear is faster and leading to the termination of tool life.

In stage-I, 0 to 8.5 min of the machining operation the

tool wear is in the range of 0-8.3 �m only.

But in the stage-II and stage-III the crater wear in-

creases with increase in machining time and the tool fails

at about 20 min from the starting of the machining opera-

tion.

4.2 Crater Wear vs. Cumulative Mean AE

Parameters

The variation of Crater wear with mean AE para-

Investigate the Crater Wear Monitoring of Single Point Cutting Tool Using Adaptive Neuro Fuzzy Inference System 267

meters (mean average value, mean RMS value and mean

area) are shown in Figure 3.

This reflects the random variations of mean AE

parameters such as, mean average value, the mean RMS

value and the mean area of the AE signal [15]. The be-

havior of random variation is due to cumulative crater

wear, whereas AE parameters are due to release of AE

signals at that instant. The variation of cumulative mean

AE parameters and cumulative mean area with crater

wear is shown in Figure 4.

Crater wear occurred in three stages is shown in Ta-

ble 2. Cumulative AE parameters plotted against the

crater wear are shown in Figure 4. By noting the limiting

values of cumulative mean AE parameters at these tran-

sition stages, the state of crater wear can be monitored.

From the Table 2, the crater wear in stage-I occurs if

the cumulative mean average value, mean RMS value

and mean area of AE signal are in the range of 0485 mV,

0�711 mV and 0�1150 mV.s, respectively. The same in

stage-II, varies in the range of 485 to 900 mV, 711 to

1254 mV and 1150 to 2029 mV.s and in stage-III, the re-

spective cumulative mean AE parameters are above 900

mV, above 1254 mV and above 2029 mV.s, respectively.

4.3 Time vs. Cumulative Mean AE Parameters for

Crater Wear

Figure 5 shows the variation of cumulative mean AE

parameters and cumulative mean area with cutting time.

Like crater wear vs cumulative mean AE parameters, this

plot does not clearly identify the three stages of wear, but

there is still minor identification of the stages as marked

by arrows in Figure 5, at the corresponding transitions in

the curve.

However, unlike the transitions in the crater wear vs.

cumulative mean AE parameters curve (Figure 4), these

transition points slightly differ from that of the wear

268 P. Kulandaivelu and P. Senthil Kumar

Table 1. Crater wear along with cumulative mean AE

parameters

Mean AE parameters

S.NoTime

(min)

Crater

wear

(�m)

Average

value

(mV)

RMS

value

(mV)

Area

(mVs)

01 00.5 00.3 0000.678 0000.789 0002.135

02 01.0 00.6 0040.801 0051.030 0086.878

03 01.5 01.3 0057.185 0089.024 0141.050

04 02.0 02.0 0081.642 0137.212 0213.365

05 02.5 02.5 0108.966 0184.554 284.73

06 03.0 03.2 0137.603 0233.334 0358.108

07 03.5 03.6 0169.134 0279.664 0432.188

08 04.0 04.1 0203.643 0327.330 0508.354

09 04.5 04.5 0231.677 0369.477 0577.911

10 05.0 04.8 0266.762 0420.753 0658.973

11 05.5 05.6 0301.015 0470.360 0735.116

12 06.0 06.0 0336.085 0515.459 0811.220

13 06.5 06.5 0366.738 0553.200 0876.022

14 07.0 06.9 0398.014 0590.843 0936.460

15 07.5 07.3 0432.699 0634.156 1010.180

16 08.0 07.8 0462.756 0674.556 1082.118

17 08.5 08.2 0485.942 0711.514 1142.157

18 09.0 09.0 0511.928 0746.468 1201.156

19 09.5 09.8 0543.832 0789.503 1272.217

20 10.0 10.5 0574.655 0835.704 1342.756

21 10.5 11.2 0604.313 0877.954 1412.127

22 11.0 12.0 0630.302 0914.031 1471.594

23 11.5 12.6 0658.343 0950.074 1530.632

24 12.0 13.6 0685.432 0986.124 1590.385

25 12.5 14.1 0715.118 1019.570 1640.430

26 13.0 14.6 0741.878 1050.227 1685.941

27 13.5 15.4 0763.015 1078.199 1732.081

28 14.0 16.2 0793.631 1113.536 1795.102

29 14.5 16.9 0821.855 1149.216 1855.299

30 15.0 17.3 0848.973 1182.755 1912.568

31 15.5 18.4 0875.062 1217.805 1970.321

32 16.0 19.4 0900.248 1254.763 2029.360

33 16.5 21.1 0928.472 1293.443 2095.557

34 17.0 22.9 0957.684 1332.685 2159.771

35 17.5 24.9 0986.047 1370.901 2223.029

36 18.0 27.1 1014.743 1406.913 2283.239

37 18.5 29.9 1042.108 1442.119 2340.902

38 19.0 31.2 1072.377 1479.360 2405.798

39 19.5 34.2 1102.589 1524.144 2476.798

40 20.0 41.2 1134.120 1569.474 2550.878

Figure 2. Variation of crater wear with time.

Investigate the Crater Wear Monitoring of Single Point Cutting Tool Using Adaptive Neuro Fuzzy Inference System 269

Figure 3. Crater wear with mean AE parameters.

Figure 4. Crater wear with cumulative mean AE parameters.

Figure 5. Time with cumulative mean AE parameters.

Table 2. Trend of crater wear with cumulative mean AE parameters

Stages of crater wear Stage I Stage II Stage III

Crater wear (�m) 0�8.3. 08.3�19.4 .Above 19.4Time (min) 0�8.5. 8.5�16. Above 160Cumulative mean average value (mV) 0�485 485�900 Above 900Cumulative mean RMS value (mV) 0�711 0711�1254 0Above 1254Cumulative mean area (mV.s) 00�1150 1150�2029 0Above 2029

curve (Figure 3). First transition point coincided with

the wear curve, but the second one occurred at the 14th

min instead of the 16th min. Even then, these transitions

in the time vs. cumulative mean AE parameters curve are

also suggested as a measure to monitor the stage of crater

wear due to the following reasons [16].

In crater wear vs. cumulative means AE parameters,

it is necessary to measure the wear initially to study the

trend, whereas in time vs. cumulative mean AE para-

meters, without any addiction on the wear, this curve can

be plotted during the process and the transitions in the

curve can be monitored to identify the stages of wear.

The research findings are further confirmed by not-

ing the patterns of the signals in these three stages of

wear, which are presented in Figures 6, 7 and 8, respec-

tively. These figures clearly show that the signals contain

many variations in amplitude level in the first two stages

of wear, whereas the variation in amplitude level is less

in the third stage of wear [17].

The reason for this change in the type of signals can

be attributed to the rate of wear. The rate of crater wear is

slow in the first stage and moderate in the second stage.

Hence the acoustic emission due to wear occurs with

definite time gap between each emission and so, the sig-

nals are of burst type in these stages of wear. In the final

or terminal stage of wear, the wear rate is faster loading

to successive emissions without any time gap between

the events and overlapping one over the other. Conse-

quently, the final (third) stage of wear emits continuous

emission [18].

On-line tool condition monitoring is suggested to

monitor the co-efficient of variation of RMS value of the

signal. In this case, the crater wear would be in stage-I,

stage-II and stage-III, when the co-efficient of variation

of RMS value is above 0.061, between 0.061 and to

0.042 and below 0.042, respectively.

5. ANFIS Based Crater Wear Prediction

Adaptive Neuro Fuzzy Inference System (ANFIS) is

a class of adaptive network that act as a fundamental

framework for adaptive fuzzy inference system [19].

Figure 9 shows the typical ANFIS architecture. Assume

that fuzzy inference system has two inputs X and Y and

one output F. Every input has two fuzzy sets A1 A2 and

B1 B2. The rule base system has two if-then rules of

Takagi-Sugeno’s type are [20,21].

Rule 1: If x is A1 and y is B1, then f1 = p1 x + q1 y + rl

Rule 2: If x is A2 and y is B2, then f2 = p2 x + q2 y + r2

The proposed ANFIS based crater wear predication

system was developed using MATLAB. Initially the sys-

tem was developed a with three inputs parameters, aver-

270 P. Kulandaivelu and P. Senthil Kumar

Figure 6. AE wave form captured crater wear at stage-I (5min and 50 s after start of machining).

Figure 7. AE wave form captured crater wear at stage-II (11min and 20 s after start of machining).

Figure 8. AE wave form captured crater wear at stage-III (16min and 25 s after start of machining).

age value, RMS value area and then later another input

time has been added to the system. The crater wear was

used as an output variable.

Three bell-shaped Member ship Function (MF), are

used for both three inputs and four input system. The lin-

guistic variable such as low (L), medium (M) and high

(H) are assigned to all the inputs. All the possible combi-

nation if-then rules were framed [22]. The system archi-

tecture of this proposed three input and four input system

are shown in Figures10 and 11 respectively.

5.1 Simulation Result

The 29 number of measured experimental data set

were taken in each stage (29 � 4 = 116) for training the

ANFIS model. The 11 number of measured experimental

data set were taken in each stage (11 � 4 = 44) for testing

the developed model.

The developed model was trained by hybrid learning

algorithm with 250 iterations. The performance of the

model was justify by root mean square value (RMSE) in-

dex and co-efficient of fitting (R). The training perfor-

mances are of the three input and four input system are

shown in Figures12 and 13 respectively.

According to training performance, the three input

system having minimum RMSE is 0.015. and four input

system having RMSE is 0.04. The co-efficient of fit (R)

values are calculated by linear curve fitting method and

the R value for three input system was 1 and R value for

four input system was 0.999 are shown in Figures 14 and

15.

The valediction of adoptiveness to the system with

input parameter is done by using ANFIS. By observing

the mean square error value from the ANFIS, the system

is adopted with the input parameter.

Based on the above performance result, the three

input system ANFIS model was most suitable for pre-

dicting the crater wear. The proposed ANFIS model

based predicated crater wear and experimentally mea-

Investigate the Crater Wear Monitoring of Single Point Cutting Tool Using Adaptive Neuro Fuzzy Inference System 271

Figure 9. Equivalent ANFIS.

Figure 10. Structure of three input ANFIS for analysis ofcrater wear.

Figure 11. Structure of four input ANFIS for analysis of craterwear.

sured value are tabulated in Table 3. The measured value

and the simulation output of the Crater wear are shown in

Figure 16.

6. Conclusion

The summary of the extensive experimental research

work on tool condition monitoring through AET is pre-

sented below.

i. The discrimination of crater wear on AE and also

enhancing the individual effects can be achieved by

positioning the sensor for crater wear on the top sur-

face of the tool holder and the sensor for flank wear

on the side surface of the tool holder, adjacent to

flank face. This has been proved experimentally.

ii. The present study was conducted with suitable sen-

sors for the frequencies between 125 kHz to 2 MHz.

Normally the effect of AE signals are dominant

when frequencies range beyond 200 kHz, it can be

clearly observed by the sensor.

iii. Crater wear stages can be monitored by observing

cumulative mean values of AE parameters like area,

RMS value and average value.

iv. The limiting values of AE parameters obtained to

monitor tool condition for a given cutting conditions

is found to be applicable to monitor tool condition,

even when the cutting speed is varied within �12%

by keeping all other cutting conditions constant.

v. The ANFIS has been trained by using the acoustic

emission parameters like average value, RMS value

and area or energy of the signal as an input para-

meters and crater wear as an output parameters.

vi. The simulation results show that the predicted va-

lues of the crater wear for three input network gives

almost closer value to the actual measured values.

Nomenclature

AE Acoustic Emission

272 P. Kulandaivelu and P. Senthil Kumar

Figure 12. Iteration vs. error characteristics for 3 inputs.

Figure 13. Iteration vs. error characteristics for 4 inputs.

Figure 14. Testing performance curve for 3 inputs.

Figure 15. Testing performance curve for 4 inputs.

AET Acoustic Emission Technique

rev. Revolution

RMS Root Mean Square

TCM Tool Condition Monitoring

mV Milli Volt

mV.s Milli Volt second

�m Micro meter

References

[1] Kandasami, G. S., “Application of Acoustic Emission

Technique to Studies of Machining Processes,” Ph.D.

Thesis, Anna University, pp. 127�128 (1988).

[2] Cook, N. H., “Tool Wear Sensors,” International Jour-

nal on Science and Technology of Friction, Lubrica-

tion and Wear, Vol. 62, pp. 49�57 (1980).

[3] Lidan and Mathew, J., “Tool Wear and Failure Moni-

toring Techniques for Turning - A Review,” Interna-

tional Journal of Machine Tools and Manufacture,

Vol. 30, pp. 579�598 (1990).

[4] Viktor, P. A., “The Assessments of Cutting Tool Wear,”

International Journal of Machine Tools and Manufac-

ture, Vol. 44, pp. 637�647 (2004).

[5] Hotton, D. V. and Qinghuan, Yr., “On the Effects of

Built Up Edge on Acoustic Emission in Metal Cut-

ting,” Journal of Engineering for Industry, pp. 184�

189 (1999).

[6] Jiang, C. Y., Zhang, Y. Z. and Xu, H. J., “In-Process

Monitoring of Tool Wear Stage by the Frequency Band

Energy Method,” Annals of the CIRP, Vol. 36, pp.

45�48 (1996).

[7] Ichiro Inasaki, “Application of Acoustic Emission

Sensor for Monitoring Machining Processes,” Ultra-

sonics, Vol. 36, pp. 273�281 (1998).

[8] Iwata, K. and Moriwaki, T., “An Application of

Acoustic Emission Measurement to In-Process Sen-

sing of Tool Wear,” Annals of the CIRP, Vol. 25, pp.

21�26 (1976).

[9] Jemielniak, K. and Otman, O., “Catastrophic Tool

Failure Detection Based on Acoustic Emission Signal

Analysis,” Annals of CIRP, Vol. 47, pp. 31�34 (1998).

[10] Blum, T. and Inasaki, I., “A Study in Acoustic Emis-

sion from the Orthogonal Cutting Process,” Journal of

Engineering for Industry, Vol. 112, pp. 203�211 (1990).

[11] Everson, K., Grbec, P. and Leskovar, P., “Acoustic

Emission of a Cutting Process,” Ultrasonics, Vol. 15,

pp. 17�20 (1977).

[12] Rangawala, S. and Dornfeld, D., “Study of Acoustic

Emission Generated during Orthogonal Metal Cut-

ting,” Int. J. Mech. Sciences, pp. 489�499 (1991).

[13] Jemielniak, K., “Some Aspects of AE Application in

Tool Condition Monitoring,” Ultrasonics, Vol. 38, pp.

604�608 (2000).

[14] Ravindra, H., Srinivasa, Y. and Krishnamurthy, R.,

“Acoustic Emission for Tool Condition Monitoring in

Metal Cutting,” Wear, Vol. 212, pp. 78�84 (1998).

[15] Yang, C. F. and Houghton, J. R., “Acoustic Emission

True RMS Signals Used to Indicate Wear of a High

Speed Ceramic Insert,” Journal of Mechanical Work-

ing Technology, Vol. 20, pp. 79�91 (1998).

[16] Li, X., “Acoustic Emission Method for Tool Wear

Monitoring during Turning,” International Journal of

Investigate the Crater Wear Monitoring of Single Point Cutting Tool Using Adaptive Neuro Fuzzy Inference System 273

Table 3. Crater wear comparative results

Predicted ANFIS resultsS.No.

Measured crater

wear values 3 Inputs 4 Inputs

01 02.5 2.6117 2.6969

02 03.2 3.2344 3.2307

03 03.6 3.6587 3.4952

04 04.1 4.0360 3.5518

05 04.5 4.3051 4.1191

06 12 11.804 11.939

07 12.6 12.454 12.715

08 13.6 13.314 13.701

09 14.1 13.961 14.288

10 27.1 27.166 27.764

11 29.9 28.923 28.820

Figure 16. Comparative results of crater wear.

Machine Tools and Manufacture, Vol. 42, pp. 157�165

(2002).

[17] Yang, C. F. and Houghton, J. R., “Acoustic Emission

True RMS Signals Used to Indicate Wear of a High

Speed Ceramic Insert,” Journal of Mechanical Work-

ing Technology, Vol. 20, pp. 79�91 (2001).

[18] Zhang, C. and Wang, X., “Investigation on the Feasi-

bility of Monitoring Chip Seats Using Acoustic Emis-

sion Technique,” Chinese Journal of Mechanical En-

gineers, pp. 53�59 (1989).

[19] Roger Jang, J. S., “ANFIS: Adaptive Network Based

Fuzzy Inference System,” IEEE Transactions on Sys-

tems, Man and Cybernetics, Vol. 23, pp. 665�685

(1993).

[20] Takagi, T. and Sugeno, M., “Derivation of Fuzzy Con-

trol Rules from Human Operator’s Control Actions,”

in Proceeding of IFAC Symposium Fuzzy Information

Knowledge Representation and Decision Analysis, pp.

55�60 (1983).

[21] Takagi, T. and Sugeno, M., “Fuzzy Identification of

Systems and Its Application to Modeling and Con-

trol,” IEEE Transactions on System, Man, Cybernatic,

Vol. 15, pp. 116�132 (1985).

[22] Geethanjali, M. and Mary Raja Slochanal, S., “A Com-

bined Adaptive Network and Fuzzy Inference System

(ANFIS) Approach for Over Current Relay System,”

Elsevier Publication (2008).

Manuscript Received: Jul. 19, 2011

Accepted: Dec. 2, 2011

274 P. Kulandaivelu and P. Senthil Kumar