Vous êtes sur la page 1sur 13

Wear, 152 (1992) 395-407 39.

Correlating tool life, tool wear and surface roughness by


monitoring acoustic emission in finish turning

A. E. Dir&
State University of Campinas, CP 6122, CEP 13081, Campinas, SP (Brazil)
J. J. Liu and D. A. Dornfeld
Department of Mechanical Engineering, University of California at Be&&y, Etcheverry Hall
Room 5114, Berkdqy, CA 94720 (USA)

(Received May 20, 1991; revised and accepted July 22, 1991)

Abstract

Experiments have been conducted in an attempt to monitor the changing of workpiece


surface roughness caused by the increase of tool wear, through the variation of acoustic
emission in finish turning, under different cutting conditions. The signal-processing analysis
was done on the raw signal, on the AE,, signal fiftered using a high bandpass and on
the AE,, signal fi’ltered using a smaller bandpass. The relationship among several parameters
of acoustic emission such as zero crossing rate, mean AEm, and standard deviation of
AE,, was established. The material machined was 1045 steel. The cutting force was also
monitored. The results show that acoustic emission can be a good way to monitor on-line
the growth of surface roughness in finish turning and therefore can be useful for establishing
the end of tool life in these operations. Based on the results obtained pointing out the
best acoustic emission parameters to monitor surface roughness, a set-up is proposed to
reach to this goal.

1. Introduction

Acoustic emission has been used to accomplish many goals in machining processes.
It has been used to study the fundamentals of the machining process, e.g. the chip
formation mechanism, and the contribution of some phenomena to the energy generated
in cutting, e.g. shear and plastic deformation of the chip and the friction between the
chip and tool rake face and the workpiece and tool flank face [l-4]. Another area
where acoustic emission has been evaluated extensively is for on-line monitoring of
tool wear 14-71 and detecting tool fracture [7, 81 in both turning and milling processes.
It has also been used in grinding [9-111 to monitor the spark-out operation, wheel
life, wheel dressing and in-feed control for plunge grinding. In diamond turning acoustic
emission has been used to monitor the actual depth of cut [12].
In finish turning, where the tool replacement is defined by either the workpiece
dimension or surface roughness and the tool wear and machining conditions are not
aggressive enough to fracture the tool, it is very important to monitor the surface
roughness to establish the moment to change the tool. Since the increase of surface
roughness is caused by the increase of tool wear 1133 and, as stated, acoustic emission
has been used to monitor tool wear, it may be applicable to monitor the increase of
surface roughness too. The main purpose of this work is to establish the relationship

0043-lf548/92/$5.00 0 1992 - Elsevier Sequoia. Al1 rights reserved


396

between acoustic emission and workpiece surface roughness in finish turning as a basis
for determining the moment to change the tool in these operations.

2. Surface roughness, tool wear and acoustic emission

Many factors influence the formation of surface roughness in the turning process.
These factors include chip deformation and side flow, vibration of the ma-
chine-tool-fixture-workpiece system, the geometrical contribution of the feed q) and
tool nose radius (r) and so on. This geometrical contribution can be calculated by
[I41
f"
R ccc----
i.1)
max8r
or

where R,, is the maximum peak-to-valley roughness and R, is the roughness average.
The geometry of tool wear (Fig. 1) causes a change in surface roughness as
machining time elapses. Groove and flank wear are the two kinds of wear that most
influence this change in surface roughness. The groove wear changes the tool nose
curvature and this is reflected in the workpiece surface. It also increases the chip side
flow [16]. Many authors have studied the relationship between surface roughness and
flank wear (I$). Sundaram and Lambert [17] studied turning of AISI 4140 steel with
carbide tools. Their results are shown in Fig. 2. This graph shows that there is an
increased amplitude of the surface roughness at the beginning of cut, a decreased
tendency in the middle region and again an increased tendency at the end of wear.
The relationship among R,,, R, and Vi, with cutting length (IC) was studied by
Petropoulos [18] and the results for machining AISI 1060 steel are shown in Fig. 3.
This figure shows that R, and R,, increase until V, reaches 0.2 mm. Above this value
R max oscillates around a constant value and R, oscillates as it increases. Flank wear
increases continuously. This fact proves that the flank wear formation is not enough
to explain the surface roughness changing with the cutting time.
Many authors have also successfully studied the possibility of monitoring the tool
wear using acoustic emission (AE) parameters such as AE,,, AE mode, skew, kurtosis
and spectral analysis [5-7, 19, 201. Since tool wear is responsible for the change in
surface roughness, acoustic emission may be useful for monitoring surface roughness
in the turning process. Most of these studies have only considered the flank wear

Fig. 1. Geometry of tool wear [15].


397

s
. ,. *

I i Vb
:_:: . ,” (mm)
I .
. a. * 4
r *
,*“ T
2
/
I
0
I
!OO
I
200
I
300
I
400 500
I, 0 iO 20 30 40 50 60 70 8

Tool Wear (pm) ienqtn of Cutting ikml

Fig. 2. Surface roughness (r.m.s.1 as a function of tool flank wear [17].

Fig. 3. R,, R,,, and V, as a function of length of cutting [18].

influence on the AE parameters, but Lan and Naerheim 121, 221 concluded that when
crater wear occurs along with flank wear, one influence cancels the other. AE,, tends
to become constant as cutting time elapses, since the increase in flank wear increases
A&, and the increase in crater wear decreases A&,, owing to the increase in the
effective rake angle.

3. Signal processing of AE

The cutting conditions used in this work were close to the recommended cutting
conditions in finish turning of 1045 steel. In these conditions there will be the occurrence
of both flank and crater wear, as can be seen in Section 5. Therefore, as stated above,
A,!?,, is not sufficient alone to monitor the increase in wear and surface roughness.
To overcome this, many other parameters of AE were used in an attempt to carry
out this purpose. This section talks about these parameters, their definition and physical
meaning.

3.1. Spectral analysis


The frequency spectrum of a signai is composed of an it&rite number of parameters
(each frequency has its own amplitude), which helps in the character~ation of the
phenomenon sources. The tool wear growth and consequently the surface roughness
increase can induce changes in the frequency spectra in such a way that allows their
indirect monitoring. The best-known mathematical procedure to calculate the frequency
spectra is the fast Fourier transform (FIT) algorithm, which is often used in AI?
signal processing.

3.2. Acoustic em&ion count, count rate and zero crossing rate
AEZ count is the number of times the signal overcomes a given threshold. If this
counting is done during a given period of time (usually between 0.1 s and 1 h), the
m~mum count for each time period is the count rate. This counting can be done
by software or by hardware using an electronic counter 123, 241. The dependence of
the count upon the threshold setting can lead to problems when the cutting conditions
are changed, since when this happens, the signal ~pIitude changes, which makes the
398

count meaningless if the threshold is not changed too. One way to avoid this problem
is to set the threshold at zero. The resulting parameter, called the zero crossing rate,
represents the weighted mean frequency of the signal [25].

3.3. Root mean square of the AE signal (A.&J


The r.m.s. voltage VRMs of a signal u(t) from an r.m.s. meter is usually given as
, I,2

~RMSO)
= u2(r)w(t - T) d,r

where w(t) is usually an exponential lowpass windowing function which can be expressed
as
t-r
w(t)= $exp - RC (4)
i 1
RC in this equation is the time constant which determines the corner frequency of
the lowpass filter.
This parameter has been used many times in flank wear monitoring. Since the
power of the signal is proportional to the square of AE,, [2], the increase in cutting
energy generated by the increase in tool flank wear should generate an increase in
this parameter, while the increase in crater wear should decrease its value owing to
the increase in the effective tool rake angle [21, 221. A&,,, is measured by either
software or hardware using an r.m.s. voltmeter. A time constant (AT) must be established
in the voltmeter to carry out this job.

3.4. Standard deviation, skew and kurtosis


The three parameters standard deviation, skew and kurtosis are often used to
characterize the statistical distribution of the data. Equations (S)-(7) express the
definition of standard deviation, skew and kurtosis [20] respectively:

SD = MiR (5)

where

M,= ; k&4Ek-AE) (8)

The skew measures the symmetry of the function about its mean level while the
kurtosis is a measure of the sharpness of the peaks. A negative skew generally indicates
a shift of the distribution to the left of the mean while a positive skew indicates a
shift to the right of the mean. A high kurtosis value implies a sharp distribution while
a low kurtosis value implies virtually flat characteristics. Experimental results [20]
indicate that with increasing tool wear the skew increases while the kurtosis decreases.
These parameters can be measured either by software or hardware using a combination
of r.m.s. voltmeters.
399

4. Materials, equipment and experimental methods

A series of machining tests was run to evaluate the sensitivity of the AE signal
to changes in surface finish as in~uenced by tool wear. The material tested was AISI
1045 steel. The diameter of the workpieces varied from 127 to 100 mm during the
experiments and the machined length was 83 mm. The feed (f) and cutting speed (s)
varied from 0.14 to 0.22 mm rev-’ and from 200 to 340 m min-’ respectively as
follows: experiment 1, f= 0.18, s = 340; experiment 2, f= 0.18, s = 200; experiment 3,
f= 0.14, s = 200; experiment 4, f= 0.22, s = 200; experiment 5, f= 0.18, s = 270.
The depth of cut was maintained at d=0.5 mm for all experiments. The inserts
and tool holder used were respectively TNMP 16 04 12 K-PI5 (without coating) and
CTGNR 2525 M16. This insert was chosen because it is suitable for typical finish-
turning operations used in the experiments. The acoustic emission and cutting force
were measured with the set-up shown in Fig. 4. The total gain in AE signal (preamplifier
and amplifier) was 40 dB and a wideband AE sensor was used. The workpiece surface
roughnesses were measured using a Talysurf 10 Taylor-Hobson instrument.
An experiment was finished when the surface roughness due to the increase in
tool wear reached a value (R,) that was at least twice the value of the first cut. The
cutting experiments took place in a CNC lathe with a 5 hp spindle motor and the
tool macrophotographs were taken in a scanning electronic microscope (SEM).
The recorded raw signal was digitized using a filter with a high bandpass (SO-500
kHz). After some analysis of these data the 200-300 kHz frequency band was chosen
as the best frequency range to obtain the AE,, signal. Thus signal processing was
done with three different approaches: first, analysis of the raw signal was done; then
AEm, was analyzed with the data filtered in a SO-500 kHz band and a 20&300 kHz
band.

Fig. 4. Experimental set-up.


Fig. 5. {a) Flank and (b) rake face of a worn tooi (experiment 5); cutting time 7.5 min,
amplification 100 times.

5. Results and discussion

Figure 5 shows SEM photographs of both the rake face and flank face of a worn
tooi. It can be seen that crater and flank wear are present. These pictures were taken
using the tool of experiment 5 and are representative of tool wear in the others.
Figure 6(a) shows the behavior of flank wear and surface roughness vs. cutting time
for experiment 5. It can be seen that the growth in surface roughness was similar to
the increase in flank wear. It can also be seen in Fig. 5 that no groove wear was
formed, so that the flank wear was the main cause of the surface roughness increase,
as seen in Fig. 6.
Figure 6(b) shows the surface roughness increase as the cutting time elapses for
all cutting conditions tried. It is interesting to notice that neither feed nor cutting
speed had much influence on the R, value, at least at the beginning of the test. As
Ra (urn) 3 4 r , ,950 Flank Wear
865 Vb(P)
32
780
30
28 695
va
26 biO

24 52s
I?&
?Z 440

35s
2?
- 270
Ie
I t’ 185

14 -I:::)i ICO
0 I --p---y- 4 5 c ! 8

@I Cutting Tirnr irnirrl


Fig. 6. Flank wear and surface roughness vs. cutting time: (a) V, and R, of experiment 5; (b)
R, of ali experiments.

the cutting time elapses, the curves have a different behavior owing to the different
influence that the cutting conditions have on the tool wear growth. This figure also
shows that the feed had more influence on the surface roughness growth than did
the cutting speed. This fact can be seen by looking at the slopes of the curVes. The
curves with the same feed and different cutting speeds have a similar sIope (linear
coefficients varying from 0.22 to 0.25) while the curves with the same cutting speed
and different feeds have very different slopes and the slope increases with increasing
feed (linear coefficients varying from 0.18 to 0.41).
The small influence of cutting speed on surface roughness was expected since
this is often cited in the literature, but the small influence of feed was not expected
since there is a strong geometrical contribution of feed to surface raughness (eqns.
(1) and (2)). Shaw [14] points out the growth in specific cutting energy with decrease
in the uncut chip thickness (or the feed), which causes plastic side flow of the material.
This effect makes the influence of feed on surface roughness not as high as stated
in eqns. (1) and (2). Since in this case this influence was so small, it can be concIuded
402

that the increase in specific cutting energy was large enough to compensate tar the
geometrical contribution. The cutting force measurements carried out in the experiments
confirm this fact. It was seen that the influence of feed on cutting force was also
negligible. It is possible that the increase in specific cutting pressure (or cutting energy)
was of the order of the decrease in feed to keep the force approximately constant,
since both feed and specific cutting energy influence the force.

5.2. Raw signal analysh


Figure 7 shows the raw AE signal frequency spectrum vs. cutting time in the
50-500 l&z range of experiment 1. The other experiments showed a very similar
behavior. It can be seen from this figure that the frequency spectrum can hc split
into three regions. At low frequencies (between SO and 200 kHz) the amplitude
behavior is not meaningful because there is no trend as the cutting time goes on. In
the frequency range between 200 and 300 kHz the signal amplitude increases with
the cutting time, which is very meaningful because the tool wear and the surface
roughness (Fig. 6) also increase with cutting time. The third region (above 300 k_Hz)
gives no information.
Figure 8 shows an increasing tendency of zero crossing rate (ZCR) with surface
roughness. This is compatible with what was said above since this parameter estimates
the mean frequency of the signal and reflects the increase in amphtude in the 200-300
kHz range with cutting time and consequently with surface roughness. From this figure
it can also be seen that the feed influence on ZCR values was negligible. From these
experiments it is impossible to evaluate the cutting speed influence since the ZCR
exhibits no clear trend with cutting speed.
Despite the general increase in ZCR with surface roughness, it cannot be concluded
that it is a good parameter to monitor surface roughness because ZCR values show
a substantial scatter as they increase. However, this does not rule out the use of the
ZCR. In this study each datum point in the figure was calculated using 4096 points
sampled at 1 MHz, since this calculation was based on the raw signal. Each datum
point represents only 4 ms of cutting time, which is very littie time to represent the
cutting conditions for one workpiece. Thus it is impossible to affirm that the level of
scatter of ZCR values will continue as seen here when a larger data sample is taken.
Only the general tendency of the data can be analyzed here and the ZCR tendency

Fig. 7. Frequency spectrum vs. cutting time for experiment 1 (frequency band 50-500 kHz).
Fig. 8. ZCR vs. R,.

Fig. 9. A&,,, vs. R, (frequency band SO-500 kHz).

shows that this parameter may be a good way to monitor surface roughness growth.
When this parameter is measured on-line, the problem of small sample sizes will not
exist and then the ZCR of the raw signai should be a good parameter to use for
roughness assessment.

5.3. Root-mean-square nnaijsis (50-500 k&z band data)


Figure 9 shows the behavior of AE,, (in this frequency band) us. R,. The small
influence of feed and the large influence of cutting speed on,&&,,, values were expected
since they were cited in the literature. What is interesting to note is that the increase
in AEm, was greater when the cutting speed changed from 200 to 270 m min.-’
compared to the increase when the speed increased from 270 to 340 m min-‘. Another
observation from Fig. 9 is that the increase in AEm, with surface roughness and
consequently with tool wear was small for al1 cutting conditions tried, much smaher
than the increase in surface roughness. This is due to the fact that flank and crater
wear occurred together (Fig. 5) in the inserts of these experiments, blunting the increase
in AE,,,. Even this small increase in A&,, is most likely caused by the increase in
scatter of AE,, as can be seen in Fig. 10. This figure shows a typical sample of data
from experiments. At the beginning of the cut all data are concentrated around an
average value. When the tool is worn (and the surface roughness is greater), most of
the data are still concentrated around the same value but more scatter (higher standard
deviation) is seen.
Figure 1 I confirms this fact. The increase in .4Ern,, standard deviation with surface
roughness is much higher than the increase in Al?,,,,,. This could be the result of !l I

8oo _ ___ ;_ _ _-T.. .^ ._-,__ “__._T


.._.~_. ,

Time (8)

0 ILL _ -__A. -.. .___A .-...- ~____ ~_... ~- . 1.. _,. .--_..i_-.. _.~ !

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4


Tixne (8)

Fig. 10. Sample of AE,,, values (experiment 3j: (a) fresh tool; (b) worn tool (cutting time 16
min).

(4 ,I., ,$“” 4 tb) t:,<‘j”

Fig. 11. AE, standard deviation vs. R, (frequency band 50-500 Mlz).
405

the rubbing between the worn (and therefore irregular and rough) flank face and the
workpiece and between the rake face and the chip or (2) the change in the cutting
edge curvature affecting the shear angle and influencing chip formation. Each of these
factors could cause some instantaneous process instability generating the spikes seen
in Fig. 10. Nevertheless, more experiments should be done to explain the source of
this scattering.
Because the growth in AE,,, with surface roughness was caused mainly by the
increase in scatter, it is not correct to say that the mean AE,,, of the data filtered
in a big bandpass is a good parameter to monitor the surface roughness increase in
finish turning. The standard deviation values, however, showed a reasonable increase
with surface roughness and, very importantly, were only slightly influenced by cutting
speed and feed. This leads to the conclusion that this parameter could be very useful
in the monitoring of surface roughness. The skew, kurtosis and count rate of the AE,,
values did not show a meaningful tendency when compared with R,.

5.4. Root-mean-square ana&si.s (200-300 kHz band data)


Analysis of the r.m.s. values was carried out in this band in an attempt to remove
or minimize the effects of the insensitivity of AE,, to surface roughness when the
analysis was done in a wider frequency range. As can be seen in Fig. 12, in this
frequency band the negative influence of crater wear on AE,, values is not seen and
the increase in AE,, with surface roughness was much higher than the values obtained
in a larger frequency band. The influence of feed and cutting speed on AE,, continues
to be the same, i.e. the feed influence is negligible but the,4&,, increases when the
cutting speed increases. The scatter of AE,, is still high in this frequency band and
still increases with R,, but the increase in the standard deviation of AE,,, with R,
presents some dispersed points, as can be seen in Fig. 13, and is only slightly influenced
by feed and cutting speed. Thus AE,,, in this band shows promise in roughness
monitoring.
Once again, the skew, kurtosis and count rate did not show a meaningful tendency
in these data.

Fig. 12. AI&,, w. R, (frequency band 2OQ-300 kHz).


* v

. ’

Fig. 13. AE,,,standard deviation ‘VS.I?, (frequency band 200-300 kHz).

6.Conclusions

From the discussion above it can be seen that parameters such as the zero crossing
rate of the raw signal, the standard deviation of the A&,,, signal in the frequency
band from 50 to 500 kHz, AE,, and the standard deviation of AE,, in the frequency
band from 200 to 300 kHz showed good potential for use in the monitoring of surface
roughness growth in finish turning. In this work all these parameters were calculated
by software, since the purpose was, beginning with raw signal analysis, to find the
best AE parameters to monitor surface roughness, but in real time tasks on the shop
floor it is necessary to calculate all these parameters in hardware to reduce computing
time.
Measuring AE,,,is a well-known task and can be accomplished by using a simple
r,m.s. voltmeter. The measurement of zero crossing rate is done using an electronic
counter similar to that used to measure count rate. The standard deviation measurement
can be done by hardware through the utilization of two r.m.s. voltmeters connected
in series, with a highpass filter between them to get rid of the d.c. term of the AE
signal extracted from the first voltmeter.
From the above discussion it can be concluded that, in finish turning ol- 1045
steel with carbide tools:
(1) the more meaningful frequency band for ail cutting conditions tried was around
200-300 kHz;
(2) the increase in tool wear and consequently in surface roughness only slightly
increases the A&,,,signal at a high band frequency for all cutting conditions tried;
(3) the increase in tool wear and surface roughness generates an increase in the
scatter of AE,,,values (and consequently in standard deviation values) for all cutting
conditions tried;
(4) parameters such as the zero crossing rate of the raw signal, the standard
deviation of AE,,, filtered in the SO-500 kHz frequency band, A&,, and the standard
deviation of AE,, values in the 200-300 kHz frequency band showed good potential
for the monitoring of surface roughness growth accompanying tool wear in finish
turning;
(5) it is possible to measure ah the parameters cited above by hardware using
the set-up suggested in this study.
407

References

I T. Blum and I. Inasaki, A study on acoustic emission from the orthogonal cutting process,
Trans. ASME, .I. Eng. Znd., 11.2 (1990) 203-210.
2 E. Kannatey-Asibu and D. Dornfeld, Quantitative relationships for acoustic emission from
orthogonal metal cutting, Trans. ASME, J. Eng. Ind., 103 (1981) 33&340.
3 K. Uehara and Y. Kanda, Identification of chip formation through acoustic emission mea-
surements, Ann. CZRP, 33(l) (1984) 71-74.
4 C. L. Jiaa and D. A. Dornfeld, Experimentaf studies of sliding friction and wear via acoustic
emission signal analysis, Wear, 139 (1990) 403424.
5 E. N. Diei and D. A. Dornfeld, Acoustic emission sensing of tool wear in face milling, Trans.
ASME, J. Eng. Znd., 109 (1987) 234-240.
6 K. Iwata, T. Moriwaki and N. Takenaka, An application of acoustic emission measurement
to in-process sensing of tool wear, Ann. CZZ?P, 25(l) (1977) 21-26.
7 E. Emel and E. J. Kannatey-Asibu, Tool failure monitoring in turning by pattern recognition
analysis of AE signals, Trans. ASME, J. Eng. Znd., 110 (1988) 137-145.
8 E. N. Diei and D. A. Dornfeld, A model of tool fracture generated acoustic emission during
machining, Trans. ASME, .I. Eng. Znd., 109 (1987) 227-233.
9 D. Dornfefd and H. G. Cai, An investigation of grinding and wheel loading using acoustic
emission, Trans. ASME, J. Eng. Ind., 106 (1984) 28-33.
10 I. Inasaki, Monitoring of dressing and grinding processes with acoustic emission signals. Ann.
CIZ?P, 34(l) (1985) 277-280.
11 T. Blum and D. A. Dornfeld, Grinding process feedback using acoustic emission, Pruc. #fh
Znt Grinding Conf, 1990, SME, Detroit, MI, 1990, pp. 525/l-525/20.
12 J. R. Klaiber, D. A. Dornfeld and J. J. Liu, Acoustic emission feedback for diamond turning,
Proc. 18th No& American Manufac~~ng Research Co& 1990, SME, Detroit, MI, 1990, pp.
113-119.
13 A. E. Diniz, A rugosidade superficial em processes de torneamento em acabamento: fatores
de influencia e criteria de fim de vida da ferramento, Ph.D. Thesis, State University of
Campinas, 1989.
14 M. C. Shaw, MetaZ Cutting Principks, Oxford University Press, New York, 1984.
15 N. H. Cook, Tool wear and tool life, Trans. ASME, J. Eng. Ind., 93 (1973) 931-938.
16 W. K. Luk and R. F. Scrutton, The origin of the groove wear in the turning operation, ht.
.I. Prod. Rex, 6(3) (1968) 197-206.
17 R. M. Sundaram and B. K. Lambert, Surface roughness variability of AISI 4140 steel in
fine turning using carbide tools, Znf. J. Prod. Res., 17(3) (1979) 249-258.
18 P. G. Petropoulos, Statistical basis for surface roughness assessment in oblique finish turning
of steel components, Znt. J. Prod. Res., Z2(3) (1974) 345-360.
19 V. C. Venkatesh and M. Satchitanandam, A discussion of tool life criteria and total failure
causes, Ann. CZRP, 29 (1980) 19-22.
20 E. J. Kannatey-Asibu and D. A. Dornfeld, A study of tool wear using statistical analysis of
metal cutting acoustic emission, Wear, 76 (1982) 247-261.
21 M. Lan and Y. Naerheim, Application of acoustic emission monitoring in machining, Proc.
13th North American Manufacturing Research Conj, 1985, SME, Detroit, MI 1985, pp. 3X0-313.
22 Y. Naerheim and M. Lan, Acoustic emission reveals information about the metal cutting
process and tool wear, Proc. Z6th North American Manufacturing Research Conf, 1988, SME,
Detroit, MI, 1988, pp. 240-244.
23 R. Teti and D. A. Domfeld, Modeling and experimental analysis of acoustic emission from
metal cutting, Trans. ASME, _I. Eng. Znd., ZZZ (1989) 229-237.
24 A. G. Beattie, Acoustic emission, principles and instrumentation, .Z.Acoust. Emiss., 2 (172)
(1983) 95-128.
25 J. J. Liu and D. A. Domfeld, Monitoring of micromachining process using acoustic emission,
Tech. Rep. ESRC-91-12, 1991 (Engineering System Research Center, University of California
at Berkeley).

Vous aimerez peut-être aussi