Date post: | 10-Oct-2015 |
Category: |
Documents |
Upload: | vamc-mamidi |
View: | 15 times |
Download: | 0 times |
of 43
Accepted ManuscriptReviewMonitoring And Processing Signal Applied In Machining Processes A ReviewC.H. Lauro, L.C. Brando, D. Baldo, R.A. Reis, J.P. DavimPII: S0263-2241(14)00354-6DOI: http://dx.doi.org/10.1016/j.measurement.2014.08.035Reference: MEASUR 2962To appear in: MeasurementReceived Date: 13 May 2014Revised Date: 17 July 2014Accepted Date: 14 August 2014
Please cite this article as: C.H. Lauro, L.C. Brando, D. Baldo, R.A. Reis, J.P. Davim, Monitoring And ProcessingSignal Applied In Machining Processes A Review, Measurement (2014), doi: http://dx.doi.org/10.1016/j.measurement.2014.08.035
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customerswe are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, andreview of the resulting proof before it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
MONITORING AND PROCESSING SIGNAL APPLIED IN MACHINING PROCESSES A Review
C.H. Lauro1*; L.C. Brando2; D. Baldo2; R.A. Reis2; J.P. Davim1
1 - Department of Mechanical Engineering, University of Aveiro, Campus Universitrio de
Santiago, 3810-193 Aveiro, Portugal.
2 - Department of Mechanical Engineering, Federal University of So Joo del Rei (UFSJ), Praa Frei Orlando n 170, Centro, 36.307-352, So Joo Del Rei, Brazil.
* - Corresponding author. E-mail address: [email protected]
Abstract: In machining processes several phenomena occur during material cutting. These phenomena can affect the production through the reduction of quality or accuracy, or by increasing costs (tools, materials, time). Thus, an understanding of machining phenomena is needed not only to define the cutting parameters for maximizing production, but also to ensure worker safety. An easy way to identify these phenomena is by monitoring machining processes, such as the measurement of cutting force, temperature and vibration. The acquired signal can have information about tool life, quality of cutting and defects in the workpiece. This review paper discusses the first steps involved in choosing and defining various techniques that may be used to monitor machining processes. Furthermore, this paper also outlines the techniques to acquire and process the signals of the monitoring processes. Hence, the objective of this paper is to help the reader understand the procedures for monitoring machining processes, and define methods, parameters, targets and other factors involved in doing so.
Key-works: Monitoring; Signal processing; Machining.
Nomenclature
AE Acoustic Emission
C/N Coulomb per Newton
CNC Computer Numerical Control
CQF Conjugate Quadratic Filters CWD Choi-Williams distribution CWT Continuous Wavelet Transforms DCT Discrete Cosine Transform
DFT Discrete Fourier Transform
DWT Discrete Wavelet Transform
Fc Cutting force Ff Feed force FFT Fast Fourier Transform
Fr Radial force FRF Frequency Response Function
FT Fourier Transform
HHT Hilbert-Huang Transform
HMM Hidden Markov model
HT Hilbert Transform IMF Intrinsic Mode Function MQL Minimum Quantity Lubrication PCA Principal Component Analysis SEA Synchronous Envelope Analysis STFT Short-time Fourier Transform SVD Singular value decomposition
TCM Tool Condition Monitoring WPTs Wavelet Packet Transforms
WT Wavelet Transform
WTMM Wavelet Transform Modulus
Maxima
ZAMD Zhao-Atlas-Marks distribution
1. INTRODUCTION
Machining processes are widely used to manufacture components that require great accuracy
and/or high surface quality finish. Furthermore, machining is a process that can provide low
costs to a certain number of parts. However, to obtain these advantages it is necessary to
ensure that the set up (machine, tool, cooling, etc) is in perfect order. An option for
controlling this is to use signal monitoring of various parameters including force, noise,
temperature and vibrations.
The monitoring of machining processes can represent economy and practicality due to it
helping to identify tool wear, surface roughness, and anomalies during metal cutting that can
cause waste, damage, and other impairing factors in this process. According to Dimla [1], the
nature of tool wear monitoring is complex and diverse and should provide an indication of
when the cutting tool should be changed without varying the workpiece surface finish, the
machine integrity, and the manufactured component tolerances.
Tool Condition Monitoring (TCM) is needed to obtain not only higher productivity and better
product quality, but also to identify the risks of severe damage to workpieces or machine-tool
components. This is because the operator reaction time has become insufficient during an
emergency, and the use of high speeds can cause serious damages [2].
Machining monitoring applied to hard or brittle materials can also meet the increasing
demands in precision and quality because it provides the characterization, control, and
improvement of processes. Valuable information about the manufacturing process can supply
the dual purpose of process control and quality monitoring, and it can be the first step
towards an automated manufacturing environment [3].
Lee et al. [4] mentioned in their article that the on-line monitoring of tool wear and the
prediction of its failure is reliable and able to respond quickly to tool failure. Even though the
personal computer used in their experiments (a CPU of 10 MHz 80286) did not offer a good
response time, the on-line real-time monitoring (based on cutting forces, sound and vibration,
laser scanners, vision systems, and computer tomography), was still able to supply surface
quality feedback to CNC machines focused on the on-line adjustment of cutting parameters.
However there are some drawbacks, such as signals that may be redundant and measurement
errors that are not easy to be avoided resulting in inaccurate prediction together with a
relatively high measurement cost, despite being able to supply surface quality feedback to
CNC machines focused on the online adjustment of cutting parameters [5].
This paper discusses the usage of monitoring in machining processes. In Section 2 a brief
review of the monitoring techniques is presented. Section 3 presents some parameters that
improve signal acquisition in machining monitoring. Section 4 presents signal processing
methods commonly used in machining monitoring. Section 5 gives the main conclusions.
2. MONITORING TECHNIQUES
The monitoring of machining processes can represent economy and practicality due to it
helping to identify tool wear, surface roughness, and anomalies during metal cutting that can
cause waste, damage and other impairing factors. According to Teti et al. [6], the measuring
techniques for the monitoring of machining operations have traditionally been categorised
into two approaches:
Direct measurement: where the actual value of the variable being measured gives a
high degree of accuracy. This method has been employed extensively in research
laboratories (due to the practical limitations of access during machining, illumination
and the use of cutting fluid) to support the investigations of fundamental measurable
phenomena during machining processes.
Indirect measurement: where the actual value is subsequently deduced using
empirically determined correlations. It is less accurate than the direct method but is
also less complex and more suitable for practical applications.
The indirect measurement applied to the monitoring of tool condition uses an estimate from
the measurable signal feature that is extracted through signal processing steps, as can be seen
in Fig. (1). It is used for a sensitive and robust representation of its corresponding state.
Examples of this measurement are the cutting forces, vibrations, acoustic emission, and
motor/feed current [7].
Figure 1. The framework of TCM [7].
Current control techniques are based on post-process measurement applied to finished
products, which leads to extensive quality control inspection times and also manufacture of
defective products with increase of production costs. System optimization involves three
basic elements: the correct choice of sensors for recording the signal in the monitoring
system, accurate signal processing and characterization, and reliable predictive models with
minor/low prediction of errors [8].
2.2. Cutting Forces
The analysis and prediction of cutting forces is very important in the research of metal cutting
processes and the design of cutting tools. These researches can develop a crucial role in
thermal analysis, tool wear estimation, chatter prediction, chip form categorization, surface
roughness prediction, monitoring of tool condition and others. Furthermore, a large cutting
force means more energy consumption. This has led to the study of the reduction of cutting
force through appropriate choices of parameters and tools [9]. According to Kim & Kim [10],
the cutting force comprises:
Static force: this is an average value and focuses mostly on measuring the cutting
force.
Dynamic force: this is the superimposed fluctuation and can satisfy the needs of
higher machining accuracy, because it has very useful information about the cutting
mechanism.
Gok et al. [11] investigated optimum cutting parameter values of the cutting force in the
milling of AISI H13 (50-54 HRC) on convex and concave inclined surfaces using a ball nose
tool. They observed that an absolute difference in percentile of measured and calculated
values was lower than 3.57 in both inclined surface types.
Turchetta [12] used the measurement of cutting force components in the tangential and
normal directions to analyse the influence of different conditions of tool wear in the milling
of Coreno Perlato Royal marble using a diamond tool.
According to Childs et al. [13], the cutting forces can be measured in:
Direct measurements: these are used when the forces need to be accurately known
both in magnitude and direction. This technique involves mounting a tool or the set of
tool/workpiece on a dynamometer, which responds to the forces by using electrical
signals proportional to the applied forces.
Indirect measurements: these are less accurate than direct methods, but can be
sufficient for monitoring purposes that involve deductions from the machine tool
behaviour.
Aggarwal et al. [14] monitored the electrical current and torque of a spindle motor from the
machine controller in the milling of Certal (AlZnMgCu 0.5) using a dynamometer, in order to
develop a holistic model that allowed the correct calculation of total spindle power to be
defined. According to Kim et al. [15], the cutting force can be measured indirectly by using
the current signal of the servo motor in the range of frequencies below 70 Hz.
The measurement of motor current is commonly used in indirect methods due to the input
phase current waveform variation when feeding during cutting. Furthermore, the phase
current signal is a perfect sine wave in the case of feeding without cutting. In this method, the
bandwidth of the Hall Effect current sensor should be higher than the frequency of the
applied cutting force. They used a sensor of bandwidth of 1 kHz to eliminate the high-
frequency noise and the feed motor current should be measured through an analogue filter
having a 50 Hz cut-off frequency [16].
Li et al. [17] studied the milling of AISI 1045 with HSS tools monitoring the three-phase
currents of a feed motor using Hall Effect sensors and a low-pass filter with a cut-off
frequency of 50Hz. They observed that the permutation entropy associated with tool breakage
can be revealed by information hidden in the motor current signals. This method can
effectively work for different cutting conditions, including the entry/exit cut, the variation of
cutting parameters, and the beat in the feed-motor current signals.
According to Lee et al. [18], the spindle current signal components are as follows:
Cutting force from the machining conditions;
Additional force from the tool wear;
Force variation by non-homogeneity of the workpiece;
Electrical noise that is eliminated by using low-pass filtering.
In the monitoring of spindle current signals the linear independence of the signal components
and a hybrid approach to cutting force regulation should be considered. This procedure is
applied to successively remove the influence of each parameter not related to tool wear from
the measured signals in order to isolate the tool wear data. This technique allows the
monitoring of gradual tool wear, being indispensable for automated and unmanned CNC
machining [18].
The usage of a dynamometer is the most popular method for the measurement of cutting
forces. The torque/force transducer can be constructed using piezoelectric or classical strain
gauges. The piezoelectric effect relates to charge separation that occurs in certain materials
when subjected to mechanical force. Each force component is detected by a separate crystal
oriented relative to the force in its piezoelectric sensitive direction. Quartz is usually chosen
as the piezoelectric material because of its good dynamic mechanical properties, low loss and
piezoelectric constant of approximately 2x10-12 C/N. A charge amplifier that must have a
high input impedance is therefore necessary to create a useful output. Commercial machining
dynamometers are available with natural frequencies from 2 to 5 kHz, depending on size
[13].
Yaldz and nsaar [19] developed an analogue dynamometer consisting of strain gauges in a
Wheatstone bridge circuit configuration. They were mounted on four elastic octagonal rings
to measure the three-force components, feed force (Ff), radial force (Fr), and main cutting
force (Fc) in the turning of AISI 4140. They found the output error percentages to be -0.8%
(Ff), -0.16% (Fr) and -0.12% (Fc) with a cross-sensitivity error in the range of 0.170.92%. A
similar construction was used by Yaldz et al. [20] for milling AISI 4140 steel. The static
calibration curves showed a very high linearity, with errors of 1.3% (Ff), 1.4% (Fc) and 1.2%
(Fr) and a range of 0.61.7%. for cross-sensitivity errors. The advantage of the strain gage-
based dynamometer compared to piezoelectric methods is the cost, the former being around
20 times less expensive.
In the study of a combined-type tool dynamometer used to measure the static and dynamic
cutting forces in an ultra-precision lathe, Kim and Kim [10] used a high-pass filter to
eliminate the 60 Hz electrical component in the dynamic component measurement in the
strain gauge dynamometer.
To understand the cutting force induced errors in CNC turning, Topal and oun [21] used a
strain gauge dynamometer that was designed and constructed using a full Wheatstone bridge
configuration in order to give efficient temperature compensation. They successfully
developed an empirical model for the estimation of diametric error which, by using their
method, could be reduced by approximately 90%.
Panzera et al. [22] developed a dynamometer to independently measure deflection caused by
the three components of the turning force. They used a circular hollow bar of AISI 4340 steel
as an elastic element that was designed to withstand a maximum force of 1.5 kN and a
maximum torque of 7.5 Nm. The experiments were carried out for the turning of normalized
medium carbon steel with a maximum feed rate of 0.4 mm/rot and depth of 2 mm. Strain
gauges were connected in a full Wheatstone bridge configuration, with a voltage of 2V
applied. The output voltage had to be amplified 1,000 times using an operational amplifier
with passive filtering (low pass) and a sampling rate of 120 Hz.
2.3. Vibration
Vibration is a common phenomenon in the finishing machining of a flexible workpiece due to
its low rigidity [23]. The great industrial interest is to avoid the vibrations that produce bad
surface finish and may cause damage to the machining components [24].
According to Dimla [1], vibration signatures satisfy the conditions of robustness, reliability,
and applicability requiring fewer peripheral instruments than acoustic emission. In addition,
their signals have the necessary quick response time needed to indicate changes for on-line
monitoring. He used an analytical on-line system for TCM based on vibration signature
features in the three principal axes to correlate the tool wear and observed that it is possible to
distinguish different wear modes from an analysis of the trends in the vibration signals.
Furthermore, they confirmed that the time domain features were deemed to be more sensitive
to cutting condition than tool wear, whereas frequency based features correlated well with
tool wear.
Accelerometers are more commonly used to measure vibration. However, Devillez and
Dudzinski [24] mentioned that despite being very easy to use, accelerometers present a major
drawback in that the acceleration signal can only be used to analyse the vibration frequency
and amplitude. They proposed a method which used a non-contact displacement measuring
system based on the eddy current principle with a sensitivity of 30 V/mm, because their
interest was to directly measure tool movement in real time in order to correlate it with the
surface finish obtained. This method presented an efficient way to determine the dynamic
parameters of the tool system and to obtain the cutting tool displacement signals.
Torabi et al. [25] investigated the on-line TCM in the high speed milling of Inconel-718
using a 3-flute micro-grain tungsten carbide ball-nose tool. To acquire the vibration signal,
which is a preferred choice for signal processing regarding the lower price of the sensor and
its easy installation, they used a ceramic shear three axes accelerometer. Zhong et al. [26]
used the vibration during cutting to study the milling on AL 7050-T7451 aluminium alloy
using dry cutting and Minimum Quantity Lubrication (MQL), 150 ml/min and 300 ml/min.
They observed that the cutting fluid can speed up the attenuation of cutting vibration, and in
the finishing process it is appropriate to apply MQL, which can assure a high quality of
workpiece and reduce the usage of cutting fluid.
In the monitoring of the vibration in the ultraprecision face turning of Al 6061 aluminium
alloy, with accelerometers fixed on the spindle and tool holders, Meyer et al. [27] observed
that waviness errors caused by relative tool/workpiece vibration are a significant source of
inaccuracy. They also found that the surface finish lobes provide a systematic framework for
describing how broadband relative tool/workpiece vibrations manifest themselves on the
workpiece surface.
Salgado et al. [28] measured the cutting vibrations at a rate of 10 kHz on the turning of AISI
8620 using a triaxial accelerometer. They developed a method for determining the accuracy
of a surface roughness prediction system based on cutting vibrations.
According to Zeng et al. [23], some studies are focused on the vibration control of a flexible
workpiece, but the fixture has importance on the machining, since it has the ability to
suppress the excessive machining vibration on the workpiece and balance the cutting forces.
It presents the following advantages:
It can target the nature of the problem of machining vibration suppression on a
flexible workpiece because of its clear physical meanings;
Vibration reduction of a flexible workpiece can easily be achieved with an appropriate
fixture layout, and the capability of disturbance rejection of the workpiecefixture
cutter system can be improved using this method;
The location, the applied forces and the number of fixture elements can be
simultaneously optimized.
In their paper about the vibration signal on the milling, Bisu et al. [29] studied the
decomposition of vibration sources generally and directly related it to the degree of failure of
a mechanical component from a number of measures in various configurations achieved by
accelerometers. They highlighted that the vibration signals are the result of a mixture of
different sources that correspond to components of machines, making it difficult to identify
the state of damage to a particular component.
Lamraoui et al. [30] investigated the chatter and tool wear monitoring on the basis of the
stationary and cycle stationary tools in high speed milling. Cycle stationarity is a property
that characterizes stochastic processes whose statistical properties periodically vary with
respect to some generic variables. They used the cycle stationarity character of accelerometer
signals coming from those that were acquired simultaneously with encoder information and
the signals are resampled in the angular domain. It offered an indisputable advantage in
industrial rotating machining operations that opened up ways for using it for monitoring
machining.
Rao et al. [31] analysed the vibration signal in the turning of AISI1040 tube with length of 90
mm, outer diameter of 100 mm and inner diameter of 56 mm using a tool with nose radii of
0.8 and 0.4 mm under dry conditions. They used a Laser Doppler Vibrometer (LDV) to
obtain online data acquisition. Among the several observations, vibration amplitudes found
were to increase with the progression of tool wear.
2.4. Temperatures
The power consumed in cutting is converted into heat near the cutting edge of the cutting tool
and many of the economic and technical problems are caused directly or indirectly by this
heating action [32]. The two goals of temperature measurement in machining are, mainly, to
quantitatively measure the temperature distribution throughout the cutting region (commonly
over 700C) and the average temperature at the chip/tool point of contact [13].
According to Byrne [33], the temperature is fundamental to the process of chip removal and
perhaps it is the single most important factor influencing the efficiency of the process
showing specific characteristic factors:
The degree of plastic deformation;
The extent of tool wear;
The degree of diffusion and corrosion;
The fatigue properties;
Compositional changes in the workpiece material.
According to Sivasakthivel and Sudhakaran [34], measuring the cutting temperatures is
difficult because the temperature is a scalar field which varies throughout the system and
cannot be uniquely described by values at a point. Thermocouples are the most widely used
sensors. They can be embedded in the tool or workpiece to measure the temperature
accurately with less effort, besides being conductive, they also operate over a wide
temperature range, are rugged and inexpensive.
Several experimental methods, such as thermocouple and radiation techniques, can be
employed to measure the temperature and the prediction of heat distribution in the cutting
zone. However, due to a narrow shear band, chip obstacles and the nature of the contact
phenomena where the two bodies, tool and chip, are in continuous contact and moving with
respect to each other these measurements become extreme difficulty [35]. Table 1 shows
some researches that used these methods.
Table 1. Temperature Measurements Type
To improve the understanding of work done by a cutting tool in removing metal, OSullivan
and Cotterell [46] monitored the temperature in the turning of Al 6082-T6 aluminium alloy
tube. They used two thermocouples on the inside of the tube and an infrared thermal camera
placed 0.5 m from the workpiece on the opposite side to the cutting tool. This technique
allowed the authors to observe that the amount of heat energy flowing into the tool at the
cutting edge increased due to increase in tool wear. The net result of this was to increase the
amount of energy required to perform the cutting process.
Davoodi and Hosseinzadeh [35] used an infrared sensor to monitor the temperature in high
speed machining as it is suitable for dry conditions due to its high response rate. It has the
ability to provide temperature measurements based on the distance from the cutting zone and
it is not necessary to make a hole in the tool or workpiece to install the instruments.
Moreover, it can be used for all types of materials, but the sensor should be installed as close
as possible to the desired surface, because the distance between target surface and sensor is
very important and can affect the results. Finally, it is not possible to use liquid cooling and
the chip may come between the surface and the sensor thereby causing an error.
2.5. Sound
Takate et al. [49] maintained that the operational sound contains a lot of information. The
sounds of the machine are perceived by the operator who can use them to identify the
operation (table movements, tool changes, and machining with various cutting tools), failures
of the machine or occurrences of abnormal machining conditions. They observed in their
results that there is a high rate of recognition for the movement of the machine and various
machining operations including machining with a broken tool. The chip formation process
emits an acoustic energy at high frequencies that can be used to determine events such as
TCM. The audio signal, although at a frequency higher than that of the vibrations, can be
detected by a microphone and is useful for characterising machining dynamics [50].
Microphones are very suitable for chatter detection in milling, as their sensitivity to chatter
onset is comparable to that of expensive sensors such as plate dynamometers, displacement
probes and accelerometers. This method is affected by some limitations (directional
considerations, low-frequency response and environmental sensitivity) and for successful
application, the suppression of environmental noise is mandatory [51]. However, the
microphone is a low cost solution for detecting chatter [50].
According to Weller et al. [52], some analysis and data reduction techniques to determine the
internal condition of machinery by listening to its sound, could also be applied to metal
cutting operations. They developed experiments for the construction and operation of a
detector of tool wear that observed the development of cutting edge flank wear-land, the
increase of cutting forces, and signal amplitude for the increased sound level produced.
In their paper, Lu and Kannatey-Asibu [53] monitored the audible sound signals, force,
vibration and acoustic emission on the turning of AISI 8620 steel using different tool wear
states. They used a inch microphone (B&K 4165) mounted at a distance of 7 inches (177.8
mm) from the cutting zone, at the end of a fixture supported by a magnetic base on the turret.
They used a 12 bit data acquisition board, with a 100 kHz sampling rate. The sound
monitoring was developed as a means of enhancing process monitoring capabilities. The
results agree qualitatively with the frequency characteristics of experimental data obtained
with sharp and worn tools.
Tekner and Yelyurt [54] used the sound signal to assess machinability of AISI 304
stainless steel, analysing the flank wear, built up edge, radii of chip curl, surface roughness
and sound pressure. They measured and recorded pressure levels of sound in the machining
process using a microphone connected to a computer and observed that levels of pressure
during the cutting levels decreased in parallel with positive results occurring in chip removal.
They affirmed also that measuring the cutting sound pressure level is a suitable method for
developing an alarm system.
Weingaertner et al. [55] used a inch free field microphone (PCB 377A02) positioned in the
machine working area close to the workpiece to monitor the milling of Al 7075- T6
aluminium alloy, using a 12 mm diameter cemented carbide end-mill under dry conditions.
The microphone was chosen as a sensor to detect vibrations during the process due to its
suitable frequency bandwidth and its ability to detect vibration signals from the tool. They
observed that for roughing operations, the results of analytical and time-domain simulations
were practically the same and for finishing operations the results obtained using the two
methods showed small differences. Furthermore, the greatest differences occurred for
conditions close to resonance, which related to a low depth of cut limit in the cutting tests.
Salgado and Alonso [56] highlighted that sound signal analysis during a cutting process has
been used for a long time. In their study, they used Singular Spectrum Analysis (SSA) to
extract valuable information correlated with tool wear in the turning of AISI 1040 steel. They
chose a condenser microphone (type 40AE, G.R.A.S. Skelstedet 10B with preamplifier
PRE12H) to amplify and log their data using a sampling rate of 50 kHz over a period of 100
ms. Their conclusion was that the sound signal emitted during turning and the feed motor
current have acceptable cost-performance ratios for their industrial application in relation to
other methods proposed in the literature.
Samraj et al. [57] proposed an on-line measurement system, using Singular Value
Decomposition (SVD) of the emitted sound during the turning process, to estimate the flank
wear of a tool. They used a microphone of 0.25" diameter with a dynamic range up to 122 dB
being more suitable to record a frequency response range of 20Hz to 20 kHz at a noise
accuracy of 0.5dB. They found an increase in the SVD features as the tool flank wear
increased, i.e., the condition monitoring of tool flank wear by emitted sound was proven
possible and is a relatively simple process.
Lu and Wan [58] developed a method for tool wear monitoring in the micromilling of SK2
Steel using a microphone with bandwidth of up to 80 kHz (higher than the traditional
microphone with bandwidth of up to 20 kHz). The collected signal was transformed to the
frequency domain using Fast Fourier Transform (FFT) and applied to a Hidden Markov
model (HMM) to process the signal and determine the tool condition. This method showed
that a classification rate of 100% can be obtained by normalizing the sound signal before
conducting the features selection process.
A disadvantage of this method is that in the region between 0 and 2 kHz the influence of the
surroundings and of the noise from adjacent machines, motors, conveyors, or processes can
influence the signals [56]. Moreover, frequencies below 100 Hz cannot be measured easily
and the microphone tends to pick up high levels of background noise [50].
Rafezi et al. [59] investigated the drilling on the Al 7075 aluminium alloy using HSS without
coolant implementation. They observed that frequency components less than 10 kHz are
affected by tool wear and other regions of the frequency spectrum are similar for both sharp
and worn tools. They also commented that the environmental noises (machine tools, fans,
human voice, and others) should be eliminated. They calculated that the frequency spectrum
of the noise in the proximity of the machine under study only has frequency components of
less than 2 kHz.
2.6. Acoustic Emission
Acoustic Emission (AE) is defined as the class of phenomena whereby transient elastic waves
are generated by the rapid release of energy from localized sources within a material. It can
be found in the primary (due to chip formation); secondary (due to friction between cutting
tool and chip); and tertiary (due to friction between cutting tool flank and workpiece) cutting
zones [5]. The usage of AE sensors to monitor machining processes (turning, milling, and
grinding) is quite effective, and detecting malfunctions due to the sensor is very sensitive to
the process and more reliable. One of the ways to take full advantage of high sensitivity is the
fusion with other types of sensor such as, for example, the force sensor [60].
The AE is due to the dynamic deformation of materials accompanied by the emission of
elastic stress waves, which occur over a wide frequency range but typically from 100 kHz to
1 MHz [13]. The AE technique is considered one of the most accurate monitoring methods in
machining that some researchers used it to identify damage mechanisms, furthermore, it is
also considered to be one of the most powerful methods in the composite drilling process
[61].
The AE technique shows a great advantage over the conventional load cell, it has a relatively
superior signal-to-noise ratio and sensitivity at the ultraprecision scale, even at extremely low
depths of cut [3].
The chip formation process emits an acoustic energy at high frequencies that can be used to
determine events such as TCM. This lower frequency audio signal, although its frequency is
higher than the vibrations, can be detected by a microphone and is useful for characterising
machining dynamics. However, frequencies below 100 Hz cannot be measured easily and the
microphone tends to pick up high levels of background noise [50]. Marinescu and Axinte
[62] used the AE sensory measures efficiently for monitoring both tool malfunctions and
workpiece surface anomalies in milling of Inconel 718 and when compared with the more
traditional sensory approaches (force/acceleration) limits their applications to TCM.
Hase et al. [63] studied the correlation between the AE signals and the cutting phenomena in
the turning of AISI O1. They used a lead zirconate titanate piezoelectric ceramic sensor with
resonance frequency of 1 MHz and a frequency band of 50 kHz to 2 MHz. They applied a
500-kHz high-pass filter to eliminate noise and signals caused by phenomena such as
collisions and twining of chips that are not directly related to the cutting phenomenon. They
observed that the amplitude of the signal formed by serrate-type chips is larger than that for
flow-type chips due to the amplitude of AE waves caused by the non-uniform discontinuous
fracture being larger than that caused by uniform ductile fracture. Furthermore, they also
observed a negative correlation between the mean value and the shear angle identifying the
process of formation of chips (the cutting state) that changes with the wear of the tool and
affects the quality of the machined surface.
To detect anomalous events in abrasive water jet machining, Axinte and Kong [64] employed
multiple point AE sensing (the nozzle, workpiece fixture, and a dummy plate) using a sample
rate of 1 MHz to collect the input, utilised and idle energy-related signals. They observed that
this technique is highly versatile without disturbing its setup while coping with the harshness
of the working environment characteristic of this process.
In the face turning studies of carbon/phenolic composite, Sreejith et al. [65] mounted an AE
sensor on the tool holder configured with a pre-amplifier (160 B Model; gain = 60 db) and
filters (30 kHz to 2 MHz, 125 kHz to 250 MHz and 500 kHz to 2 MHz). Through the
frequency domain, the authors observed that the cutting tool exhibits a mixed mode of
emission consisting of low frequency burst emission and high frequency deformation modes.
Furthermore, it is also influenced by the cutting speed, which indicates that the cutting tool
cuts stably up to 200 m/min above which, degradation of the tool sets in resulting in
deteriorated performance.
2.7. Other techniques
Young et al. [66] published a paper about the in-process and on-line surface texture
measurement using the Optical Surface Measurement Techniques (Specular Reflectance,
Diffuseness, Ellipsometry and Speckle). Galante et al. [67] proposed a technique to estimate
the Ra in the turning of AISI 1040 steel with a WC tool and cutting speed of 200 m/min.
using tool on-line monitoring employing a CCD camera. They observed that this technique
offers results similar to those detected by a profilometer. Thus, this technique could be
utilized in production to maintain the integrity within assigned production specifications by
varying the feed with increasing tool wear.
Morala-Argello et al. [68] performed a quality test for the surface roughness on turned parts
using a computer vision system. It showed advantages of having the possibility to perform in
machine measuring and the chance to carry out an exhaustive control of surface finishing.
The error values are affordable in an industrial environment and the measuring is less time
consuming and therefore more economical.
Another method of monitoring was used by Gok et al. [11]. They measured the tool
deflection using an inductive sensor mounted on the cutting tool at a length of 1.5 times the
tool diameter. The sensor operated in a voltage range of 0-10 V, the output voltage value was
set to remain constant at around 5 V.
Abu-Zahra and Lange [69] investigated the ultrasonic waves in the turning of AISI 4140 (55
HRC). They described the measures as mechanical waves that propagate at a frequency above
the human audible range, i.e. 20 kHz or higher They are determined by the physical and
mechanical properties of the transmittance media, such as, temperature, pressure, density,
stiffness, and acoustic impedance. They are also similar to light waves in terms of their
propagation mechanisms such as reflection, refraction, distortion, and absorption.
3. SIGNAL ACQUISITION
To obtain success in the monitoring process, the choice of devices is very important. The user
should check if the measurement resolution and range will satisfy the process requirements.
The resolution refers to the number of binary levels of ADC that can be used to represent a
signal. The smallest detectable change in this signal determines the resolution that is required
of the device, as can be seen in Fig. (2). The BPS (bits per sample) will be kept constant at its
optimum value because during the test period it is not possible to know if the acquired signal
is in the audible range, between 20Hz and 20kHz [70].
Figure 2. Influence of bits per sample [70].
The sample frequency is another important aspect because this factor, generally, is related to
the resolution of the signal. The sample frequency should be clearly defined, because an
incorrect value can mask important values that will influence the integrity of the signal. Shaw
[71] suggests that the sample frequency should be four times the natural frequency, a
minimum, in the use of a dynamometer. Some researchers applied the Nyquist Theorem of
sampling at twice the maximum frequency, whilst others suggest sampling at least 10 times
the maximum frequency, Fig. (3) [70].
Figure 3. Example of sample rate [70].
3.2. Domain Analysis
The time domain is based on estimating the signal period and subsequent equidistant
sampling of one signal period, an integer number of signal periods, or on using measurement
time covering many signal periods that are needed to decrease sufficiently the uncertainty of
measurement caused by the energy leakage due to non-coherent sampling [72].
The time domain signal when processed by the application of the Fourier Transform, see
forward, transforms the signal data into the frequency domain, which refers to the display or
analysis of data based on frequency. In vibration analysis, the principal advantage is the
repetitive nature of the signal and are clearly displaced as peaks in the frequency spectrum at
the frequency where the repetition takes place [73]. The use of frequency analysis can verify
the monitoring of the tool rotational speed or can detect if the tool is cutting with one or
multiple blades [74].
Analysing the accelerator signal, Bisu et al. [29] realized that the convolution in the time
domain is equal to the product in the frequency domain, so the fault signal becomes a
modulated signal. They also affirmed that the fault signal can be enlarged more than ten times
at the resonant frequency. However, the fault signal is relatively diminished in the other
frequency range due to the character of the wide frequency range of impulsive signal that the
average noise signal always contains in the low-frequency range and this situation also
applies to the sensor.
Analysing the microphone and accelerometer power spectrum of signals for sharp and worn
tools, Lu and Kannatey-Asibu [53] observed that the energy distribution for sharp and worn
tools are easily discernible from the sound and vibration signals. In this regard, ignoring the
sound signals below 0.5 kHz, similar peaks were observed in both the sound and vibration
signals in the feed and cutting directions.
According to Lamraoui et al. [30], the frequency domain has a drawback due to not providing
information in the time domain. A time-frequency analysis is better as it characterises the
signal in both the time and frequency domains. In vibration analysis, it manifests in either the
frequency or the time domain and thereby gives a compromise between the frequency
resolution and the temporal resolution. They acquired the data using the angular-domain
because it is more convenient to sample the signal with respect to an angular variable , so
that the cycle stationary characteristic is preserved.
Zhang and Chen [75] developed a study of TCM in an CNC end-milling machine based on
the vibration signal collected through a microcontroller-based data acquisition system. They
observed that displaying vibration signals of the X, Y and Z directions in the time domain is
helpful in understanding the cutting condition. The vibration amplitudes in the time domain
and the frequency peaks at harmonic frequency bands of the X and Y directions can be used
as the key featured signals for monitoring the tool condition.
Sometimes the vibration and sound analysis can be analysed using frequency spectra. More
advanced signal-processing techniques needed for system, source, and path identification
problems require the computation of frequency spectra. Furthermore, if the data are random
in character, a frequency analysis in terms of power quantities per hertz greatly facilitates the
desired evaluations of the data signals. For deterministic signals that are periodic, a frequency
decomposition or spectrum is directly obtained by computing the Fourier series coefficients
of the signal over at least one period of the signal using an Fast Fourier Transform (FFT)
algorithm. Line Spectrum or Discrete-frequency Spectrum is represented by the Fourier
component magnitudes versus frequency. However, the phase information is generally
retained only in those applications where there may be a need to reconstruct the signal time
history or determine peak values. The Power Spectral Density function or Auto Spectral
Density function provides a convenient and consistent measure of the frequency composition
of random data signals. The power spectrum is most easily visualized as the mean-square
value of the signal passed through a narrow-band pass filter divided by the filter bandwidth
[76].
It is interesting to highlight however, that although the time and frequency domain are more
utilized, other domain analysis can also be applied in machining monitoring. In their paper,
Ritou et al. [2] studied the radial eccentricity of a new end mill. To study the contribution of
each tooth to the cutting force, they applied an angular approach. They acquired the force
signals in the time domain associated with the tool angular position. They observed that for
every tool revolution, a force peak is extracted for each tooth that passes and hence the cutter
eccentricity is estimated.
Lamraoui et al. [77] developed a chatter indicator method to diagnose chatter in high speed
milling of Al 7075-T6 aluminium alloy using an angular-frequency domain. They affirmed
that analysis in the angular domain is useful for observing the behaviour of cutting forces
during each revolution and provides information about the system stability. They acquired the
AC motor integrated rotational encoder data using a system with an angular sampling device.
4. SIGNAL PROCESSING
Sometimes the acquired signal can be influenced by a frequency range that is not of interest
in the analysis, and so causes the monitoring be totally impractical. An alternative is to use a
non-periodic excitation and statistical signal processing technique that requires a number of
operator decisions: the frequency range, the number of test averages, and choice of
windowing procedure [50].
Kalvoda and Hwang [78] affirmed that selecting the right data processing technique is one of
the most important items for a cutting process that is assumed to be nonlinear and non-
stationary. Kuljanic et al. [51] mentioned some analysis techniques for the signal processing
Time domain analysis (once per revolution sampling, Poincar sections);
Frequency domain analysis (Fast Fourier Transform, power spectral density);
Time-frequency domain analysis (Wavelet Transform);
Other (entropy, coarse-grained entropy rate, normalized coarse-grained information
rate).
4.1. Fourier Transform
The Fourier Transform (FT) is commonly applied in signal processing. The principle of the
Fourier Transform is to extract the fundamental frequency component of the fringe pattern in
the 1D or 2D frequency domain and its inverse transform of the filtered frequency domain
signal which then provides the modulo 2pi phase of the fringe pattern [79].
To study the high-precision machining, Kono et al. [80] applied the Fourier series in the
frequency domain to analyse geometric errors from other errors using an artefact and a laser
displacement sensor. In their study of the analysis using a multi-sensor in high speed
machining, Kang et al. [81] monitored the spindle vibration and analysed the rotation
frequency and tooth frequency of the acceleration signal transformed by a Fourier Transform.
In the literature it is possible to find several variations of the Fourier Transform applied to
machining signal processing. The Discrete Fourier Transform (DFT) and Discrete Cosine
Transform (DCT) are efficient forms of the Fourier Transform often used in various
applications including tool condition monitoring [82].
The Gabor Transform, also called the short-time Fourier Transform (STFT), is a time-
frequency technique used to deal with non-stationary signals that have a short data window
centred on time. Its implementation for AE signal processing is efficient when it is used to
locate and characterise events with well-defined frequency patterns, not overlapping and long
relative to the window function [83].
According to Gu et al. [84], from among the several approaches available, the short-time fast
Fourier Transform (STFFT) is often used for non-stationary signal analysis, but it is a trade-
off between time and frequency resolutions. They proposed an approach based on the Choi-
Williams time-frequency distribution analysis and singular value decomposition that gives
satisfactory results and can be used for on-line condition monitoring and diagnosis of
machines.
Zhu et al. [7], affirm that although the Fast Fourier Transform (FFT) is the standard method
for observing signals in the frequency domain and has been widely studied, it has certain
serious theoretical drawbacks in processing machining signals. Liu et al. [85] used FFT to
avoid excessive vibrations noise in the machining signal and to filter the undesired frequency
components in the interpolation points, which guaranteed the exact trajectory generation and
shock free motion simultaneously.
4.2. Wavelet Transform
Wavelet Transform (WT) decomposes a single signal series in the time domain into a two-
dimensional function, where each of the decomposed signals is a mixture of source signals. It
can be considered as a series of band pass filters, whose results could be regarded as different
mixtures of independent source signals [86].
According to Zhu et al. [7], the Wavelet Transform was developed in the late 1980s to meet
the needs of adaptive time-frequency analysis in applied mathematics, physics, and
engineering. It has also been used for machinery fault diagnostics and TCM. Its great
potential in detecting abrupt changes of tool conditions can be explained by:
Sparse representation of signal, the wavelet expansion coefficients cj,k and dj,k drop
rapidly with increase in j and k, and only a few large coefficients exist while the
others are small.
Setting a suitable threshold, the undesired noise is filtered, the real essence of wavelet
denoising, and compression.
The localization of the time and frequency description of the signal that reveals the
signal behaviour in real time and its corresponding frequency property.
Liao et al. [82] highlighted the properties because the Wavelet Transforms are more powerful
and versatile than the Fourier Transform:
Some WTs have compact support, thus are able to capture local time-dependent
properties of data, whereas Fourier Transforms can only capture global properties.
WTs are more efficient even when compared with the FFT.
The WT is hierarchical and allows much fine tuning for a variety of applications.
Unlike the Fourier Transform, Wavelet Transform has an infinite set of possible basis
functions.
The Wavelet Transform has been applied to many engineering studies with great success and
can equally be applied for monitoring machining processes. It occurs due to the wavelet
exhibiting a natural shape, which is more descriptive of most natural processes than the sine
function used in Fourier analysis. Furthermore, the Wavelet Transform is capable of
revealing aspects of data that other signal analysis techniques miss, like trends, breakdown
points, discontinuities in higher derivatives, and self-similarity [87].
Different Wavelet features are used in tool condition monitoring. Continuous Wavelet
Transforms (CWT) are recognized as being effective tools for both stationary and non-
stationary signals, but they involve much redundant information and are computationally very
slow. The Discrete Wavelet Transform (DWT) has a fast algorithm based on Conjugate
Quadratic Filters (CQF) [7].
Grzesik and Brol [88] used the CWT (Mexican hat and Morlet) which is also capable of
detecting the fractal or partial fractal (multifractal) properties of the roughness profile on the
hard turning using optionally standard and wiper ceramic inserts. Kasashima et al. [89] used
the DWT to investigate the cutting forces in the milling of stainless steel 304.
Zhong et al. [26] adopted the Wavelet Packet Transforms (WPTs), which is often used to
analyse the frequency of a signal and provide a possibility of time-frequency localization, to
analyse vibration signals in order to show the effect of cutting fluid on different frequency
bands.
Wang and Liang [90] developed a non-dimensional chatter index based on the Wavelet
Transform Modulus Maxima (WTMM) and statistical analysis which included as advantages
the random and statistical nature of the metal cutting process. The sensitivity to chatters is
well known to be effective in detecting singularities; less susceptible to process changes; it
varies between 0 and 1 independent of cutting processes and hence can be used in different
machining processes.
Torabi et al. [25] applied CWT analysis of force and vibration in various scales to search for
the appropriate number of peaks according to cutting conditions, sampling rate and time
duration. They observed that wavelet features of force and vibration can be extracted for use
in a clustering method. It filters out the signal before feature extraction and makes the model
robust against the effects of noise.
Xu et al. [91] studied a TCM method using the vibration signal of the tool wear, which was
identified by pattern recognition technology, thus the conditions were identified. They
observed that calculating and classifying the speed of neural network and wavelet packet
analysis is very fast and a fault diagnosing system can be utilized to build up practical on-line
fault diagnosis, so it has broad prospects in many application areas. Table 2 shows some
researches that used WT to process monitoring signals.
Table 2. Some machining researches using Wavelet.
4.3. Hilbert and Hilbert-Huang Transform
The usage of the Hilbert Transform (HT), or its derivations, may be found in the literature for
machining research. The HT is one of the integral transforms like Laplace and Fourier, Eq.
(1). It was first introduced to solve a special case of the integral equations. Investigators of
digital algorithms for the realization of the HT made a major contribution when the digital
revolution started, and digital computers and digital signal procedures appeared everywhere.
In 1985, the HT was included as a typical signal procedure for the Brel and Kjr two
channel analyser. HT application to the initial signal provides some additional information
about an amplitude, instantaneous phase and frequency of vibrations. It also can be employed
for solving an inverse problem. In the frequency response function (FRF) of a linear structure
the original FRF is reproduced, and any departure from this, i.e. a distortion, can be attributed
to nonlinear effects [98].
[ ] ( )
== dtx
pi
1x(t)H(t)x Eq. (1)
Bisu et al. [29] used HT to develop a method based on synchronous envelope analysis (SEA),
which is a very useful tool for monitoring and also used to create special signals called
analytic signals. It is especially important in simulation and in this case a model was
developed with the purpose of using simulation to determine the teeth evolution in the
machining process. Furthermore, they affirmed that HT can be interpreted in the frequency
domain to highlight the different dynamic phenomena. This form of analysis is a very useful
tool for monitoring and creating special signals called analytic signals which are especially
important in simulation. They affirmed that the method to compute the HT of a function is a
frequency-domain approach, Eq. (2).
( )
= dty
pi
1(t)y Eq. (3.2)
Cao et al. [99] decomposed Acoustic Emission (AE) signals via a lifting scheme and
extracted features from the wavelet coefficients using HT to identify salient features of
different tool states (normal conditions, slight breakage and serious breakage) in the milling
of AISI 1045. They observed that tool breakage can be detected successfully through the
recognition of these features.
A derivation of the Hilbert Transform (HT), the Hilbert-Huang Transform (HHT), is also
used to process the signals. According to Cao et al. [100], the HHT consists of the following:
Empirical Mode Decomposition , when a complicated signal is decomposed into a
series of simple oscillatory modes, designated the intrinsic mode function (IMF), and
a residue;
The Hilbert Transform is then invoked for each intrinsic mode function (IMF) to
obtain the instantaneous frequencies and the instantaneous magnitudes, which
comprise the Hilbert-Huang spectrum of the signal.
Cao et al. [100] monitored the vibration signals from the milling of aluminium 7050 with a
carbide end mill cutter (two flutes) using a spindle speed of 8.5 krpm (1.5 m/min) and a
sampling frequency of 6.4 kHz. They used the HHT to analyse the reconstructed signals and
obtain the Hilbert-Huang spectrum, from which their mean value and standard deviation were
used to calculate the chatter indices.
Kalvoda and Hwang [78] used the HHT to analyse the cutting forces and vibration from the
milling of aluminium alloy using an end-mill with diameter of 12 mm and four flutes. They
used a sampling frequency of 9 kHz for both acquisition signals with a cutting speed of
204m/min (5.4 krpm). They used the HHT to correlate the tool wear/breakage by observing
the change in the frequency peak with the change in cutting geometry of a cutter tool. They
considered the shift of the main frequency peak into lower frequency together with higher
frequency fluctuations to be a cutter tool wear indicator. Although the HHT method is a new
method, it has shown good results and it has attracted other researchers to use it. Table 3
shows some researchers who have used this method.
Table 3. Some machining researches using HHT.
4.4. Other signal processing methods
Kim et al. [15] applied a real-time Kalman filter in the motor current signals in order to filter
out the undesired current component due to the acceleration or deceleration from the
measured servo motor current signal. A Kalman filter is a linear and minimum error variance
recursive algorithm for optimally estimating the unknown state of a dynamic system from a
noisy environment.
In their paper, Marinescu and Axinte [104] developed a monitoring method based on
combinations of time-frequency domain analysis of the AE by use of advanced processing
techniques, Choi-Williams distribution (CWD), Zhao-Atlas-Marks distribution (ZAMD) and
formant analysis methods. They mentioned that ZAMD produces good resolution in time and
frequency domains and reduces the interference obtained in the cross-terms present in multi-
component signals. It is useful in resolving small spectral peaks and capturing non-stationary
and multi-component signals.
To process the signal, Simeone et al. [105] used Principal Component Analysis (PCA), also
known as the Karhunen Loeve Transformation, the purpose of which is to reduce the high
dimensionality of sensor signal data, consisting of a large number of interrelated variables, by
extracting significant signal features. The usage of PCA consists of a signal pre-processing of
mean centring which needs to be performed by calculating the mean of each variable and
subtracting it from the original data to generate a zero-mean distribution. Signals oscillating
around zero are obtained to ensure that the first principal component describes the direction
of maximum variance. They observed a success rate that varied between 88% to 100%
confirming the capability of PCA to extract valuable sensory features for on-line residual
stress monitoring.
5. CONCLUSION
Monitoring applied to machining processes can improve the process through an increase in
tool life and surface quality with a simultaneous decrease of electric energy and waste
material. However, the choice of the monitoring method requires great care due to
implementation cost and requirements, besides the objectives to be analysed, i.e., in certain
cases, the tool wear monitoring using a dynamometer can be as efficient as the accelerometer,
which is more inexpensive. Furthermore, signal interpretation is fundamental. The user
should match the best method of analysis to the objective, i.e., if the process uses variable
revolution, the vibration or sound monitoring in the frequency domain can be unsuitable if
the user does not revise the revolution/frequency ratio. Although machining monitoring
requires great care, its usage brings excellent results for both industrial and academic
research.
6. ACKNOWLEDGEMENT
The authors would like to thank the Ministry of Educations Coordination for the
Improvement of Higher Education Personnel (CAPES), the University of Aveiro and the
Department of Mechanical Engineering of the Federal University of So Joo del Rei.
Additional thanks go to Dr. Juan Carlos Campos Rubio, from the Federal University of Minas
Gerais, Brazil.
7. REFERENCES
[1] D. E. Dimla, The Correlation of Vibration Signal Features to Cutting Tool Wear in a Metal Turning Operation, Int. J. Adv. Manuf. Technol., vol. 19, no. 10, pp. 705713, Jun. 2002.
[2] M. Ritou, S. Garnier, B. Furet, and J. Y. Hascoet, Angular approach combined to mechanical model for tool breakage detection by eddy current sensors, Mech. Syst. Signal Process., vol. 44, no. 12, pp. 211220, Feb. 2014.
[3] X. Han and T. Wu, Analysis of acoustic emission in precision and high-efficiency grinding technology, Int. J. Adv. Manuf. Technol., vol. 67, no. 912, pp. 19972006, Nov. 2012.
[4] K. S. Lee, L. C. Lee, and S. C. Teo, On-line tool-wear monitoring using a PC, J. Mater. Process. Technol., vol. 29, no. 13, pp. 313, Jan. 1992.
[5] C. Lu, Study on prediction of surface quality in machining process, J. Mater. Process. Technol., vol. 205, no. 13, pp. 439450, Aug. 2008.
[6] R. Teti, K. Jemielniak, G. O. Donnell, and D. Dornfeld, Advanced monitoring of machining operations, CIRP Ann. - Manuf. Technol., vol. 59, no. 2, pp. 717739, Jan. 2010.
[7] K. Zhu, Y. S. Wong, and G. S. Hong, Wavelet analysis of sensor signals for tool condition monitoring: A review and some new results, Int. J. Mach. Tools Manuf., vol. 49, no. 78, pp. 537553, Jun. 2009.
[8] E. Garca-Plaza, P. J. Nez, D. R. Salgado, I. Cambero, J. M. H. Olivenza, and J. G. Sanz-Calcedo, Surface Finish Monitoring in Taper Turning CNC Using Artificial Neural Network and Multiple Regression Methods, Procedia Eng., vol. 63, pp. 599607, Jan. 2013.
[9] W. J. Deng, Q. Li, B. L. Li, Y. T. He, W. Xia, and Y. Tang, Study on the cutting force of cylindrical turning with novel restricted contact tools, Int. J. Adv. Manuf. Technol., Jun. 2013.
[10] J. Kim and D. Kim, Development of a combined-type tool piezo-film accelerometer for an ultra-precision lathe, J. Mater. Process. Technol., vol. 71, no. 3, pp. 360366, 1997.
[11] A. Gok, C. Gologlu, and H. I. Demirci, Cutting parameter and tool path style effects on cutting force and tool deflection in machining of convex and concave inclined surfaces, Int. J. Adv. Manuf. Technol., no. 3, Jun. 2013.
[12] S. Turchetta, Cutting force and diamond tool wear in stone machining, Int. J. Adv. Manuf. Technol., vol. 61, no. 58, pp. 441448, Nov. 2011.
[13] T. H. C. Childs, K. Maekawa, T. Obikawa, and Y. Yamane, Metal Machining Theory and Applications. London: Arnold, 2000, p. 400.
[14] S. Aggarwal, N. Nei, and P. Xirouchakis, Cutting torque and tangential cutting force coefficient identification from spindle motor current, Int. J. Adv. Manuf. Technol., vol. 65, no. 14, pp. 8195, May 2012.
[15] T.-Y. Kim, J. Woo, D. Shin, and J. Kim, Indirect cutting force measurement in multi-axis simultaneous NC milling processes, Int. J. Mach. Tools Manuf., vol. 39, no. 11, pp. 17171731, Nov. 1999.
[16] G. D. Kim and C. N. Chu, Indirect Cutting Force Measurement Considering Frictional Behaviour in a Machining Centre Using Feed Motor Current, Int. J. Adv. Manuf. Technol., vol. 15, no. 7, pp. 478484, Jul. 1999.
[17] X. Li, G. Ouyang, and Z. Liang, Complexity measure of motor current signals for tool flute breakage detection in end milling, Int. J. Mach. Tools Manuf., vol. 48, no. 34, pp. 371379, Mar. 2008.
[18] K. J. Lee, T. M. Lee, and M. Y. Yang, Tool wear monitoring system for CNC end milling using a hybrid approach to cutting force regulation, Int. J. Adv. Manuf. Technol., vol. 32, no. 12, pp. 817, Mar. 2006.
[19] S. Yaldz and F. nsaar, A dynamometer design for measurement the cutting forces on turning, Measurement, vol. 39, no. 1, pp. 8089, Jan. 2006.
[20] S. Yaldz, F. nsaar, H. Salam, and H. Ik, Design, development and testing of a four-component milling dynamometer for the measurement of cutting force and torque, Mech. Syst. Signal Process., vol. 21, no. 3, pp. 14991511, Apr. 2007.
[21] E. S. Topal and C. oun, A cutting force induced error elimination method for turning operations, J. Mater. Process. Technol., vol. 170, no. 12, pp. 192203, Dec. 2005.
[22] T. H. Panzera, P. R. Souza, J. C. C. Rubio, A. M. Abro, and T. R. Mansur, Development of a three-component dynamometer to measure turning force, Int. J. Adv. Manuf. Technol., vol. 62, no. 912, pp. 913922, Dec. 2011.
[23] S. Zeng, X. Wan, W. Li, Z. Yin, and Y. Xiong, A novel approach to fixture design on suppressing machining vibration of flexible workpiece, Int. J. Mach. Tools Manuf., vol. 58, pp. 2943, Jul. 2012.
[24] A. Devillez and D. Dudzinski, Tool vibration detection with eddy current sensors in machining process and computation of stability lobes using fuzzy classifiers, Mech. Syst. Signal Process., vol. 21, no. 1, pp. 441456, Jan. 2007.
[25] A. J. Torabi, Er Meng Joo, Li Xiang, Lim Beng Siong, Zhai Lianyin, San Linn, Gan Oon Peen, and Ching Chuen Teck, Application of classical clustering methods for online tool condition monitoring in high speed milling processes, in 2012 7th IEEE Conference on Industrial Electronics and Applications (ICIEA), 2012, pp. 12491254.
[26] W. Zhong, D. Zhao, and X. Wang, A comparative study on dry milling and little quantity lubricant milling based on vibration signals, Int. J. Mach. Tools Manuf., vol. 50, no. 12, pp. 10571064, Dec. 2010.
[27] P. A. Meyer, S. C. Veldhuis, and M. A. Elbestawi, Predicting the effect of vibration on ultraprecision machining surface finish as described by surface finish lobes, Int. J. Mach. Tools Manuf., vol. 49, no. 15, pp. 11651174, Dec. 2009.
[28] D. R. Salgado, F. J. Alonso, I. Cambero, and a. Marcelo, In-process surface roughness prediction system using cutting vibrations in turning, Int. J. Adv. Manuf. Technol., vol. 43, no. 12, pp. 4051, Sep. 2008.
[29] C. F. Bisu, M. Zapciu, O. Cahuc, A. Grard, and M. Anica, Envelope dynamic analysis: a new approach for milling process monitoring, Int. J. Adv. Manuf. Technol., vol. 62, no. 58, pp. 471486, Dec. 2011.
[30] M. Lamraoui, M. Thomas, and M. El Badaoui, Cyclostationarity approach for monitoring chatter and tool wear in high speed milling, Mech. Syst. Signal Process., vol. 44, no. 1, pp. 177198, 2014.
[31] K. V. Rao, B. S. N. Murthy, and N. M. Rao, Cutting tool condition monitoring by analyzing surface roughness, work piece vibration and volume of metal removed for AISI 1040 steel in boring, Measurement, vol. 46, no. 10, pp. 40754084, 2013.
[32] E. Trent and P. Wright, Metal Cutting, 4th ed. Woburn: ButterworthHeinemann, 2000, p. 446.
[33] G. Byrne, Thermoelectric signal characteristics and average interfacial temperatures in the machining of metals under geometrically defined conditions, Int. J. Mach. Tools Manuf., vol. 27, no. 2, pp. 215224, 1987.
[34] P. S. Sivasakthivel and R. Sudhakaran, Optimization of machining parameters on temperature rise in end milling of Al 6063 using response surface methodology and genetic algorithm, Int. J. Adv. Manuf. Technol., vol. 67, no. 912, pp. 23132323, Dec. 2012.
[35] B. Davoodi and H. Hosseinzadeh, A new method for heat measurement during high speed machining, Measurement, vol. 45, no. 8, pp. 21352140, Oct. 2012.
[36] E. Bagci and B. Ozcelik, Analysis of temperature changes on the twist drill under different drilling conditions based on Taguchi method during dry drilling of Al 7075-T651, Int. J. Adv. Manuf. Technol., vol. 29, no. 78, pp. 629636, Jul. 2005.
[37] R. Li and A. J. Shih, Spiral point drill temperature and stress in high-throughput drilling of titanium, Int. J. Mach. Tools Manuf., vol. 47, no. 1213, pp. 20052017, Oct. 2007.
[38] L. C. Brando, R. T. Coelho, and C. H. Lauro, Contribution to dynamic characteristics of the cutting temperature in the drilling process considering one dimension heat flow, Appl. Therm. Eng., vol. 31, no. 1718, pp. 38063813, Dec. 2011.
[39] T. Ueda, H. Tanaka, A. Torii, and T. Matsuo, Measurement of Grinding Temperature of Active Grains Using Infrared Radiation Pyrometer with Optical Fiber, CIRP Ann. - Manuf. Technol., vol. 42, no. 1, pp. 405408, 1993.
[40] C. Wei, K. Xu, R. Li, and D. Hu, Temperature modeling in end grinding of coated workpieces, J. Shanghai Jiaotong Univ., vol. 15, no. 3, pp. 319322, May 2010.
[41] A.-M. O. Mohamed, A. Warkentin, and R. Bauer, Variable heat flux in numerical simulation of grinding temperatures, Int. J. Adv. Manuf. Technol., vol. 63, no. 58, pp. 549554, Feb. 2012.
[42] C. H. Lauro, L. C. Brando, and S. L. M. Ribeiro Filho, Monitoring the temperature of the milling process using infrared camera, Sci. Res. Essays, vol. 8, no. 23, pp. 11121120, 2013.
[43] L. C. Brando and R. T. Coelho, Temperature and heat flow when tapping of the hardened steel using different cooling systems, Rev. Chil. Ing., vol. 17, no. 2, pp. 267274, 2009.
[44] G. Fromentin, a. Bierla, C. Minfray, and G. Poulachon, An experimental study on the effects of lubrication in form tapping, Tribol. Int., vol. 43, no. 9, pp. 17261734, Sep. 2010.
[45] S. Bhowmick, M. J. Lukitsch, and a. T. Alpas, Tapping of AlSi alloys with diamond-like carbon coated tools and minimum quantity lubrication, J. Mater. Process. Technol., vol. 210, no. 15, pp. 21422153, Nov. 2010.
[46] D. OSullivan and M. Cotterell, Temperature measurement in single point turning, J. Mater. Process. Technol., vol. 118, no. 13, pp. 301308, Dec. 2001.
[47] F. M. Aneiro, R. T. Coelho, and L. C. Brando, Turning Hardened Steel Using Coated Carbide at High Cutting Speeds, J. Brazilian Soc. Mech. Sci. Eng., vol. XXX, no. 2, pp. 104109, 2008.
[48] M. Hadad and B. Sadeghi, Minimum quantity lubrication-MQL turning of AISI 4140 steel alloy, J. Clean. Prod., vol. 54, pp. 332343, Sep. 2013.
[49] S. Takate, J. H. Ahn, M. Miki, Y. Miyao, and T. Sata, A Sound Monitoring System for Fautt Detection of Machine and Machining States, in Annals of the ClRP, 1986, vol. 35, pp. 14.
[50] K. Cheng, Machining Dynamics - Fundamentals, Applications and Practices. London: Springer, 2009, p. 328.
[51] E. Kuljanic, G. Totis, and M. Sortino, Development of an intelligent multisensor chatter detection system in milling, Mech. Syst. Signal Process., vol. 23, no. 5, pp. 17041718, Jul. 2009.
[52] E. J. Weller, H. M. Schrier, and B. Weichbrodt, What sound can be expected from a worn tool?, J. Eng. Ind., vol. 91, no. 3, pp. 525534, 1969.
[53] M. C. Lu and E. Kannatey-Asibu, Analysis of Sound Signal Generation Due to Flank Wear in Turning, J. Manuf. Sci. Eng., vol. 124, no. 4, p. 799, 2002.
[54] Z. Tekner and S. Yelyurt, Investigation of the cutting parameters depending on process sound during turning of AISI 304 austenitic stainless steel, Mater. Des., vol. 25, no. 6, pp. 507513, Sep. 2004.
[55] W. L. Weingaertner, R. B. Schroeter, M. L. Polli, and J. de Oliveira Gomes, Evaluation of high-speed end-milling dynamic stability through audio signal measurements, J. Mater. Process. Technol., vol. 179, no. 13, pp. 133138, Oct. 2006.
[56] D. R. Salgado and F. J. Alonso, An approach based on current and sound signals for in-process tool wear monitoring, Int. J. Mach. Tools Manuf., vol. 47, no. 14, pp. 21402152, Nov. 2007.
[57] A. Samraj, S. Sayeed, J. E. Raja, J. Hossen, and A. Rahman, Dynamic Clustering Estimation of Tool Flank Wear in Turning Process using SVD Models of the Emitted
Sound Signals, in World Academy of Science, Engineering and Technology, 2011, vol. 56, pp. 11511155.
[58] M.-C. Lu and B.-S. Wan, Study of high-frequency sound signals for tool wear monitoring in micromilling, Int. J. Adv. Manuf. Technol., vol. 66, no. 912, pp. 17851792, Aug. 2012.
[59] H. Rafezi, J. Akbari, and M. Behzad, Tool Condition Monitoring based on sound and vibration analysis and wavelet packet decomposition, in 2012 8th International Symposium on Mechatronics and its Applications, 2012, pp. 14.
[60] I. Inasaki, Application of acoustic emission sensor for monitoring machining processes, Ultrasonics, vol. 36, no. 15, pp. 273281, Feb. 1998.
[61] N. Zarif Karimi, H. Heidary, G. Minak, and M. Ahmadi, Effect of the drilling process on the compression behavior of glass/epoxy laminates, Compos. Struct., vol. 98, pp. 5968, Apr. 2013.
[62] I. Marinescu and D. A. Axinte, A critical analysis of effectiveness of acoustic emission signals to detect tool and workpiece malfunctions in milling operations, Int. J. Mach. Tools Manuf., vol. 48, no. 10, pp. 11481160, Aug. 2008.
[63] A. Hase, M. Wada, T. Koga, and H. Mishina, The relationship between acoustic emission signals and cutting phenomena in turning process, Int. J. Adv. Manuf. Technol., vol. 70, no. 58, pp. 947955, Oct. 2013.
[64] D. A. Axinte and M. C. Kong, An integrated monitoring method to supervise waterjet machining, CIRP Ann. - Manuf. Technol., vol. 58, no. 1, pp. 303306, Jan. 2009.
[65] P. S. Sreejith, R. Krishnamurthy, and S. K. Malhotra, Effect of specific cutting pressure and temperature during machining of carbon/phenolic ablative composite using PCBN tools, J. Mater. Process. Technol., vol. 183, no. 1, pp. 8895, Mar. 2007.
[66] R. D. Young, T. V. Vorburger, and E. C. Teague, In-Process and On-Line Measurement of Surface Finish, CIRP Ann. - Manuf. Technol., vol. 29, no. 1, pp. 435440, Jan. 1980.
[67] G. Galante, M. Piacentini, and V. F. Ruisi, Surface roughness detection by tool image processing, Wear, vol. 148, no. 2, pp. 211220, Aug. 1991.
[68] P. Morala-Argello, J. Barreiro, and E. Alegre, A evaluation of surface roughness classes by computer vision using wavelet transform in the frequency domain, Int. J. Adv. Manuf. Technol., vol. 59, no. 14, pp. 213220, Jul. 2011.
[69] N. H. Abu-Zahra and J. H. Lange, Tool Chatter Monitoring in Turning Operations Using Wavelet Analysis of Ultrasound Waves, Int. J. Adv. Manuf. Technol., vol. 20, no. 4, pp. 248254, Aug. 2002.
[70] National Instruments, How to Choose the Right DAQ Hardware for Your Measurement System. pp. 15, 2012.
[71] M. C. Shaw, Metal Cutting Principles, 2a ed. USA: Oxford University Press., 2004, p. 672.
[72] M. Novotny and M. Sedlacek, Measurement of active power by time domain digital signal processing, Measurement, vol. 42, no. 8, pp. 11391152, Oct. 2009.
[73] A. R. Bhende, G. K. Awari, and S. P. Untawale, Assessment of bearing fault detection using vibration signal analysis, VSRD Tech. Non-Technical J., vol. 2, no. 5, pp. 249261, 2011.
[74] A. Morska, M. Matuszak, and P. Waszczuk, Zeszyty Naukowe Experimental sensor system implementation for selected micromilling-related parameters, Sci. Journals Marit. Univ. Szczecin, vol. 31, no. 103, pp. 134139, 2012.
[75] J. Z. Zhang and J. C. Chen, Tool condition monitoring in an end-milling operation based on the vibration signal collected through a microcontroller-based data acquisition system, Int. J. Adv. Manuf. Technol., vol. 39, no. 12, pp. 118128, Sep. 2007.
[76] I. L. Vr and L. L. Beranek, Noise and vibration control engineering: principles and applications. John Wiley & Sons, 2006, p. 966.
[77] M. Lamraoui, M. Thomas, M. El Badaoui, and F. Girardin, Indicators for monitoring chatter in milling based on instantaneous angular speeds, Mech. Syst. Signal Process., vol. 44, no. 12, pp. 7285, Feb. 2014.
[78] T. Kalvoda and Y.-R. Hwang, Analysis of signals for monitoring of nonlinear and non-stationary machining processes, Sensors Actuators A Phys., vol. 161, no. 12, pp. 3945, Jun. 2010.
[79] L. Huang, Q. Kemao, B. Pan, and A. K. Asundi, Comparison of Fourier transform, windowed Fourier transform, and wavelet transform methods for phase extraction from a single fringe pattern in fringe projection profilometry, Opt. Lasers Eng., vol. 48, no. 2, pp. 141148, Feb. 2010.
[80] D. Kono, A. Matsubara, I. Yamaji, and T. Fujita, High-precision machining by measurement and compensation of motion error, Int. J. Mach. Tools Manuf., vol. 48, no. 10, pp. 11031110, Aug. 2008.
[81] M. C. Kang, J. S. Kim, and J. H. Kim, A monitoring technique using a multi-sensor in high speed machining, J. Mater. Process. Technol., vol. 113, no. 13, pp. 331336, Jun. 2001.
[82] T. W. Liao, C.-F. Ting, J. Qu, and P. J. Blau, A wavelet-based methodology for grinding wheel condition monitoring, Int. J. Mach. Tools Manuf., vol. 47, no. 34, pp. 580592, Mar. 2007.
[83] E. M. Rubio, R. Teti, and I. L. Baciu, Advanced signal processing in acoustic emission monitoring systems for machining technology, in Intelligent Production Machines and Systems, 2006, pp. 16.
[84] S. Gu, J. Ni, and J. Yuan, Non-stationary signal analysis and transient machining process condition monitoring, Int. J. Mach. Tools Manuf., vol. 42, no. 1, pp. 4151, Jan. 2002.
[85] X. Liu, F. Ahmad, K. Yamazaki, and M. Mori, Adaptive interpolation scheme for NURBS curves with the integration of machining dynamics, Int. J. Mach. Tools Manuf., vol. 45, no. 45, pp. 433444, Apr. 2005.
[86] H. Shao, X. Shi, and L. Li, Power signal separation in milling process based on wavelet transform and independent component analysis, Int. J. Mach. Tools Manuf., vol. 51, no. 9, pp. 701710, Sep. 2011.
[87] C. Scheffer and P. S. Heyns, Wear monitoring in turning operations using vibration and strain measurements, Mech. Syst. Signal Process., vol. 15, no. 6, pp. 11851202, Nov. 2001.
[88] W. Grzesik and S. Brol, Wavelet and fractal approach to surface roughness characterization after finish turning of different workpiece materials, J. Mater. Process. Technol., vol. 209, no. 5, pp. 25222531, Mar. 2009.
[89] N. Kasashima, K. Mori, G. H. Ruiz, and N. Taniguchi, Online Failure Detection in Face Milling Using Discrete Wavelet Transform, CIRP Ann. - Manuf. Technol., vol. 44, no. 1, pp. 483487, Jan. 1995.
[90] L. Wang and M. Liang, Chatter detection based on probability distribution of wavelet modulus maxima, Robot. Comput. Integr. Manuf., vol. 25, no. 6, pp. 989998, Dec. 2009.
[91] C. Xu, Z. Liu, and W. Luo, A Frequency Band Energy Analysis of Vibration Signals for Tool Condition Monitoring, in 2009 International Conference on Measuring Technology and Mechatronics Automation, 2009, vol. 1, pp. 385388.
[92] Y. S. Tarng and B. Y. Lee, Amplitude demodulation of the induction motor current for the tool breakage detection in drilling operations, Robot. Comput. Integr. Manuf., vol. 15, no. 4, pp. 313318, Aug. 1999.
[93] K. Mori, N. Kasashima, J. C. Fu, and K. Muto, Prediction of small drill bit breakage by wavelet transforms and linear discriminant functions, Int. J. Mach. Tools Manuf., vol. 39, no. 9, pp. 14711484, Sep. 1999.
[94] a. Velayudham, R. Krishnamurthy, and T. Soundarapandian, Acoustic emission based drill condition monitoring during drilling of glass/phenolic polymeric composite using wavelet packet transform, Mater. Sci. Eng. A, vol. 412, no. 12, pp. 141145, Dec. 2005.
[95] Z. Yang and Z. Yu, Grinding wheel wear monitoring based on wavelet analysis and support vector machine, Int. J. Adv. Manuf. Technol., vol. 62, no. 14, pp. 107121, Dec. 2011.
[96] Y. Choi, R. Narayanaswami, and a. Chandra, Tool wear monitoring in ramp cuts in end milling using the wavelet transform, Int. J. Adv. Manuf. Technol., vol. 23, no. 56, pp. 419428, Mar. 2004.
[97] J. Bhaskaran, M. Murugan, N. Balashanmugam, and M. Chellamalai, Monitoring of hard turning using acoustic emission signal, J. Mech. Sci. Technol., vol. 26, no. 2, pp. 609615, Apr. 2012.
[98] M. Feldman, Hilbert transform in vibration analysis, Mech. Syst. Signal Process., vol. 25, no. 3, pp. 735802, Apr. 2011.
[99] H. Cao, X. Chen, Y. Zi, F. Ding, H. Chen, J. Tan, and Z. He, End milling tool breakage detection using lifting scheme and Mahalanobis distance, Int. J. Mach. Tools Manuf., vol. 48, no. 2, pp. 141151, Feb. 2008.
[100] H. Cao, Y. Lei, and Z. He, Chatter identification in end milling process using wavelet packets and HilbertHuang transform, Int. J. Mach. Tools Manuf., vol. 69, pp. 1119, Jun. 2013.
[101] T. Kalvoda and Y.-R. Hwang, A cutter tool monitoring in machining process using HilbertHuang transform, Int. J. Mach. Tools Manuf., vol. 50, no. 5, pp. 495501, May 2010.
[102] a. M. Bassiuny and X. Li, Flute breakage detection during end milling using HilbertHuang transform and smoothed nonlinear energy operator, Int. J. Mach. Tools Manuf., vol. 47, no. 6, pp. 10111020, May 2007.
[103] Z. Yang, Z. Yu, C. Xie, and Y. Huang, Application of HilbertHuang Transform to acoustic emission signal for burn feature extraction in surface grinding process, Measurement, vol. 47, pp. 1421, Jan. 2014.
[104] I. Marinescu and D. Axinte, A timefrequency acoustic emission-based monitoring technique to identify workpiece surface malfunctions in milling with multiple teeth cutting simultaneously, Int. J. Mach. Tools Manuf., vol. 49, no. 1, pp. 5365, Jan. 2009.
[105] A. Simeone, T. Segreto, and R. Teti, Residual Stress Condition Monitoring via Sensor Fusion in Turning of Inconel 718, Procedia CIRP, vol. 12, pp. 6772, Jan. 2013.
Figure 4. The framework of TCM [7].
Figure 5. Influence of bits per sample [70].
Figure 6. Example of sample rate [70].
Table 4. Temperature Measurements Type
Researcher Material Method Target
Dril
ling [36] Al 7075-T651 aluminium alloy Thermocouple Tool
[37] Ti-6Al-4V Thermocouple Tool [38] AISI H13 Thermocouple Workpiece
Grin
ding [39] AISI 1055 Annealed and Hardness Infrared Radiation Pyrometer Tool
[40] Steel with WC-Co coating Thermocouple Workpiece [41] AISI 4140 Infrared Thermal Camera Workpiece
Mill
ing [35] CuZn40Al12 Infrared Sensor Workpiece
[34] Al 6063 aluminium alloy Thermocouple Workpiece [42] Al 7050 aluminium alloy Infrared Thermal Camera Workpiece
Tapp
ing [43] AISI H13 Thermocouple Workpiece
[44] AISI 1070 Thermocouple Workpiece [45] Al 319 aluminium silicon alloy Infrared Thermometer Workpiece
Turn
ing [46] Al 6082-T6 aluminium alloy Thermocouples Workpiece Infrared Thermal Camera
[47] AISI 4340 Thermocouple Tool [48] AISI 4140 Thermocouples Tool
Table 5. Some machining researches using Wavelet.
Researcher Monitored Signal Material Spindle Speed (rpm) Sampling Frequency
Dril
ling [92] Electric Current S45C steel 1000, 1200 1 kHz
[93] Cutting Force Stainless steel 316 2000 1 kHz [94] Acoustic Emission Laminated composite 1762.95
Grin
ding
[82] Cutting Force
Ceramic 4500
4906 Hz Vibration 4906 Hz
Spindle Power 4906 Hz Acoustic Emission 1 MHz
[95] Acoustic Emission ASTM 1045 1 MHz
Mill
ing
[96] Cutting Force AISI 1018 1000, 2000 500 Hz [17] Electric Current AISI 1045 300, 450, 600, 900, 1200 1 kHz
[26] Vibration Al 7050-T7451 aluminium alloy 3000 - 18000 2 MHz
Turn
ing [69] Vibration AISI 4140 200 - 1600 100 MHz Ultrasound waves 10 MHz
[68] Image AISI 6150 [97] Acoustic Emission AISI-D3 171, 245, 318 8333 Hz
Table 6. Some machining researches using HHT.
Researcher Monitored Signal Material Spindle Speed (rpm) Sampling Frequency
Mill
ing [101] Vibration SAE 1045 773, 1345, 2004 2 kHz
[102] Current SAE 1045 300, 450, 600, 900, 1200 1 kHz
Gri
ndi ng
[103] Acoustic emission AISI 1045 24 2 MHz
Vibration 2 MHz Voltage, Current 10 kHz