A Review of Sensor System and Application in Milling Process for Tool Condition Monitoring

This study presents a review of the state-of-the-art in sensor technologies and its application in milling process to measure machining signal for Tool Condition Monitoring (TCM) systems. Machining signals such as cutting force, torque, vibration, acoustic emission, current/power, sound and temperature from milling operation are briefly reviewed with the goal of indentifying the parameters for TCM. Sensors reviewed include both commercial and research devices that can measure machining signals. In this study describes trends in the sensor systems used and its potential for future research.


INTRODUCTION
The milling operation is a machining process that is very versatile, capable of producing a flat or curved surface, or complex geometric parts, by feeding the work piece against a rotating cutter containing a single or a number of cutting edges.Many phenomena affecting productivity, such as tool wear, chatter and even tool breakage naturally occur during milling operations.These phenomena if not monitored, degrade the quality of machined parts and cause dimensional errors.Under certain conditions, when tool wear or chatter has been excessive, they can cause machine downtime, as a result of catastrophic tool failure.Catastrophic tool failures in machine tools should be avoided, because they damage the machined parts and the machine tool itself and expose operating personnel to significantly hazardous situations (Cho et al., 2005).Downtime always has a significant effect on productivity and causes a rise in production costs.It has been reported that tool failure contributes up to 6.8% to the downtime of machining processes (Yeo et al., 2000).Certain types of downtime that are caused by excessive wear or chatter during the milling operation can be avoided by implementing a Tool Condition Monitoring (TCM) system.This system attempts to maintain the machined part quality and to avoid catastrophic tool failure by detecting wear status, chatter, tool breakage and even tool collision.Previous studies reported that implementation of a reliable TCM system increases the productivity by about 10-50% and increases savings by up to 40% (Najafi and Hakim, 1992;Lim, 1995;Rehorn et al., 2005).
Traditionally, the measuring techniques in TCM have been implemented in two categories of approach: direct and indirect.In the direct method, the actual geometric parameters of the cutting tool, for example, tool wear, are measured.Tool wear is measured by using a microscope, a CCD (Charge-Coupled Device) camera, radioactive isotopes, laser beams and electrical resistance (Kurada and Bradley, 1997;Wang et al., 2007).The direct method has become a conventional method that requires more time for dismantling and tool set up purposes and can only be used in laboratory techniques.This is due to the practical limitations caused by access problems during machining, illumination and the use of cutting fluids.However, direct measurement has a high degree of accuracy and has been employed extensively in research laboratories to support the investigations of fundamental measurable phenomena during the machining processes (Teti et al., 2010).
The indirect methods are achieved by empirically determined correlations or deducing suitable sensor signals to the actual quantity of tool wear.In this method, a variety of sensors are deployed to determine the machining signals, such as cutting force, torque, vibration, acoustic emission, motor current or power, temperature, sound and so on (Byrne et al., 1995;Sick, 2002).The advantages are that it has a less complicated setup and is suitable for practical applications but it is less accurate than the direct method.In contrast to the conventional method of tool-wear measurement, the approach is that the machining processes are continuously monitored via sensing devices to quantify the process performance or provide information for process optimization (Teti et al., 2010).
The main objective in this study is to present an overview of sensor systems and their applications in TCM during milling operations.The remainder of this study comprises three sections.Section one introduce the sensor system in milling process.Second, focuses on sensor systems and their applications currently used in milling operations.Third, provides a critical review of sensor systems in the milling process and identifies the trend of future sensor systems used i milling operations.Finally, concludes.

MACHINING SIGNALS IN MILLING PROCESS
The milling process is the removal of metal by feeding the work piece past a rotating multi cutter as shown in Fig. 1.The action of the milling cutter is different from a drill or a turning tool.In turning and drilling, the tools are kept continu contact with the material to be cut, whereas milling is an intermittent process, where the thickness of the cut for each cutting edge varies during the cut.Besides, the cutting tool in the turning process is singular and is always stationary in the tool holder, whereas the cutting tool in the milling process can be singular or with a multi cutting edge.Therefore, the cutting force generated from the process is also different.The magnitude of the cutting force in the turning process is always constant since the edge of the cutting tool is continuously engaged with the work piece during cutting.But, the cutting force in the milling process is a periodic series of interrupted cuts.This depends on the number of tool edges, the geometry of the milli cutters and the operation itself (Altintas, 2000) important parameters measured in the milling process for monitoring application are torque, vibration, acoustic emission, sound, ultrasound, current, power and also temperature.

SENSOR SYSTEM AND APPLICATIONS
There are several sensors system have been developed and applied in milling process in the previous, both commercial and research device.The sensor system deployed based on machining signals measurement including cutting force, torque, v acoustic emission, current/power and temperature.

Cutting force:
The cutting force is one of the most sensitive indicators of machining setup and is suitable for practical applications but it is ccurate than the direct method.In contrast to the wear measurement, the approach is that the machining processes are continuously monitored via sensing devices to quantify the process performance or provide information for .The main objective in this study is to present an overview of sensor systems and their applications in TCM during milling operations.The remainder of this study comprises three sections.Section one introduce or system in milling process.Second, focuses on sensor systems and their applications currently used in milling operations.Third, provides a critical review of sensor systems in the milling process and identifies the trend of future sensor systems used in TCM during

MACHINING SIGNALS IN MILLING PROCESS
The milling process is the removal of metal by feeding the work piece past a rotating multi-toothed cutter as shown in Fig. 1.The action of the milling cutter is different from a drill or a turning tool.In turning and drilling, the tools are kept continuously in contact with the material to be cut, whereas milling is an intermittent process, where the thickness of the cut for each cutting edge varies during the cut.Besides, the cutting tool in the turning process is singular and is he tool holder, whereas the cutting tool in the milling process can be singular or with a multi cutting edge.Therefore, the cutting force generated from the process is also different.The magnitude of the cutting force in the turning process is edge of the cutting tool is continuously engaged with the work piece during cutting.But, the cutting force in the milling process is a periodic series of interrupted cuts.This depends on the number of tool edges, the geometry of the milling (Altintas, 2000).Another important parameters measured in the milling process for monitoring application are torque, vibration, acoustic emission, sound, ultrasound, current, power

AND APPLICATIONS
There are several sensors system have been developed and applied in milling process in the previous, both commercial and research device.The sensor system deployed based on machining signals measurement including cutting force, torque, vibration, acoustic emission, current/power and temperature.
The cutting force is one of the most machining performance,  (Kistler, 2012) including tool condition and chip formation and also because it optimizes the machining process and the design of proper machining tools.Recently, a commercial force transducer and a dynamometer for the milling process have been developed, both a table an rotating dynamometer.A number of these techniques have evolved mainly due to the efforts of the Kistler Company over the past 40 years (Kistler, 2012) work piece table dynamometers have been widely used in industry and research laboratories for cut measurement in face and end milling as shown in Fig. 2a.The principle of the dynamometer is pressure detection using piezoelectric materials that are used in dynamometer construction as the main element and are converted to a proportional electric charge in three directions as shown in Fig. 2b.The most well piezoelectric material is quartz crystal (SiO rings are employed, including two shear quartz pairs, for Fx (yellow) and (violet), a pressure-sensitive pair.The pressure sensitive quartzes are arranged in the middle so that they lie in the neutral axis under bending.All the and z channels respectively are electrically paralleled.This makes the measurement independen momentary force application point.including tool condition and chip formation and also because it optimizes the machining process and the design of proper machining tools.Recently, a commercial force transducer and a dynamometer for the milling process have been developed, both a table and number of these techniques have evolved mainly due to the efforts of the Kistler (Kistler, 2012).The work piece table dynamometers have been widely used in industry and research laboratories for cutting force end milling as shown in Fig. 2a.The principle of the dynamometer is pressure detection using piezoelectric materials that are used in dynamometer construction as the main element and are lectric charge in three The most well-known piezoelectric material is quartz crystal (SiO 2 ).Quartz rings are employed, including two shear-sensitive (yellow) and Fy (blue) and Fz sensitive pair.The pressuresensitive quartzes are arranged in the middle so that they lie in the neutral axis under bending.All the x, y channels respectively are electrically paralleled.This makes the measurement independent of the The table dynamometer based on piezoelectric material has been widely used for cutting force measurement in the milling process.sensors used in machining to detect changes in tool wear, tool breakage and tool chatter.For instance, and Lin (1996) used a piezoelectric dynamometer to measure cutting force signals during a face milling operation to estimate tool wear.Yan investigated indices of tool wear based on the milling force using the dynamometer.Ghani et al used this type of dynamometer to measure the cutting force in estimating tool wear.Cho et al dynamometer to detected tool breakage by recording the three-channel cutting force in the milling Kim et al. (2007) measured the cutting force signal using a piezoelectric dynamometer and compared it with the prediction model for detecting chatter in a ball end milling operation.The other authors who applied the piezoelectric dynamometer in toolinclude Choudhury andRath (2000) (2001), Susanto and Chen (2003), Choi Chen and Chen (2004), Bhattacharyya Chen and Li (2009), Girardin et al. (2010Girardin et al. ( ) (2010) ) and Čuš and Župerl (2011).On the other hand, it is also used for detecting tool breakage Huang et al., 1999;Huang and Chen, 2000) chatter (Lacerda and Lima, 2004;Pongsathornwiwat and Tangjitsitcharoen, 2010;Huang Huang et al., 2012a).
Likewise, a rotating dynamometer has also been developed by the Kistler Company (Stirnimann 2012) based on a piezoelectric sensor that is equipped with quartz crystals.It enables measurement of the torque and axial force on a rotating tool rpm.The data are transferred by induction from the rotating sensor to a spatially fixed receiver.Kuljanic and Sortino (2005)  , Choi et al. (2004), , Bhattacharyya et al. (2007a), ( 2010), Kious et al. .On the other hand, it is also used for detecting tool breakage (Ko et al., 1995;., 1999;Huang and Chen, 2000) and tool (Lacerda and Lima, 2004;Pongsathornwiwat et al., 2012b; Likewise, a rotating dynamometer has also been Besides measuring the cutting force in the milling process using the commercial dynamometers that exist in the market, it can also be measured based on elastic deformation of the materials by using a strain gauge.The resistance of the strain gauge changes deformation that occurs in a material under mechanical load.The amount that the resistance changes depends on how the gauge is deformed, the materials from which it is made and the design of the gauge and Beasley, 2000).A strain gauge is a low cost, simple sensor that can be mounted directly onto the surface of the tool holder to measure the cutting force in turning (Ghani et al., 2011).However, milling needs metal based components, such as a table dynamometer, on which to place the strain gauge.For instance, Korkut (2003) has successfully developed a three component analogue dynamometer based on a strain gauge sensor.This dynamometer consists of four elastic octagonal rings, made from AISI 1040 steel, Milling operation and its generated machining signal (Yaldız et al., 2007) that was clamped between the cutter and the spindle.The dynamometer signals were stored in a computer by using a data acquisition board consisting of the main , the cutting perpendicular force F cN , the M c .They investigated the characteristics of cutting forces in face milling and wear indicators by simple analysis of the feature parameters of the cutting force signals.In (2009) used the rotating er for chatter detection by measuring the , the feed force F f and the feed rotating dynamometer has several advantages over the fixed or table , such as the cutting forces can be independently of the size of work piece and measurement can be performed on (four or five axis milling) (Kaya Besides measuring the cutting force in the milling process using the commercial dynamometers that exist in the market, it can also be measured based on elastic deformation of the materials by using a strain gauge.The resistance of the strain gauge changes due to the deformation that occurs in a material under mechanical load.The amount that the resistance changes depends on how the gauge is deformed, the materials from which it is made and the design of the gauge (Figliola ge is a low cost, simple sensor that can be mounted directly onto the surface of the tool holder to measure the cutting force in turning .However, milling needs metalbased components, such as a table dynamometer, on strain gauge.For instance, Korkut has successfully developed a three-force component analogue dynamometer based on a strain gauge sensor.This dynamometer consists of four elastic ngs, made from AISI 1040 steel, on which strain gauges were mounted, clamped between upper and lower plates, forming a platform, as shown in Fig. 3.The precision of this dynamometer was 5 N, with dynamic response bandwidth 192.2 Hz and maximum loading 1,500 N.There were differences from the dynamometer designed by Yaldiz et al. (2007).They developed a table dynamometer to measure three perpendicular cutting force components and torque.The system combined the strain gauge and piezoelectric accelerometer to measure static and dynamic cutting forces.Saglam and Unuvar (2003) used a strain gauge-based dynamometer for TCM in face milling.They correlate between the cutting parameters and the cutting force components with flank wear and surface roughness and the cutting force data were taken as a reference in order for on-line observation.
A spindle-integrated force sensor is one of the rotating cutting force transducers in the milling process that has been developed by Scheer et al. (1999) and Park (2004).The force transducers that in the form of a force ring sensor integrated to the spindle housing.The machine spindle is retrofitted with a flange force ring, which can be placed between the spindle flange and the spindle suspension as shown in Fig. 4a.This flat ring contains several piezoelectric force sensors as shown in Fig. 4b, which measure the process forces in the x, y and z directions and the torque in the axial direction.The data is transmitted from the rotating part of the sensor to a stator via telemetry.Byrne and O'Donnell (2007) combined the spindle-integrated force sensor for the drilling process with two piezoelectric force sensing rings, a flange and a bearing sensor ring.The flange sensor ring consisted of a precision machined steel housing in which a combination of compressive and shear force sensors were installed.The flange sensor ring was located behind the spindle housing, measuring forces in three directions.The construction of the bearing sensor ring consisted of a precision machined steel housing in which 22 piezoelectric elements were incorporated with a single output signal representing axial force.The bearing sensor ring was located behind the front thrust bearing measuring the force components in the feed direction.As shown in Fig. 5, the sensors were arranged in diagonally opposite pairs in the flange sensor ring in order to account for thermal instability.
However, integration of the force ring into the spindle machine structure raises some concerns (Byrne et al., 1995).These include the possibility of weakening the spindle dynamics, increased drift due to temperature changes inside the spindle, higher cross-talk between X, Y and Z channels due to structural deformation and the effect of spindle-internal forces on the measured signal.However, Jun et al. (2002) evaluated the effect of integration of the force ring on the dynamics of the spindle, the drift behavior due to temperature change inside the spindle, the cross-talk between the channels and the existence and compensation of the spindle- internal forces.The results show that the dynamic stiffness of the spindle machine tool decreased by less than 10% after the installation of the force ring, the drift behavior also varied by less than 10% and the average value of the force ring cross-talk in any direction was less than 9%.So, they concluded that the piezoelectricbased force ring is a strong transducer to be used in process model-based monitoring and fault diagnosis techniques.
The cutting force measurement on multi cutting edges in face milling has also been successfully developed.Adolfsson and Stahl (1995) have been pioneering the building of equipment for measuring cutting force components at each cutting edge.They developed a rotating dynamometer, based on strain gauges, measuring the main cutting force and feed cutting force components at each cutting edge.Strain gauges were mounted on a modified standard face mill, with a diameter of 125 mm and originally eight cutting edges.Every second cutting edge position was removed, giving space for the mounting of strain gauges, as well as weakening the tool in order for the strain gauges to be subjected to a larger strain.The signals from the strain gauges were processed in a serial stream and sent via a fibre optic cable to an LED (Light Emitting Diode).Then, the signals were transmitted by Infrared (IR) light signals over an air gap in the rear of the hollow spindle, at a speed of 1.28 Mbps.This equipment has been used for experimental studies of cutting force variation in face milling (Andersson et al., 2011).In contrast, Totis et al. (2010) used three direction (3D) piezoelectric for measuring triaxial cutting force components.Each cutting edge is provided with an integrated triaxial force sensor.Kistler 9251A, clamped between the modular cartridge and the cutter body by means of the preloading screw.The signals transmission is via a telemetry system with a 12 bit resolution and a sampling rate of 13 kHz.The rotating pulse test result shows that the frequency bandwidth is greater than 1.5 kHz.
The third method of measuring the cutting force in the milling process is the measurement of the elastic deflection of the spindle machine tool.Deflection is the displacement of a structural element from its static position as a result of force acting on the body.Matsubara et al. (2000) deployed two displacement sensors near a set of front bearings in the spindle unit so that they could detect a small scale cutting force of the spindle in the X and Y directions.A thin collar with a fine cylindrical surface was attached to the spindle as a sensor target.An electrically conductive sensor, based on the Eddy current principle with a dynamic response of about 1.3 kHz was used.Sarhan et al. (2004) studied the performance of this sensor compared to the table dynamometer and indicated an error between the displacement sensor and the dynamometer of about 20-50 N.They found that the relationship between the spindle displacement sensor output and the cutting force signals from the dynamometer showed a nonlinear characteristic, for which the cutting force can be estimated with a 90% confidence interval of ±30.9 N in the X-axis direction and ±35.4 N in the Y-axis direction.As well as force acting on the body, displacement also can occur in structures due to temperature changes.Therefore, a displacement sensor can be potentially compensated for by controlling the temperature during the machining process (Sarhan et al., 2006).The inductance sensor effect of Villari has also been proposed by Aoyama and Ishii (2004) to detect cutting force elements, cutting torque and tool deflections.In addition, tool deflection due to the cutting force in the milling operation can also be measured by using a thin film PVDF piezoelectric strain sensor and a wireless data transmitting system (Ma et al., 2010).
On the other hand, the use of capacitance sensors to detect the spindle shaft displacement due to the cutting load was investigated by Albrecht et al. (2005).A capacitive displacement sensor was integrated into the spindle to measure the static and dynamic variations of the gap between the sensor head and the rotating spindle shaft under load.The bandwidth of the uncompensated indirect force sensor system is limited to 350 Hz.A Kalman filter-based scheme was used to compensate for the spindle dynamic effects (Park and Altintas, 2004).They indicated that the use of a Kalman filter, which quasi inverts a model of the dynamic compliance between the tool tip and the displacement sensor reading point, increased the bandwidth of the proposed sensor system from 350 to almost 1,000 Hz.Auchet et al. (2004) also measured the shaft displacement of the motor spindle using command voltages of magnetic bearing to indirectly determine the cutting forces.The bandwidth of the indirect force measurement using active magnetic bearings is about 4 kHz and the cutting forces calculated from the command voltages were found to be in good agreement with those of a platform dynamometer.However, as reported by Castillo-Castañeda (2003), the displacement of the milling cutter can also be measured using a fibre optic (photonic) sensor for on-line toolwear detection.
Another technique that can be used to measure the cutting force is indirect measurement using a current sensor.Lee et al. (1995) proposed the indirect cutting force sensing method using the current of the AC servo drive.Kim et al. (1999) presented the indirect cutting force measurement method in the milling processes by using current signals of the servo motors.Jeong and Cho (2002) also reported that the cutting force can be estimated by measuring the feed-motor current.Meanwhile, Spiewak (1995) used a three-component accelerometer as a sensor to develop an instrumented milling cutter for measuring dynamic cutting forces.The system converts accelerations from the rotating spindle into stationary coordinates.Torque: The measurement of torque is an important part of the milling operation due to the rotating cutting tool spindle.Traditionally, milling torque has been determined from the measurement of the rotational speed of the spindle motor and the power consumption, or from the measurement of the motor current (Lee et al., 1997).Spindle power related to rotational speed and torque may be written as: where, P = The spindle power (W) ω = The rotational speed (rad/s) T = The torque (Nm) The milling torque is directly proportional to the tangential cutting force.The instantaneous torque, T, is expressed as the sum of the products of the tool holder radius and tangential cutting force for each tooth engaged in the cut (Schmitz and Smith, 2009).Generally, torque is measured once with cutting force by using a commercial dynamometer for toolwear monitoring (Kuljanic and Sortino, 2005;Kaya et al., 2011).These sensors are based on piezoelectric elements as used by Dini and Tognazzi (2007).They measured cutting torque by using a commercial rotating torque dynamometer (Montronix Accu Torque TM sensor) mounted on a standard ISO40, DIN69871-A tool holder.The power was supplied to the sensing system by the induction effect and the signal was transmitted by wireless telemetry to the stator, properly placed near the spindle.The tool holder was modified by reducing the external diameter from 48 to 26 mm for a total length of 24 mm, so the natural frequencies of the modified tool holder were decreased from 1867 to 1028 Hz.A piezoelectric composite element was also used by Tansel et al. (2011).They designed and developed mechanical hardware to measure torque using a piezoelectric composite element that was held between two extensions separately attached to the lower and upper halves.The system is called a Torque-based Machining Monitor (TbMM) and they used the system for chatter detection and wear estimation from the torque signals (Tansel et al., 2012).
Another technique to measure the torque in the rotating spindle uses a strain gauge.Smith et al. (1998) developed a milling torque dynamometer based on a strain gauge sensor.They installed the strain gauge between the tool and the tool holder on a conventional milling tool.The sensor provided virtually distortionless torque measurements over a bandwidth from DC to 2 kHz for a 100 mm diameter face milling cutter with sampling frequency of the A/D set at 10 kHz.The signals were transmitted by LEDs or the optical telemetry technique with a bandwidth of 0.5 MHz.Suprock and Nichols (2009) also developed sensor-integrated tooling based on a strain gauge for measuring torque in an end milling cutting tool as shown in Fig. 6.A Bluetooth A2DP transmitter and an AD 623 instrumentation amplifier are integrated into an ABS plastic shroud mounted on the exterior of a C 40 set screw-type tool holder.The tool holder was modified to house the female mini DIN connector and route the signal cables to the amplifier circuit.The system contains an analogue band-pass filter from 10 to 20 kHz, with the observed response attenuates at approximately 17 kHz.Deyuan et al. (1995) have applied torque signals from a strain gauge in detecting tool breakage in the milling process.
Vibration: A wide variety of transducers exist for use in sensing vibration.Vibration measurements are also referred to as displacement, velocity or acceleration measurements.Accelerometers are sensing transducers that provide an output proportional to acceleration and vibration.The piezoelectric accelerometer is the most common type in the vibration sensing of machining operations.Its transducer has the highest frequency response and range for the dynamic events (Figliola and Beasley, 2000).The piezoelectric element is placed in such a way that when the unit is in vibration, a mass applies a force, proportional to the acceleration, to the piezoelectric element.The element acts as a spring, which has a stiffness k and connects the base of the accelerometer to the seismic masses.The frequency response of the sensor is determined by the resonant frequency of the sensor, which can generally be modelled as a simple single-degree-of-freedom system.Using this system, the resonant frequency (ω) of the sensor can be estimated by (Wilson, 2005): Application of an accelerometer in the milling operation for TCM has been attempted over a number of decades to detect tool wear, tool breakage, surface roughness and chatter.Mehta et al. (1983) investigated the effect of the interaction between tool wear and vibrations using a piezoelectric accelerometer in a vertical milling machine using a single-tooth milling cutter with a diameter of 125 mm.Chen and Chen (1999) applied an ICP (Integrated Circuit Piezoelectric) accelerometer to detect tool breakage in the milling operation.Lou and Chen (1999) used a PCB accelerometer to measure vibration signals in the end milling operation for surface roughness prediction using a neural fuzzy system.Kim et al. (2007) predicted the chatter in a ball-end milling operation using an accelerometer and dynamometer.The chatter can be detected from the calculated cutting forces and their frequency spectra.Other authors, including Atlas et al. (2000), Yesilyurt and Ozturk (2007), Sun et al. (2007), Orhan et al. (2007), Zhang and Chen (2008), Huang et al. (2008), Wright et al. (2008), Chuangwen et al. (2009), Kalvoda and Hwang (2010b) and Insperger et al. (2008), applied an accelerometer for TCM to detect tool wear and breakages in the milling process by placing the accelerometer in various locations, such as on the side surface of the work piece in the feed direction, on the fixture or vice jaws and even on the lower bearing of the spindle housing (Norman et al., 2007;Kalvoda and Hwang, 2010a).
Optical techniques like laser displacement sensors for vibration detection have also been used in the milling process for TCM (Ryabov et al., 1998).The main advantage of this method is that it is possible to perform remote non-contact measurements on a rotating milling tool very close to the cutters.Tatar and Gren (2008) proposed the measurement technique of milling cutter vibration using laser vibrometry.They used a Polytec PSV 300 scanning laser vibrometer, standing steadily on a support located at about 1,800 mm from the measurement point.The vibration signals were compared with the cutting force and spindle head vibrations and the result showed that vibration velocities or displacements of the tool can be obtained with high temporal resolution during the cutting load.This technique is suitable for monitoring vibration, chatter and tool wear in a rotating spindle (Nakagawa et al., 2008).

Acoustic emission:
Piezoelectric ceramics are most commonly used in AE transducers due to their high sensitivity and response, primarily for normal displacement at the surface of contact.The transducer can detect stress-wave motions that cause local dynamic material displacement and convert this displacement to an electrical signal.The frequency range in an AE transducer varies between 20 kHz and 1 MHz and the meaningful information of the AE signals in the machining are mostly above 100 kHz (Li, 2002 ).An AE transducer consists of several parts, including: • A piezoelectric ceramic element with electrodes on each face • One electrode connected to an electric ground, the other to a signal lead • A backing material behind the element designed to minimize reflections back to the element and to damp the signal around the resonance frequency • A case providing an integrated mechanical package that may also serve as a shield to minimize the electromagnetic interference (Muravin, 2012) AE transducers have been successfully applied to detect tool wear in the micro end-milling operation (Tansel et al., 1998).Its application to the milling operation has been less straightforward, because the milling operation is an interrupted cutting operation with the rotating cutting tool causing pulse shock loading during the entry and exit of each individual tooth to the work piece.The magnitude of these shock pulses is possibly equivalent to those generated by tool fracture during end milling (Li and Guan, 2004).However, AE signals have high frequency and low amplitude, resulting in some studies focusing on the AE signal transmission system.Inasaki (1998) and Govekar et al. (2000) reported that coolant can be used as a transmission path by locating the AE transducer on the coolant supply nozzle.Hutton and Hu (1999) developed a liquid-coupled sensor system mounted on the end of the spindle drive shaft to collect AE signals.Li and Yuan (1998) also developed a simple device to collect AE signals from the tool holder through a magnetofluid.But, for the detection of tool wear and breakage, some researchers have used AE transducers in the milling operation directly mounted on the machine vice (Pai and Rao, 2002;Mathew et al., 2008), on the machine table (Jakobsen et al., 2005), on the workpiece (Tansel et al., 1998;Cao et al., 2008) and even on main machine spindle or on the cutting tool holder (Xiqing and Chuangwen, 2009;Haber et al., 2004).
However, a fibre optic interferometer has been used to measure AE in the machining process by McBride et al. (1993).The system is based on a Michelson interferometer that generates antiphase outputs and incorporates some novel polarization features.The light from a laser diode is a 780 nm wavelength.The laser output is coupled into the input fibre.The guided beam is then amplitude divided at a directional coupler into the signal and reference arms of the interferometer.The signal arm terminates in a rugged optical probe containing a lens to focus the light onto the workpiece surface, with a 40 µm spot size at a working distance of 25 mm (Carolan et al., 1997a).Carolan et al. (1997aCarolan et al. ( , 1997b) ) reported the application of a fibre optic interferometer for the in-process measurement of AE during face milling to provide tool wear information with AE energy and frequency analysis.Wilkinson et al. (1999) even used two kinds of AE sensors, a fibre optic sensor and a piezoelectric transducer, for tool-wear prediction.
Sound and ultrasound: Sound generated from a machining process is a common result of friction between the tool and the work piece.But, if there has been tool wear or chatter, the normal sound will be disturbed.Therefore, sound pressure levels from the cutting zone can be a parameter in detecting tool wear, breakage or chatter.Generally, sound signals can be measured by an electric microphone and then analyzed using a spectrum analyzer.Applications of sound signals for tool wear monitoring in the milling process have been conducted by Ai et al. (2012).The milling sound signal is extracted using acoustic spectrum characteristic parameters and analyzed with a Linear Predictive Cepstrum Coefficient (LPCC) to correlate to tool wear.A sound signal has also been used by Kopac and Sali (2001) and Nuawi et al. (2009) in the turning process and by Rubio and Teti (2009) in monitoring the milling process.Weingaertner et al. (2006) also used audio signals to evaluate stability based on work piece surface roughness.Because sound waves can travel through the air, sound signals are often used both in turning and milling by putting the microphone as close as possible to the cutting zone.Audible sound signals have an accepted standard range of frequencies of 20 to 20,000 Hz.If the sound frequency is greater than 20 kHz, it is ultrasonic and is beyond the upper limit of human hearing.The use of ultrasonics has also been attempted by Abu-Zahra and Yu (2000), Nayfeh et al. (1995) and Nuawi et al. (2007) for tool wear monitoring in the turning process, but there has been no report in the milling process.Usually, application sound signals are combined with AE, cutting force, current, vibration (Ghosh et al., 2007;Abbas et al., 2011) and laser techniques (Mannan et al., 2000).
Motor current and power: Current and power from the spindle motor are needed to drive the tool holder in the milling process.Their signals can be detected by power sensors to measure spindle power or current consumption.When the tool is worn or broken, as compared to the normal cutting condition, the current consumption changes (Sevilla-Camacho et al., 2011;Shao et al., 2011).Generally, the current signal is measured by the Hall Effect and an inductive sensor.The current conductor passes through a magnetically permeable core that concentrates the conductor's magnetic field.The Hall effect device is mounted in the core at a right angle to the concentrated magnetic field.A constant current in one plane excites the Hall device.When the energized Hall device is exposed to a magnetic field from the core, it produces a voltage that can be measured and amplified into process-level signals such as 4-20 mA.The inductive sensor has a wire-wound core and a signal conditioner.The current conductor passes through the core that magnifies the conductor's magnetic field.The AC current constantly changes potential from positive to negative and back again, generally at the rate of 50 or 60 Hz.The expanding and collapsing magnetic field induces current in the windings.This secondary current is converted to a voltage and conditioned to output process-level signals.
The Hall effect sensor (PCB Mounting Hall Effect Current Transducer) has been used by Li and Guan (2004), Li (2001Li ( , 2002)), Li and Yuan (1998), Li et al. (2008) and Sevilla-Camacho et al. (2011) to collect the current signal to monitor tool breakage in end milling by using a threshold method.They compared the marked feature with the threshold value; if the marked feature was over the threshold value, an alarm signal was generated.Lee and Tarng (1999) and Lee et al. (1997;1999) also used a current-to-voltage sensor (LEM Module LA 50-P) that works based on the Hall effect to monitor tool failure in end-milling operations, while Bhattacharyya et al. (2007b) used it for continuous on-line estimation of tool wear during face milling.In addition, Jeong and Cho (2002) used the Hall effect sensor to estimate the cutting force on a milling machine.Axinte and Gindy (2004) assessed the spindle power for tool wear monitoring.They used a dynamometer to collect the feed force and the feed perpendicular force.Based on their relationship, cutting forces can be obtained at a certain feed motion angle.Then, based on the relationship between the cutting force and cutting speed, the cutting power can be estimated.Tseng and Chou (2002) reported that the workload fluctuates by 1.5% when the tool is worn and torn.They extracted the workload of the spindle motor from the CNC controller and then transmitted the data via the I/O card for further processing.Jesús et al. (2003) implemented the band-pass analogue filter to minimize the spurious signals and to enhance the cutting force signal from the driver current signal.Lee et al. (2007) also reported that acquiring the spindle current is enabled by regulating the cutting force.
Temperature: Temperature and heat generated at the tool-chip and the tool-work piece interface during machining operations have been recognized as major factors that influence tool performance and work piece geometry.High temperatures at the interface accelerate the tool wear mechanisms and promote plastic deformation on the machined surface.Therefore, the increasing temperature during the cutting process can be used as reference for TCM (Young, 1996).There are two basic types of temperature sensing: • Contact temperature sensing requiring the sensor to be in direct physical contact with the media or object being sensed.• Non-contact measurement which interprets the radiant energy of a heat source in the form of energy emitted in the IR portion of the electromagnetic spectrum (Wilson, 2005).
For instances of contact cutting temperature measurement, Wright et al. (2006) used Resistive Temperature Detectors (RTDs) installed on the rear of the end-mill inserts and data were transmitted using a wireless system.RTDs are temperature sensors that contain a resistor that changes resistance value as its temperature changes.RTD elements consist of a length of fine coiled wire wrapped around a ceramic or glass core.In contrast, thermocouples consist of two dissimilar metals, joined together at one end.When the junction of the two metals is cooled or heated a voltage is produced that can be correlated back to the temperature.Kim et al. (2001) used a K-type thermocouple to measure the work piece temperature implanted into the work piece with a 0.5 mm diameter hole.They studied the influence on cutting temperature according to the change in the cutting environment and compared the temperature and tool wear.Ng et al. (2000) also used an embedded thermocouple technique to measure the cutting temperature when milling Inconel 718 with ball-nose coated tools.Dewes et al. (1999) measured the tool/workpiece interface and chip temperatures during ball-end-milling of H13 hardened steel.The measurements were performed by using thermocouple and IR camera techniques.They also studied the effect of workpiece angle and tool wear on the interface temperature.The results show that a worn tool generates higher temperatures than a new tool.Coz et al. (2012) implanted a K-type thermocouple into the rotating end-mill tool holder.They used a wireless connection system with a transmitter integrated into the tool holder and a radio frequency antenna placed near the tool holder.
A non-contacting device that intercepts and measures thermal radiation is called an IR pyrometer.IR sensors measure the IR energy emitted from an object in the 4-20 micron wavelength and convert the reading to a voltage.A typical IR pyrometer uses an optical system to concentrate radiated energy onto a thermopile.The resulting voltage output is amplified and conditioned to provide a temperature reading (Wilson, 2005).A previous study by Ueda et al. (2001) used a two-color pyrometer with an optical fibre to measure the temperature of the flank face of a cutting tool in high-speed milling.The fibre was inserted into a fine hole drilled into the workpiece and accepted IR energy that was radiated from the flank face of the cutting tool when it passed above the hole.Then, Sato et al. (2007) and Ueda et al. (2008) developed an IR radiation pyrometer with two optical fibres connected by a fibre coupler and applied it to the measurement of the tool-chip interface temperature in end milling with a binderless CBN tool.The IR rays radiated from the tool-chip interface are accepted by one optical fibre, which is inserted into the tool and fixed into the main spindle of the machine tool.Then the IR energy is sent to the other fibre, which is fixed to the column of the machine tool and led to the two-color pyrometer.The cyclical temperature variation in the cutting tool during cutting and non-cutting periods in end milling using this measurement technique was also studied (Sato et al., 2011).However, the use of cutting temperature for TCM in the milling operation is still rare and, in addition, few studies relate to the turning process.

CONCLUSION
The applications of sensors in TCM are many and varied.The use of sensors in machining provides significant information about tool conditions, including tool wear, breakage and chatter, which influence the quality of the machined surface and the production time.In order to be applied in industry, the sensors should be low cost, small in size, robust, non-invasive and should not interfere with the working space.The value of this technology can be best realized when reliable sensors are deployed in milling operations with rotating tools using wireless data transmission systems.Implementation of innovative sensors and their integrations into systems in TCM will be a research challenge in the future to enhance machining systems and their operations.

RECOMMENDATIONS AND CHALLENGES
After describing sensor systems and their applications in the milling process, it is possible to determine the state and direction of sensor systems in TCM research and to suggest some directions that should be developed in future research.
As was found in the literature related to milling, currently available sensors are adequate to detect machining signals, such as force, torque, vibration, AE, sound, current and temperature.In order to be widely used by machining industries, the sensor systems in TCM must be low cost, not reduce the stiffness of the machine tools and not restrict the working space; the sensor must be as close as possible to the machining point, have a reliable application and have good dynamic characteristics.The problem lies in how to deploy the sensors to pick up the signals on a rotating spindle during the milling operation.Certainly, the sensors based on cables for data transmission cannot be deployed on a rotating tool holder.However, machining signals, such as the cutting force and audible sound, can be detected without placing the sensor on the rotating tool.The cutting force is a popular signal for TCM and many authors used a table dynamometer.However, commercial dynamometers are not suitable instruments for industrial use due to their high cost.As well as sound signals, they are sensitive to other events including coolant flow, noise on the workshop floor and parallel machining operations.Current and power signals are also considered to be robust, but these signals are greatly influenced by the viscous damping of the feed system and the friction of the mechanical system.The use of vibration and AE signals can also be directly detected by the piezoelectric transducers that are placed on the workpiece.However, the quality of the signal is not stable, because the distance between the cutting zone and the sensor is constantly changing due to the relative movement of the cutting spindle in the three axes during cutting.The cutting zone temperature can also be measured by a non-contacting device, like an IR pyrometer, but it can be costly.Alternatively, many researchers use thermocouples to measure the cutting zone temperature by implanting them into the workpiece, so making the temperature reading not completely illegible during the cutting.
Therefore, a wireless data transmission system is essential since the sensor has to be mounted on a rotating tool holder.Recently, two standard technologies for Wireless Sensor Networks (WSN) have been developed: ZigBee and Bluetooth.Both operate within the Industrial Scientific and Medical (ISM) band of 2.4 GHz (Callaway, 2004).Wireless data acquisitions in TCM have been used in previous studies.Suprock and Nichols (2009) and Suprock et al. (2008) developed a wireless system using Bluetooth for transmitting machining signals in an end-milling tool holder.They used an A2DP transmitter as a strain gauge sensor that measured torque and a Jabra BT 350 transmitter as an accelerator to measure vibration in the tool tip.The system is capable of sending a stereo audio signal to the Bluetooth receiver at 48.0 kHz.Wright et al. (2006) used MEMS (Micro-Electro-Mechanical Systems) for tool temperature measurement.They used a move that is integrated between signal conditioning and radio units and RTDs as a sensor.Coz et al. (2012) also measured temperature by using the embedded thermocouple in the tool, thus data is sent to an amplifier using a transmitter integrated in the tool holder and a Radio Frequency (RF) antenna placed near the tool holder.Ma et al. (2010) applied a thin film PVDF piezoelectric strain sensor to measure deflection in the tool holder and wireless data as the transmitting platform built upon the IEEE 802.15.4 standard.Totis et al. (2010) used a wireless telemetry system for transmitting a force form load cell transducer in the face milling tool holder.However, the number of publications in TCM research that used wireless data transmission as compared to publications using a cable system and a commercial dynamometer are very few.Moreover, integrated multiple sensors in one wireless system with a tool holder has yet to be reported, whereas sensor fusion ensures a more precise and accurate measurement.More specifically, if multisensors are employed, the overall uncertainty of the resulting measurement can be reduced and thus, the resolution and precision of the system can be improved.
The system provides a better estimation of accuracy over a wide range of operating conditions.The combined sensory information also provides more accurate, more complete, more robust and more reliable than a single sensor.Therefore, applications of wireless data transmission and sensor fusion for TCM systems in the milling process will provide good research opportunities in the future.

(
Stirnimann et al., based on a piezoelectric sensor that is equipped with quartz crystals.It enables measurement of the torque and axial force on a rotating tool up to 25,000 rpm.The data are transferred by induction from the rotating sensor to a spatially fixed receiver.Kuljanic deployed a rotating dynamometer that was clamped between the cutter and the spindle.The dynamometer signals were stored in a computer by using a data acquisition board consisting of the main cutting force F c , the cutting perpendicular force axial cutting force F a and torque M the characteristics of cutting forces in face milling and determined the tool-wear indicators by simple analysis of the feature parameters of the cutting force signals.In another study, Kuljanic et al. (2009) dynamometer for chatter detection by measuring the axial force F a , the feed force perpendicular force F fN .The rotating several advantages over the fixed or table dynamometers, such as the cutting forces can be measured on the rotating tool independently of the size of work piece and measurement can be performed on any spatial position (four or five axis milling) et al., 2011).
Fig. 4: Construction of a spindle-integrated forces sensor, (a) position of the flange ring, (b) flange force ring with piezoelectric sensors(Scheer et al., 1999)

Fig. 6 :
Fig.6: Construction of spindle for torque measurement based on a strain gauge(Suprock and Nichols, 2009)