vibration-based tool condition monitoring systems 2005 scheffer

26Vibration-Based Tool

Condition MonitoringSystems

C. SchefferUniversity of Pretoria

P.S. HeynsUniversity of Pretoria

26.1 Introduction .................................................................... 26-1149

26.2 Mechanics of Turning .................................................... 26-1150General Terms † Chatter Vibrations † Tool Wear

26.3 Vibration Signal Recording ........................................... 26-1155Direct and Indirect Systems † Sensor Requirements

for Tool Wear Monitoring † Force Measurement †

Acceleration Measurement † Acoustic Emission

Measurement † Sensor Comparisons

26.4 Signal Processing for Sensor-Based ToolCondition Monitoring ................................................... 26-1159Feature Extraction † Feature Selection

26.5 Wear Model/Decision-Making for Sensor-BasedTool Condition Monitoring .......................................... 26-1163Trending, Threshold † Neural Networks † Fuzzy

Logic † Other Methods

26.6 Conclusion ...................................................................... 26-1168

Summary

Despite the high level of technology built into every aspect of modern metal cutting operations, the phenomenon oftool wear still hampers the reliability and complete automation of machining processes. Tool wear is the loss ofmaterial on the edge of the cutting tool. This chapter concerns sensor-based tool condition monitoring (TCM), andspecifically those methods that are based on vibration related properties such as force, acceleration, and acousticemission (AE). References are made to systems proposed in the literature and also to commercially availablehardware. The chapter focuses on turning operations. The mechanics of turning are briefly discussed. Variousmethods of obtaining vibration signals from turning operations are described. The vibration signal has to beprocessed in order to estimate the level of wear in the cutting edge of the tool, and several state-of-the-art approachesare discussed. Effective methods of constructing a model relating sensor data and the tool wear, using processedvibration signals, are described. The chapter concludes by indicating some important points that should beconsidered when using vibration-based systems for TCM, and some interesting topics for future research in this fieldof study. Chapters 25 and Chapter 27 present further information of the present subject.

26.1 Introduction

Millions of products are manufactured daily by a variety of processes. A basic method to form bulk metal

into a desired final shape is through the process of metal cutting, also referred to as machining. Metal

cutting is essentially the removal of excess material from a workpiece by moving a working tool over the

0-8493-1580-8/05/$0.00+$.50q 2005 by CRC Press LLC 26-1149

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

surface of the workpiece at a certain depth, speed, and feed rate. Conventional machining operations are

turning, milling, and drilling.

Despite the high level of technology built into every aspect of modern metal cutting operations, the

phenomenon of tool wear still hampers the reliability and complete automation of machining processes.

Tool wear is the loss of material on the edge of the cutting tool. Although tool wear can be minimized, it

cannot be eliminated. Unfortunately, excessive or even a small quantity of tool wear may cause a defect in

a machined component, and therefore it is always necessary to be aware of the extent of the current tool

wear before machining can commence. Economic losses due to tool wear occur as a result of the

scrapping of expensive parts and the nonoptimal use of tool inserts. A conservative approach is often

taken, and the insert is recycled long before it should have been. Furthermore, secondary damage due to

tool wear can be extreme and even catastrophic. For this reason, many approaches to tool condition

monitoring (TCM) have been proposed through the years. There exist sensorless and sensor-based TCM

approaches. Sensorless approaches are generally tool-life equations and not monitoring methods. Thus,

sensorless approaches attempt to determine the optimal tool life under certain machining conditions.

These are often extended versions of the famous Taylor equation, which is described by

vTn ¼ C ð26:1Þ

where v is the cutting speed, T is the tool life, and n and C are constants that must be determined

experimentally for a given tool and workpiece combination.

This chapter is focused on sensor-based TCM, and specifically those methods that are based on

vibration related properties such as force, acceleration, and acoustic emission (AE). These sensor types

are known to be most effective for TCM. Furthermore, discussions will be focused on the application of

TCM in turning operations, though reference will be made to other machining operations as well. Besides

vibration-based approaches, other sensor based TCM methods are:

* Use of noncontact capacitive sensors* Vision systems* Measurement of the motor current* Surface roughness monitoring* Ultrasonic monitoring* Temperature monitoring* Laser scatter methods* Audible emission monitoring

The reader is also referred to other excellent overviews of sensor-assisted TCM, published by Dan and

Mathew (1990), Byrne et al. (1995), Scheffer and Heyns (2001a), and Dimla (2001). A TCM database wasQ1

also published by the CIRP, supervised by Teti (1995), which includes more than 500 research papers

focusing only on TCM.

26.2 Mechanics of Turning

26.2.1 General Terms

A typical turning operation is schematically shown in Figure 26.1. The cutting tool moves parallel to the

workpiece and spindle, and hence reduces the diameter of the shaft. The most important machining

parameters are:

* Cutting speed (usually expressed in m/min)* Feed rate (usually expressed in mm/rev)* Depth of cut (usually expressed in mm)

The force response on the tool tip due to the turning operation consists of three components: Fx, Fy,

and Fz. These forces consist of a static and a dynamic part, as shown in Figure 26.2. The static forces are

governed by the static pressure between the tool and workpiece, and are a function of the machining

Vibration and Shock Handbook26-1150

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

parameters. The dynamic forces are governed by forced and free vibrations due to excitation from the

cutting operation. Analytical models exist that can describe the static forces for basic machining

operations (Merchant, 1945). The dynamic behavior is more difficult to model theoretically, although

there is also continuous research in this area (Kapoor et al., 1998).

One of the main difficulties of monitoring tool wear with vibration is to identify the frequency range

that is influenced by tool wear, since machining processes entail various mechanisms that produce

vibrations that are not related to tool wear. The frequency range of vibrations produced during ordinary

machining operations usually falls between 0 and 10 kHz. From the literature, it can be concluded that

the frequency range sensitive to tool wear depends entirely on the type of machining operation, and must

be determined experimentally for each individual case. There are two important vibration frequencies

present during cutting:

* The natural frequencies of the tool holder and its components* The frequency of chip formation

Dynamic tests should be conducted to identify the dynamic properties of tool holders (Scheffer and

Heyns, 2002a). However, the interaction of the

working tool engaged into the rotating workpiece

complicates the situation, and as a result the

dynamic behavior during cutting could be differ-

ent from the expected behavior obtained from off-

line tests. Scheffer and Heyns (2004) compared

continuous cantilever models with modal hammer

tests for different tool holder overhang lengths.

The natural frequency of the first mode as a func-

tion of overhang length is plotted in Figure 26.3

(for a specific tool holder). It can be seen that a

continuous fixed-free cantilever beam model

corresponds well with the results obtained with

hammer tests.

FIGURE 26.1 Turning operation.

dynamic cuttingforce

0times [S]

forc

e [N

]

static cutting force

FIGURE 26.2 Static and dynamic forces.

Vibration-Based Tool Condition Monitoring Systems 26-1151

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

The chip formation frequency can be calculated

with simple equations that take the machining

conditions into account (Lee et al., 1989). The tool

holder natural frequencies and chip formation

frequency are independent. Generally, tool wear

has a larger effect on the vibration amplitudes at

the tool holder natural frequencies but can

influence chip formation frequencies as well.

26.2.2 Chatter Vibrations

Another phenomenon important to machining

operations is tool chatter vibrations. These are

self-excited vibrations resulting from the gener-

ation of different chip thicknesses during machin-

ing. Initially, cutting forces excite a structural

mode of the machine–workpiece system. This

leaves a wavy surface finish on the workpiece.

During the next revolution, another wavy surface

is produced in the same way. Depending on the

phase shift between these two waves, the maximum chip thickness can grow and oscillate at a particular

frequency that is close to that of a structural mode. This is called the regenerative chatter frequency.

Chatter cause a poor surface finish and can also lead to tool breakage.

The analysis and prediction of chatter has been the subject of research for many years. Morimoto et al.

(2000) developed a piezoelectric shaker/actuator to regenerate the vibrations of the cutting process. In

this way, unwanted vibrations such as chatter can be attenuated. The system is also helpful to determine

the dynamic properties of the machine tool. Koizumi et al. (2000) used a very interesting approach called

the correlation integral in the time domain to identify chatter onset. Lago et al. (2002) designed a sensor

and actuator integrated tool for turning and boring to control chatter. The tool holder shank vibrations

are sent to the actuator via a digital controller. An adaptive feedback control system is used to perform

broadband vibration attenuation up to 40 dB at different frequencies simultaneously.

26.2.3 Tool Wear

26.2.3.1 Tool Failure Mechanisms

Tool wear is caused by mechanical loads, thermal loads, chemical reactions, and abrasive loads. The load

conditions are in turn influenced by the cutting conditions and materials. The different loads can cause

certain wear mechanisms that may occur in combination. These mechanisms have either a physical or

chemical characteristic that causes loss or deformation of tool material. Tool wear mechanisms can be

classified into several types, summarized as follows (Du, 1999):

* Abrasive wear resulting from hard particles cutting action* Adhesive wear associated with shear plane deformation* Diffusion wear occurring at high temperatures* Fracture wear due to fatigue

Other wear mechanisms are plastic deformation and oxidation, which are not very common in

industry. It is estimated that 50% of all tool wear is caused by abrasion, 20% by adhesion, and the other

10% by the other mechanisms (Kopac, 1998). Abrasion is basically the grinding of the cutting toolQ2

material. The volume of abrasive wear increases linearly with the cutting forces. Higher hardness of the

tool material can reduce the amount of abrasive wear. During adhesion, the high pressures and

temperatures on the roughness peaks on the tool and the workpiece cause welding. These welding points

7

6

5

4

3

2

1

040 45 50 55 60

1st m

ode

freq

uenc

y [k

Hz]

overhang length [mm]

fixed-free cantilever beamhammer tests

FIGURE 26.3 Frequency of first tool holder mode.

Source: Scheffer, C. and Heyns, P.S., Mech. Syst. Signal

Process, Elsevier, 2004. With permission.


154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

are broken many times every second due to the workpiece movement and as a result cause removal of the

tool material (Kopac, 1998). Diffusion wear occurs at even higher cutting speeds, where very high

temperatures are present (especially when using hard metal tools).

26.2.3.2 Tool Failure Modes

Tool wear will generally occur as a combination of a number of wear modes, with one mode predominant.

The dominant mode will depend on the dominant wear mechanism, which in its turn is influenced by the

machining conditions and the choice of tool and workpiece material. For a given tool and workpiece

combination, the dominant wear mode can be determined at different cutting speeds using the product of

the cutting speed and the undeformed chip thickness (Dimla, 2000). The common wear modes are:

* Nose wear* Flank wear* Crater wear* Notch wear* Chipping* Cracking* Breakage* Plastic deformation

Figure 26.4 is a graphical representation of the different tool failure modes. The consequences of tool

wear are deviations in shape and roughness of the machined part that cause the part to be discarded because

it is out of tolerance. Most wear modes cause an increase in cutting forces, although this is not always the

case for all tool and workpiece combinations. The most widely researched tool failure modes for turning

with single point tools are flank wear, breakage (fracture), and crater wear. Flank and crater wear are

accepted as normal tool failure modes, because the other failure modes can be avoided by selecting the

proper machining parameters. The growth of flank and crater wear is directly related to the total cutting

time, unlike some of the other failure modes, which can occur unexpectedly even with a new tool.

FIGURE 26.4 Tool failure modes. Source: Scheffer, C. and Heyns, P.S. 2001. COMA 01, University of Stellenbosch.

Q7

With permission.


205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

26.2.3.3 Tool Wear Measurement

Wear measurements of tool inserts are done

through the implementation of an appropriate

international standard, ISO 3685. Flank wear is

quantified in terms of VB, which is the mean of the

wear height on the tool flank. The length of flank

wear is also measured in terms of b. Crater wear is

quantified in terms of the crater depth, K. The

parameters are depicted in Figure 26.5, which is a

scanning electron microscope (SEM) picture of a

worn turning insert.

26.2.3.4 Tool Wear Stages

It is assumed by most authors that tool wear

consists of an initial, a regular, and a fast wear stage

(Zhou et al., 1995). Some authors divide tool wear

into five distinct stages (Bonifacio and Diniz, 1994):

1. Initial stage of wear

2. Regular stage of wear

3. Microbreakage stage of wear

4. Fast wear stage

5. Tool breakage

It has been established by various researchers

that the initial and fast (before tool breakage)

stages occur more rapidly than the regular stage.

Bonifacio and Diniz (1994) explain that, during

the fast wear stage with coated carbide tools, the

tool loses its coating and the tool substrate (which

has less resistance) begins to perform the cut and

wears faster. During the initial stage, the tool edge

loses its sharp edge rapidly, after this the process

stabilizes for a given time. Flank wear in relation to

total cutting time will typically appear as depicted

in Figure 26.6.

The geometrical growth and rate of wear is

unique for every tool insert, even those used with

the same machining parameters. Wear measure-

ments conducted on the shop floor of a piston

manufacturer by Scheffer and Heyns (2004) are

shown in Figure 26.7. It was found that the tools

last between 1000 and 6000 components, which

makes the optimal use of the tool extremely

problematic if the wear is not monitored on-line.

The reason for this behavior is mainly attributed to

fluctuating conditions on the shop floor, for

example, the rate at which components are

manufactured. If the time allowed for the tool to

cool down between workpieces is not constant,

large variations in the tool life can be expected.

FIGURE 26.5 Tool wear parameters.

1 2 34

5initial regular fast

cutting time

flank wear

FIGURE 26.6 Flank wear in relation to cutting time.

0.25

0.2

0.15

0.1

0.05

0 40002000 6000

flan

k w

ear

VB

[m

m]

number of workpieces

FIGURE 26.7 Typical variations in tool life. Source:

Scheffer, C. and Heyns, P.S., Mech. Syst. Signal Process,

Elsevier, 2004. With permission.


256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

26.3 Vibration Signal Recording

The information from vibration sensors can be treated in numerous ways. The overall aim of a tool

condition monitoring system (TCMS) is to utilize the best processing techniques to extract the relevant

information from sensor signals. Generally, a TCMS consists of the steps depicted in Figure 26.8. Various

methods that could be used in each step will be discussed in more detail.

26.3.1 Direct and Indirect Systems

TCMSs can be divided in two categories, namely, direct and indirect. Direct methods are concerned with

a measurement of volumetric loss at the tool tip, while indirect methods use a pattern in sensor data to

detect a failure mode (Byrne et al., 1995). Direct methods do not utilize vibration and will not be

discussed here. In general, direct methods are sensitive to dirt and cutting chips, and consequently they

are not commonly accepted in industry. Indirect methods have found more acceptance in industry due to

the fact that most indirect methods are easily interpreted, cost-effective, and often more reliable than

direct methods. Also, for some applications, it might not be possible to use a direct monitoring method

due to the nature of the process.

26.3.2 Sensor Requirements for Tool Wear Monitoring

Machine tools represent very hostile environments for sensors. Sensors used for TCM (also see Chapter

15) must meet certain requirements, such as (Byrne et al., 1995) the following:

* Must measure as close as possible to the point of metal removal* Must not cause a reduction in the stiffness of the machine tool* Must not cause a restriction of the working space of the machine* Should be wear and maintenance free, easy to replace, and of low cost* Must have resistance to dirt, chips, and electromagnetic and thermal influences* Should function independent of tool and workpiece* Must provide reliable signal transmission, e.g., from rotating to fixed machine components

26.3.3 Force Measurement

Worn tools cause an increase in the cutting force components. It is also known that both the dynamic and

static components generally increase with tool wear due to frictional effects. The three components of the

cutting force each responds uniquely to varying machining parameters and the different wear modes.

Depending on the type of process that is investigated and the specific experimental setup, results among

researchers vary. This can be attributed to dynamic effects of the machine tool and the measurement

equipment. There are a number of different sensor configurations to collect forces from machining

operations and these are described below.

26.3.3.1 Direct Measurement Dynamometers

Tool holder dynamometers are by far the most popular method for collecting cutting forces. These

sensors utilize the piezoelectric effect and can measure quasistatic and dynamic cutting forces very

accurately. However, dynamometers are very expensive and bulky instruments and are not practical for a

sensor selectionand deployment

signal recordingand conditioning

generate signalfeatures

select wearsensitive features

model featuresand wear

relationship

FIGURE 26.8 TCMS steps.


307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

typical shop floor. Furthermore, their usable

frequency range is limited to approximately

1 kHz. An example of a tool holder dynamometer

is shown in Figure 26.9.

Tarmal and Opavsky (2000) investigated the

dynamics of a conventional force dynamometer

for machining operations. It was found that the

dynamometer has significant amplitude distortion

in the frequency range that is quoted as the

operating range by the manufacturer. The authors

suggest that the dynamic characteristics of the

dynamometer (while clamped as it would be

during measurements) be identified with a

modal test and the effect of dynamometer

dynamics be compensated for after measurements

are made to obtain the true cutting force.

26.3.3.2 Indirect Force Sensors

There are numerous small force sensors available

for the purpose of force measurement on machine

tools. These measure forces in load-carrying

components of the machine tool and are thus

not direct force measurement devices. The advan-

tages of these sensors are their size, low cost, and

significantly higher operational frequency range.

A disadvantage is that a suitable position for the

sensor can only be determined experimentally.

These sensors are suitable for tool breakage

monitoring in rough machining or detection of

other catastrophic events such as collisions. An

example of a three-component force sensor is

shown in Figure 26.10.

26.3.3.3 Piezoelectric Strain Sensors

The use of piezoelectric strain sensors for wear

monitoring of synthetic diamond tool inserts was

reported by Scheffer and Heyns (2000a). These

sensors are ultrasensitive to changes in cutting

forces if they are installed in an appropriate

location. The best location for the sensor must

once again be determined experimentally, but

generally it should be installed on a load-carrying

component of the machine as close as possible to

the tool tip, for example, on the tool holder itself

(Scheffer and Heyns, 2001b). An example of a

piezoelectric strain sensor that can be used on machine tools is shown in Figure 26.11.

26.3.3.4 Resistance Strain Gauges

A quite simple method to estimate both the static and dynamic components of cutting forces without any

distortion is to use resistance strain gauges (see Chapter 15). These comply with most of the requirements

for TCM sensors, and they can accurately follow the static and dynamic response of a system up to

FIGURE 26.9 KISTLER force dynamometer type 9121.

Source: KISTLER Brochure 2002. Courtesy of Kistler

Instrumente AG.

FIGURE 26.10 KISTLER three-component force sen-

sor type 9251A. Source: KISTLER Brochure 2003.

Courtesy of Kistler Instrumente AG.

FIGURE 26.11 KISTLER strain sensor type 9232A.

Source: KISTLER Brochure 2004. Courtesy of Kistler

Instrumente AG.


358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

50 kHz. Scheffer and Heyns (2002a) developed a sensor-integrated tool holder using strain gauges. It was

shown that the system is robust, cost-effective, and fit for an industrial TCMS. The physical layout of the

strain gauges on a boring bar is shown in Figure 26.12. The system was calibrated with a special device to

directly obtain the three cutting forces from the strain gauge signals.

26.3.3.5 Customized Force Sensors

There are a number of customized force sensors available that can be used with specific machining

operations. These are:

* Force measuring plates, pins, and bearings* Special force measuring bolts* Force and torque measuring rings that fit on spindles

26.3.4 Acceleration Measurement

Piezoelectric accelerometers can measure the machine vibration caused by oscillations of cutting forces. It

is well known that high-frequency vibrations (higher than 1 kHz) yield large acceleration levels, giving

accelerometers an advantage over force-based monitoring. Accelerometers fulfill the environmental

requirements for tool wear monitoring because they are resistant to the aggressive media present during

machining. Accelerometers are also less expensive than force dynamometers and can measure vibration

levels within a very wide frequency range, typically 5 Hz to 10 kHz.

Various authors have shown that acceleration levels change with tool wear. Li et al. (1997) found that

the coherence function of two crossed accelerations can be used as an easy and effective way to identify

tool wear and chatter. They found that with progressive tool wear, the autospectra of the two

accelerations and their coherence function increase gradually in magnitude around the first natural

frequencies of the cross-bending vibration of the tool shank. As the tool approaches a severe wear stage,

the peaks of the coherence function increase to values close to unity. Scheffer et al. (2003) reported on the

use of an accelerometer for wear monitoring during hard turning. It was found that certain frequencies

show repeatable amplitude increase with increasing tool wear. These frequencies corresponded to the tool

holder natural frequencies. Some authors, for example, Bonifacio and Diniz (1994), also found that aQ3

wear sensitive frequency will increase with increasing tool wear and then suddenly decrease near the end

of tool life. This can be attributed to an increased damping effect due to plastic deformation and

microbreakage of the cutting edge.

26.3.5 Acoustic Emission Measurement

Cutting processes produce elastic stress waves that propagate through the machine structure. Different

sources in the cutting process generate these stress waves known as acoustic emission (AE). Sources of AE

FIGURE 26.12 Application of resistance strain gauges. Source: Scheffer, C. and Heyns, P.S., Mech. Syst. Signal. With

permission.


409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

in metal cutting are:

* Friction on the tool face and flank* Plastic deformation in the shear zone* Crack formation and propagation* Impact of the chip at the workpiece* Chip breakage

A typical AE sensor for use on machine tools is

shown in Figure 26.13.

The fact that crack formation generates AE

makes AE ideal for tool breakage detection.

Collection of the AE requires special hardware that can bandpass filter the signals to the AE

range (between approximately 50 and 250 kHz). Furthermore, amplification is required and an

analogue root-mean-square (RMS) circuit with a short time constant is generally also included to collect

the AERMS. The different steps required to collect AE are depicted in Figure 26.14 (adapted fromQ4

Jemielniak, 2000).

Araujo et al. (2000) investigated sliding friction as a possible source of AE during metal cutting.

The AERMS values in different frequency ranges were collected for different widths of cut and also

with the tool rubbing against the workpiece without cutting. It was found that the level of AE

remains almost constant for all width of cut conditions, and hence it was concluded that the main

mechanism for AE during metal cutting is the sliding friction between the tool and workpiece.

Consequently, an increase or decrease of AE can be expected with tool wear depending on the effect

on the sliding friction due to that tool wear. Furthermore, it is believed that the cutting

temperatures will affect the AE due to thermal expansion effects. Chiou and Liang (2000)

investigated AE with tool wear and chatter effects in turning. A model is presented that can predict

the chatter AERMS amplitude with certain levels of flank wear. Good correlation was found between

the model and the experimental results. Kim et al. (1999) reported on the use of AE to monitor the

tool life during a gear shaping process. The AERMS is collected and used in a software program to

predict the remaining tool life.

Li (2002) presented an overview of using AE for TCM in turning operations. It is stated the AE is

heavily dependent on cutting conditions and, as a result, methods should be employed to handle this

problem effectively. Some methods are proposed that include advanced signal processing, sensor fusion

and modeling techniques. Many other AE-based tool wear and breakage monitoring systems have been

implemented successfully in research. One problem still lies with an appropriate interpretation of the AE

frequency spectrum. In most studies, an explanation for the choice of certain frequencies and their

advantages are not given or not investigated. In fact, Jemielniak (2000) found that using the average value

of AE (or AERMS) is the most suitable. A similar conclusion was made by Scheffer et al. (2003), who

compared different processing methods of the AE signal during hard turning.

FIGURE 26.13 Kistler AE sensor type 8152B. Source:

PCB Website 2002. Courtesy of Kistler Instrumente AG.

FIGURE 26.14 Steps for collecting AE during turning. Source: KISTLER brochure 2004. Adapted from Jemielniak,

K., Ultrasonics, 2000.


460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

510

26.3.6 Sensor Comparisons

Future research should be directed towards directly comparing different sensors for tool wear

monitoring. Choi et al. (1999) developed a single sensor for parallel measurement of force and AE. A

finite element analysis was carried out to determine the optimal position for the sensor away from the

tool holder because the sensor obstructed the working space of the machine. The approach was successful

for breakage detection but no wear estimations are reported. Barrios et al. (1993) compared AE,

vibration, and spindle current for TCM during milling. It was found that the spindle current is the most

sensitive sensor for detecting tool wear, with AE the least sensitive. However, contradictory results are

reported in other publications, and hence more research would be required to determine ultimately

which sensor is the best for which machining operation. Govekar et al. (2000) compared force and AE

methods for TCM, and concluded that the best result is achieved when sensory information is combined.

Dimla and Lister (2000a) compared the use of force and vibration signals for TCM and also combined the

information in a single decision-making technique (Dimla and Lister, 2000b). Similar comparative

studies were reported by Scheffer et al. (2003).

26.4 Signal Processing for Sensor-Based Tool ConditionMonitoring

Using the sensor information from the different sensor systems described in the previous section, a

decision must be made with respect to the tool condition. This decision is generally referred to as the data

classification. It is often better to combine sensory information to solve a complex problem such as TCM.

Such a combined approach is referred to as sensor fusion. Sick (2002) proposed a generic sensor fusion

architecture for TCM, which summarizes the various sensor fusion levels of a TCMS. These are:

* Analogue preprocessing* Digital preprocessing* Feature extraction* Wear model* Decision making

Fusion of sensor information can occur at any of these levels. Analogue and digital preprocessing are

activities such as signal amplification, conditioning, filtering, calibration, and temperature compen-

sation. The feature extraction step is probably the most important step, because here the sensor signals

must be condensed and reduced to only a few appropriate wear sensitive values. Many different methods

are available to achieve this. The wear model level establishes a relationship between the chosen features

and the tool condition. In many cases, neural networks (NNs) are used in this step, and sensor fusion

takes place within the NN. A decision level can also be included where a final decision is made with

respect to the tool condition, for instance a “competing experts” formulation if a TCMS is used in

conjunction with a tool-life equation. Discussions on the various techniques follow.

26.4.1 Feature Extraction

Most decision-making techniques for process monitoring are based on signal features. Through

appropriate signal processing, features can be extracted from signals that show consistent trends with

respect to tool wear. Features are mainly derived through processing in the time, frequency, or joint time-

frequency domain or statistical analysis.

26.4.1.1 Time Domain

Features extracted from the time domain are usually fundamental values such as the signal mean or RMS.

Other techniques include the shape of enveloping signals, threshold crossings, ratios between


511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559

560

561

time-domain signals, peak values, and polynomial

approximations of time-domain signals. Examples

of time-domain features from an interrupted

cutting operation are shown in Figure 26.15.

It has been found that some of the time-domain

features show good correlation with tool wear and

are easy to implement (Scheffer, 1999). Bayramo-

glu and Dungel (1998) investigated the use of

several different force ratios (calculated from the

static cutting forces). It was found that certain

force ratios can be used to monitor tool wear

under a wide range of cutting conditions. Most

commercial TCMS rely on time-domain infor-

mation. However, time-domain features are

known to be sensitive to disturbances and should

be complemented with features from another

domain.

26.4.1.2 Frequency Domain

The power or energy of certain frequency bands

in the fast Fourier transform (FFT) is often

suggested as a feature for TCM. It is very

challenging to identify spectral bands that are

sensitive to tool wear. It is even more difficult to

determine exactly why these frequencies are

influenced by tool wear. Power in certain bands

will often increase due to higher excitation forces

because of the increase in friction when the tool

starts to wear. Sometimes a peak in the FFT will

also shift due to changing process dynamics as a

result of tool wear. An early frequency-domain

approach is reported by Jiang et al. (1987), in

which frequency-band energy is determined from

the power spectral density (PSD) function as a

feature for tool wear.

Some authors suggest that two frequency ranges be identified from the original signal (Bonifacio and

Diniz, 1994). The one range must be sensitive to tool wear, the other must be insensitive. For instance, if

the measurement was made from 0 to 8 kHz, it must be split (using appropriate filters) into a 0 to 4 kHz

signal and a 4 to 8 kHz signal. If the lower range is more sensitive to tool wear, a ratio between the two

ranges can be calculated. If this ratio exceeds a certain pre established value, it can be concluded that the

end of the tool life has been reached. This can also apply for a ratio between the signal recorded from a

fresh tool to that compared with a worn tool. Examples of frequency-domain features from cutting forces

are shown in Figure 26.16.

One difficulty with frequency-domain approaches is that the dynamics of the operation and

measurement hardware is not always fully understood. The fact that measurement hardware dynamics

instead of process dynamics are often measured was also recently identified by Warnecke and Siems

(2002). The response of a force dynamometer is influenced by its clamping condition, which may cause it

to experience nonlinearities at relatively low frequencies. There are also some uncertainties when using

these instruments, relating to their calibration and other varying parameters. A model for expressing the

uncertainties when collecting cutting forces with a dynamometer was proposed by Axinte et al. (2001).

These uncertainties might be responsible for the scatter of force components often reported in the

FIGURE 26.15 Simple time-domain features.

FIGURE 26.16 Frequency-domain features.


562

563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

612

literature. An interesting study is also reported by Bahre et al. (1997), concerning determination of the

natural frequencies of the machine tool components using the finite element method (FEM). These are

taken into account for interpretation of the vibration/AE signal.

26.4.1.3 Statistical Processing

In the case of statistical features, signals are assumed to have a probabilistic distribution, and

consequently, useful information can be extracted from the statistics of the distribution. Hence, the signal

is regarded as a random process. Generally, machining processes are nonstationary but are assumed to be

stationary for the short periods during which features are calculated. Several statistical features have been

investigated for TCM and can be applied to several machining operations. The main features are those

that describe the probability distribution of a random process (variance, standard deviation, skewness,

kurtosis, etc.) and coefficients of time-series models. There are also miscellaneous other statistical

features, such as cross-correlations, the coherence function, and the harmonic mean.

One useful approach is the use of autoregressive (AR) and autoregressive moving average (ARMA)

coefficients. AR coefficients computed for a signal represent its characteristic behavior. When the signal

changes during the cutting operation due to tool wear, the model coefficients also change and can then be

utilized to monitor the progressive tool wear. Baek et al. (2000) report on the use of an eighth order AR

model for tool breakage detection during end milling. It was found that the AR approach is somewhat

more accurate than the frequency band energy method. Yao et al. (1990) used the ARMA method to

decompose the dynamic cutting force signals, and wear-sensitive frequencies were identified. This

assisted in identifying the importance of certain vibration modes with respect to TCM.

The use of statistical process control (SPC) methods is also reported by some authors. Jun and Suh

(1999) considered the X-bar and exponentially weighted moving average (EWMA) for tool breakage

detection in milling. Jennings and Drake (1997) used statistical quality control charts for TCM. Different

statistical parameters are calculated and examples of one-, two- and three-variable control charts are given.

26.4.1.4 Time–Frequency Domain

The most common time–frequency domain processing method in TCM applications is wavelet analysis.

A comprehensive discussion on the advantages and disadvantages of wavelet analysis for TCM is

described by Sick (2002). It is often stated that wavelets are used because they provide information about

the localization of an event in the time as well as in the frequency domain. However, locating discrete

frequency-related events in the time domain is rarely of importance with respect to tool wear (which is a

gradually increasing phenomenon). In contrast, tool breakage will have a large localized effect in the time

domain, but this can be monitored more effectively using time-domain techniques. Furthermore,

wavelets are time variant and the exact contribution of a particular frequency at any given time can never

be determined accurately due to Heisenberg’s uncertainty principle.

Despite the above arguments, the use of wavelet analysis for TCM is reported in several publications.

Lee and Tarng (1999) use the discrete wavelet transform for cutter breakage detection in milling and find

that the technique is reliable even under changing machining conditions. Luo et al. (2002) published

results of a TCMS using wavelet analysis of vibration signals. In this case, the wavelet is used as a filter to

enhance wear-sensitive features in the signals. However, the results are not compared with conventional

digital filtering. A comparative study between wavelets and digital filtering for tool wear monitoring was

carried out by Scheffer (2002). It was found that, although the wavelet packets act as automated filters, a

very similar (if not better) result could be achieved with appropriate digital filtering. The use of wavelets

increase the complexity of the TCMS, which is a disadvantage for shop-floor implementations.

Furthermore, the results from digital filtering can be physically related to the machining operation and

tool wear, whereas the behavior of wavelet packets is more difficult to interpret.

Another method of time–frequency analysis that can be applied for TCM is spectograms (e.g., the

Gabor distribution). Spectograms are very useful to identify stationarity in dynamic signals, and

for detection of disturbances that may be time-localized in signals. The use of the Choi–Williams

Q5 time–frequency distribution for TCM during multimilling is described by James and Tzeng (2000).


613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

660

661

662

663

Wear-sensitive regions on the time–frequency

distribution are calculated and used as inputs to

a NN for wear classification. An example of a

change in the dominant chip curl frequency during

hard turning is shown in Figure 26.17. It is obvious

that, due to some disturbance (perhaps tool wear),

the dominant dynamic force frequency “jumps”

from 27 to 9 Hz.

26.4.2 Feature Selection

Various authors attempt to generate features that

are sensitive to tool wear but insensitive to

changing machining parameters. For most oper-

ations, the machining parameters can be included

in the wear model and hence the sensitivity of the

features is not such an important issue. There are

also other techniques to normalize sensor data

with respect to machining parameters, for instance

the use of a theoretical model (Sick, 1998). This is

very useful if the machining conditions change so often that not enough data can be collected for training

or calibrating a model. Numerous techniques exist to select the most wear-sensitive features or to reduce

the input feature matrix to a lower dimension. The main techniques for feature selection and reduction

are listed below:

* Principal component analysis (PCA)* Statistical overlap factor (SOF)* Genetic algorithm (GA)* Partial least squares (PLS)* Automatic relevance determination (ARD)* Analysis of variance (ANOVA)* Correlation coefficient* Simulation error calculations

Al-Habaibeh et al. (2000) presented a TCMS for a parallel kinematics machine tool for high-speed

milling of titanium. An interesting approach to feature selection is employed, called self-learning

automated sensors and signal processing selection (ASPS). This approach is based on an on-line self-

learning methodology, whereby a certain feature will be selected automatically based on a correlation

with tool wear. A linear regression is performed on each feature in the sensory feature matrix to detect the

sensitivity of each feature with respect to tool wear. A very interesting cost analysis is then preformed to

determine if the installation of a sensor justifies its costs.

Ruiz et al. (1993) proposed the use of a discrimination power for feature selection in a TCM

application. The method is similar to that of the SOF. An automated version is proposed that also checks

for linear correlation between features. It is difficult to assess the success rate of the automated procedure

because the experiments/simulations are not described in enough detail. Lee et al. (1998) describe the use

of ANOVA to determine the best force ratio for TCM statistically. Several ratios between the three main

cutting forces are computed and the influence of controllable parameters (e.g., machining conditions) on

these ratios are investigated by means of ANOVA.

Du (1999) describes the use of a blackboard system, which is a knowledge-based approach for feature

selection and decision-making. An advantage is the fact that a physical interpretation of a feature can be

linked to phenomena in the machining operation. The method is also flexible, but suffers from the

disadvantage of requiring a large quantity of data and expertise to establish the knowledge-based rules.

FIGURE 26.17 Time-frequency distribution of cutting Q7

force signal. Source: Scheffer, et al., Int. J. Mach. Tools

Manuf., Elsevier, 2003. With permission.


664

665

666

667

668

669

670

671

672

673

674

675

676

677

678

679

680

681

682

683

684

685

686

687

688

689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

713

714

Mdlazi et al. (2003) compared the performance of ARD and PCA for feature selection for two damage

detection case studies. It was found that the performance of the methods is similar, but one might

perform better on a particular data set. Generally speaking, the PCA yields better results for damage

detection problems. Scheffer and Heyns (2002b) compared several feature selection methods for TCM,

such as SOF, PCA, GA, ANOVA, and the linear correlation coefficient. It was found that the correlation

coefficient approach and the SOF should be preferred for TCM applications. PCA could also be of

assistance, but the feasibility of PCA for on-line applications is still questionable. The correlation

coefficient and SOF is expressed as percentages in Figure 26.18 (from Scheffer and Heyns, 2004) for 30

different wear monitoring features in a turning tool wear case study. Ideally, a feature with a high level of

correlation and SOF should be selected.

As a last step, engineering judgment is required for proper feature selection because automated

methods will often select features that are dependant on one another, thus not achieving the goals of

sensor fusion. The following rules can be used as a guideline for selecting features for TCM:

* Select features from the static and dynamic parts of force signals.* Select features measured in different directions.* Use time- and frequency-domain features.* Features based on simple signal processing methods are preferred.* There should be a reasonable physical explanation for the behavior of a feature with respect to

tool wear.

26.5 Wear Model/Decision-Making for Sensor-Based ToolCondition Monitoring

26.5.1 Trending, Threshold

A very simple decision-making technique is to

trend features and to establish threshold values.

When a certain feature or set of features crosses a

threshold value, an estimation of the tool con-

dition can be made. Unfortunately, these threshold

values can only be determined experimentally.

The difficulty with this method is to determine

the correct threshold value, especially under

diverse cutting conditions. Furthermore, the

method is extremely sensitive to disturbances.

The trend of the mean feed force with increasing

flank wear is shown in Figure 26.19, and two

thresholds are shown as examples. It is clear that

this technique is not very reliable due to the large

variance in the trend.

FIGURE 26.18 Comparison of correlation coefficient and SOF for feature selection. Source: Scheffer, C. and Heyns,

Q7

P.S., Mech. Syst. Signal Process, Elsevier, 2004. With permission.

threshold-replace

threshold-warning

flank wear VB [mm]

feat

ure

valu

e [n

orm

alis

ed]

0.04 0.06 0.08

−1

−0.5

0

0.5

1

FIGURE 26.19 Example of trend and thresholds.

Source: Scheffer, C. and Heyns, P.S., South African Inst.

Q7

Tribol. 2002. With permission.


715

716

717

718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

734

735

736

737

738

739

740

741

742

743

744

745

746

747

748

749

750

751

752

753

754

755

756

757

758

759

760

761

762

763

764

765

26.5.2 Neural Networks

The use of NNs as a secondary, more sophisticated signal processing and decision-making technique is

often found in TCM applications. The simultaneous utilization of many features and the robustness

towards distorted sensor signals are two of the most attractive properties of NNs. Neural networks also

assist in the fusion of sensor information for TCM. In other words, combining features from acceleration,

AE, and force signals in a NN can result in a method that can predict the tool condition with increased

accuracy (Silva et al., 1998). The successful implementation of NNs is dependent on the proper selection

of the network structure, as well as the use of the correct training and testing methods.

It is important to make a distinction between supervised and unsupervised NN paradigms.

Unsupervised NNs are trained with input data only, and are usually used for discrete classification of

different stages of tool wear. Supervised NNs are trained with input and output data, and these are used

for continuous estimations of tool wear. Furthermore, a distinction should be made between dynamic

and static NNs. In the case of dynamic NNs, temporal (time) information is included in the network with

the aim to model a time series. This can be done explicitly by using a time-based feature as an input to the

network, or implicitly by using recurrent networks or networks with tapped delay lines (TDLs). Dynamic

networks are preferred for TCM because tool wear is time-dependent (tool wear is a monotonically

increasing parameter that is partly a function of machining time).

26.5.2.1 Unsupervised Networks

There are two basic network paradigms for unsu-

pervised classifications, namely adaptive resonance

theory (ART) and the self-organizing map (SOM).

ART is based on competitive learning, addressing

the stability–plasticity dilemma of NNs. The main

advantage is its ability to adapt to changing

conditions. ART networks also have self-stability

and self-organization capabilities. The SOM is

actually a data-mining method used to cluster

multidimensional data automatically. A high-

dimensional feature matrix can be displayed on a

two-dimensional grid of neurons that are arranged

in clusters with similar feature values. Clusters for new and worn tools can be formed, and these are used for

automatic classification of the tool condition. A SOM is depicted schematically in Figure 26.20.

There are many practical advantages for using unsupervised networks. One is the fact that the

machining operation is not interrupted for tool wear measurements during the training phase. There is

also the advantage of practical implementation if machining conditions change very often and

appropriate training samples for supervised learning cannot be collected. Furthermore, the numerous

different combinations of tool and workpiece materials and geometries can make supervised learning

impossible. Normally, unsupervised NNs are used to identify discrete wear classes and cannot be used for

a continuous estimation of tool wear.

Silva et al. (2000) investigated the adaptability of the SOM and ART for tool wear monitoring during

turning with changing machining conditions. It was found that, with appropriate training, the methods

have enough adaptive capabilities to be employed in industrial applications. Govekar and Grabec (1994)

use the SOM for drill wear classification, where the SOM is used as a kind of empirical modeler. It was

found that the adaptability of the SOM and its ability to handle noisy data makes the technique viable for

on-line TCM. Scheffer and Heyns (2000b, 2001b) showed how a TCMS can be adaptable using SOMs.

Different network sizes were compared with define discrete classes of new and worn tools. Larger networks

yielded more continuous results. The TCMS using SOMs was applied to monitoring synthetic diamond

tools for an industrial turning operation. It was found that the SOM can be used for industrial applications,

especially if tool wear measurements are not available.

FIGURE 26.20 Schematic representation of the SOM.


766

767

768

769

770

771

772

773

774

775

776

777

778

779

780

781

782

783

784

785

786

787

788

789

790

791

792

793

794

795

796

797

798

799

800

801

802

803

804

805

806

807

808

809

810

811

812

813

814

815

816

Different NN paradigms were compared on a wear monitoring application for aluminum turning by

Scheffer and Heyns (2002b). It was shown that the SOM is useful to identify discrete wear classes, as

shown in Figure 26.21. If an exact value of the tool wear is required, supervised networks will yield better

results but will require proper training samples.

26.5.2.2 Supervised Networks

Common supervised NNs used for TCM are the multilayer perceptron (MLP), multilayer feedforward

(FF) network, recurrent neural network (RNN), supervised neuro-fuzzy system (NFS-S), time delay

neural network (TDNN), single layer perceptron (SLP) and the radial basis function (RBF) network. The

use of an SLP for TCM is described by Dimla et al. (1996), using the perceptron learning rule for training.

The SLP is useful to identify discrete classes of the tool condition. FF networks are usually trained with

the backpropagation algorithm. However, backpropagation should not always be the preferred choice

because other methods are known that outperform this technique in terms of training time and

generalization. The size of the hidden layers in multilayer networks should be optimized for performance.

Many contradictory statements about the use of MLP networks can be found in the literature. One of the

main problems is the selection of the number of input features, size of the network, and the number of

training examples that should be used.

A multilayer feedforward (FF) network is shown schematically in Figure 26.22. Normally, a nonlinear

activation function should be used in the first layer, and linear neurons in the subsequent layers. In the

case of the FF networks, the backpropagation algorithm is often used for training. Backpropagation is an

optimization algorithm based on steepest gradient descent.

The use of FF networks with the backpropaga-

tion training rule is reported by authors such as

Zhou et al. (1995), Das et al. (1996), and

Zawada-Tomkiewicz (2001). Cutting conditions

can also be included in such networks. Lou and Lin

(1997) describe the use of a FF network using a

Kalman filter to avoid the training problems

encountered with backpropagation for a TCM

application. The proposed method is less sensitive

to the network initializations that often cause

convergence problems with backpropagation.

Monitoring a dynamic system such as a cutting

process should be done with a dynamic modeling

technique such as dynamic NN paradigms, for

example, recurrent networks, TDNNs, or explicit

FIGURE 26.21 Unsupervised approach to wear monitoring with the SOM. Source: Scheffer, C. and Heyns, Q7

P.S., South African Inst. Tribol. 2002. With permission.

FIGURE 26.22 Multilayer FF network.


817

818

819

820

821

822

823

824

825

826

827

828

829

830

831

832

833

834

835

836

837

838

839

840

841

842

843

844

845

846

847

848

849

850

851

852

853

854

855

856

857

858

859

860

861

862

863

864

865

866

867

inclusion of temporal information in static networks. Recurrent NNs have feedback connections

from their output to their input. There are various types of recurrent NNs that are useful for

specific applications. Elman networks are quite interesting. Generally, they are two-layer networks

with feedbacks from the first layer output to the first layer input. This type of network can be used to

learn and model temporal patterns. A recurrent network and an Elman network are shown schematically

in Figure 26.23.

Liu and Altintas (1999) report on the use of a FF network using a combination of TDLs and recurrent

connections. Machining conditions are also included. It is stated that the system was integrated into an

industrial TCMS, but was never put to use due to lack of “… robust, practical cutting force sensors …”

(Liu and Altintas 1999). Scheffer and Heyns (2002b) report on the use of an Elman NN for TCM. It was

found that the Elman network has a very smooth response and yielded better results than static NN

paradigms. It should be mentioned that the Elman network requires more time for training, but because

this is done off-line, training time should not be a criterion for evaluating NNs.

Neuro-fuzzy systems (NFS-S) attempts to combine the learning ability of NNs with the interpretation

ability of fuzzy logic. A TCMS using an NFS-S can be generated almost automatically because the fuzzy

rules can be learned by the NN. A combination of supervised and unsupervised training is used for

NFS-S. An in-process NFS-S system to monitor tool breakage was designed and implemented

successfully by Chen and Black (1997), concentrating on end milling operations. Xiaoli et al. (1997)

as well as Chungchoo and Saini (2002) also propose some of the advantages of using an NFS-S for TCM.

RBF networks are often preferred because of the convergence properties of the training algorithm.

In essence, convergence can be guaranteed and is often achieved much faster than in MLPs. The

accuracy of RBFs depends on the choice of the centers for the basis functions, and should be treated

with care. Pai et al. (2001) reported on the use of a resource allocation network (RAN) for TCM. The

RAN is a RBF network utilizing sequential learning. The RAN is compared with the MLP for wear

estimation during face milling. It was found that the RAN has faster learning ability but the MLP is

more robust.

TDNNs have delay elements in the feedforward connections, called TDLs. One advantage of

TDNNs over RNNs is that stability problems are avoided. An investigation towards the inclusion of

one and two phase delays for a TCM application was reported by Venkatesh et al. (1997). Different

network sizes were also investigated, and it was found that the NNs with temporal memory generally

perform better than those without memory. It is also stated that new algorithms should be

investigated for training. Sick and Sicheneder (1997) also describe the use of TDNNs for TCM in

turning. The TDNN is compared with the MLP and a significant improvement was found when using

FIGURE 26.23 Recurrent networks: feedback connection (left) and Elman network (right). Source: Scheffer, C. and Q7

Heyns, P.S., South African Inst. Tribol. 2002. With permission.


868

869

870

871

872

873

874

875

876

877

878

879

880

881

882

883

884

885

886

887

888

889

890

891

892

893

894

895

896

897

898

899

900

901

902

903

904

905

906

907

908

909

910

911

912

913

914

915

916

917

918

TDNNs. In another instance, Sick et al. (1998) compare the SOM, NFS-S, and MLP networks for

wear estimation. The following critical questions are used to evaluate the different NN paradigms

(Sick et al., 1998):

* Are the generalization capabilities of the NN sufficient (tested on previously unseen data)?* What rate of correct classification can be achieved for different wear stages?* Are the results repeatable (e.g., with a new initialization)?

In the case study presented by Sick et al. (1998), the best results were found with MLPs. It is

stated, however, that the results can be improved when using TDNNs, and such results are reported in

Sick (1998).

A novel combined approach is suggested by Sick (1998) to handle the effect of machining parameters.

An empirical model is used to normalize the data with respect to machining parameters before the data

are entered into the NN. Thus, machining parameters are not included in the NN itself. This approach

solves the extrapolation limitations encountered when an NN is tested with data recorded with

machining parameters it was not trained with. Although many authors test their NNs’ paradigms in such

a way, NNs cannot be expected to extrapolate. NNs should instead be tested with previously unseen data

recorded with same machining parameters it was trained with (hence an interpolation effect). This is a

problem because training and testing patterns for each condition must be supplied. However, if data can

be normalized with respect to machining parameters, training is only required for the normalized

condition. This was in effect achieved by Sick (1998). A difficulty still lies with establishing an appropriate

model, and in many cases it will also require a large number of experimental tests. A possible solution lies

in the incorporation of numerical models, for example, finite element models.

Scheffer (2002) presented another approach to tool wear monitoring of turning operations, using a

combination of static and dynamic NNs. Static networks are trained off-line to model selected features

from cutting forces. A dynamic NN that uses explicit temporal information is then trained on-line

with the particle swarming optimization algorithm (PSOA). The training goal of the dynamic NN

is to minimize the errors between the outputs of the static NNs and the on-line measurements.

The method was tested on various turning operations and was also tested on an industrial shop floor.

It was found that the method is more accurate and reliable than other NN paradigms and can be used

with cost-effective hardware (Scheffer and Heyns 2002a, 2004). The method is depicted schematically

in Figure 26.24.

FIGURE 26.24 Combined static and dynamic NN approach for turning. Source: Scheffer, C. and Heyns, P.S., South Q7

African Inst. Tribol. 2002. With permission.


919

920

921

922

923

924

925

926

927

928

929

930

931

932

933

934

935

936

937

938

939

940

941

942

943

944

945

946

947

948

949

950

951

952

953

954

955

956

957

958

959

960

961

962

963

964

965

966

967

968

969

26.5.3 Fuzzy Logic

Many authors have investigated the use of fuzzy logic to classify tool wear. It has been shown that fuzzy

logic systems demonstrate great potential for use in intelligent manufacturing applications. While NN

models cannot directly encode structured knowledge, it is often stated that fuzzy systems can directly

encode structured knowledge in a numerical framework. Additionally, fuzzy systems are capable of

estimating functions of a system with only a partial description of the system’s behavior.

Du et al. (2002) propose a very interesting method called transition fuzzy probability, which was

applied to a boring operation. This formulation can deal with the uncertainty of process conditions.

The method performs well because TCM has two uncertainties: that of occurrence and that of

appearance. The transition fuzzy probability solves this issue through the use of temporal information,

similar to dynamic NNs. The method was shown to outperform a backpropagation NN, although

very few details are given. It would be interesting to compare this method with dynamic NNs such

as TDNNs.

Fu et al. (1997) combined force, vibration, and AE in a fuzzy classifier for TCM during milling. Time-

and frequency-domain features were used, and it was found that combining the sensory information

achieved the best result. This is done within the fuzzy classifier. Li and Elbestawi (1996) and Kuo and

Cohen (1998) combine fuzzy modeling steps with NNs at different levels for TCM. The latter combined

force, vibration and AE in a multisensor approach with satisfactory results.

26.5.4 Other Methods

There are also a number of other decision-making and modeling methods that have been applied to

TCM, and these include:

* Knowledge-based expert systems (Du, 1999)* Pattern recognition algorithms (Kumar et al., 1997)* Dempster–Shafer theory of evidence (Beynon et al., 2000)* Hidden Markov models (Ertunc and Loparo, 2001; Ertunc et al., 2001)

Of these four approaches, only hidden Markov models have the potential possibly to outperform NNs

and fuzzy systems. However, not enough comparable research has been conducted in this area, and is it

certainly a worthwhile topic for future research.

26.6 Conclusion

Techniques for achieving TCM with vibration-based properties were presented in this chapter. The

sensing methods that have proved to be effective for TCM are force, acceleration, and AE. The sensors

employed must comply with certain requirements such as robustness and cost-effectiveness. Sensors

must be installed as close as possible to the point of metal removal in order to avoid signal-to-noise ratio

problems. Various techniques exist to condition and process the signals in analogue and digital formats.

The aim of signal processing is to generate wear sensitive features from the vibration signals. This could

be done by time, frequency, joint time–frequency, and statistical analysis. Feature selection can be

automated with a variety of procedures, but care must be taken when using these to avoid selection of

linearly dependent data.

The selected features can be used to establish a model of tool wear. Numerous research papers have

shown that NNs should be used due to the many advantages of NN modeling. The training and testing

procedures of NNs are of utmost importance if the system is considered for industrial implementation.

Care must be taken not to overtrain the networks because they will lose their ability to generalize.

Furthermore, NNs cannot be expected to perform well if they are tested with previously unseen

machining parameters. They should also be trained with the minimum and maximum tool wear that is


970

971

972

973

974

975

976

977

978

979

980

981

982

983

984

985

986

987

988

989

990

991

992

993

994

995

996

997

998

999

1000

1001

1002

1003

1004

1005

1006

1007

1008

1009

1010

1011

1012

1013

1014

1015

1016

1017

1018

1019

1020

expected. Future work should be directed towards incorporating numerical machining models into the

wear monitoring system to normalize the data with respect to machining parameters. If this can be

achieved, the amount of training data required for an effective TCMS will be reduced, which in turn will

provide a better solution to TCM for the manufacturing industry.

References

Al-Habaibeh, A., Gindy, N., and Radwan, N. 2000. An automated approach for monitoring gradual tool

wear in high speed milling of titanium, pp. 371–380. In Proceedings of the 13th International

Congress on Condition Monitoring and Diagnostic Engineering Management (COMADEM 2000),

Houston, TX, 3–8 December.

Araujo, A.J.M.M., Wilcox, S.J., and Reuben, R.L. 2000. Sliding friction as a possible source of acoustic

emission in metal cutting, pp. 381–387. In Proceedings of the 13th International Congress on

Condition Monitoring and Diagnostic Engineering Management (COMADEM 2000), Houston, Tx,

3–8 December.

Axinte, D.A., Belluco, W., and De Chiffre, L., Evaluation of cutting force uncertainty components in

turning, Int. J. Machine Tools Manuf., 41, 719–730, 2001.

Baek, D.K., Ko, T.J., and Kim, H.S., Real time monitoring of tool breakage in a milling operation using a

digital signal processor, J. Mater. Process. Technol., 100, 266–272, 2000.

Bahre, D., Muller, M., and Warnecke, G. 1997. Basic characteristics on cutting effects in correlation to

dynamic effects, pp. 21–26. 1997 Technical Papers of the North American Manufacturing Research

Institution of SME.

Barrios, L.J., Ruiz, A., Guinea, D., Ibanez, A., and Bustos, P., Experimental comparison of sensors for

tool-wear monitoring on milling, Sensors Actuators A, 37–38, 589–595, 1993.

Bayramoglu, M. and Dungel, U., A systematic investigation on the force ratios in tool condition

monitoring for turning operations, Trans. Inst. Measur. Control, 20, 92–97, 1998.

Beynon, M., Curry, B., and Morgan, P., The Dempster–Shafer theory of evidence: an alternative

approach to multicriteria decision modelling, Omega, 28, 37–50, 2000.

Bonifacio, M.E.R. and Diniz, A.E., Correlating tool wear, tool life, surface roughness and tool vibration in

finish turning with coated carbide tools, Wear, 173, 137–144, 1994.

Byrne, G., Dornfeld, D., Inasaki, I., Ketteler, G., Konig, W., and Teti, R., Tool Condition Monitoring

(TCM) — The status of research and industrial application, Ann. CIRP, 44, 541–567, 1995.

Chen, J.C. and Black, J.T., A Fuzzy-Nets-In-Process (FNIP) system for tool breakage monitoring in end-

milling operations, Int. J. Machine Tools Manuf., 37, 783–800, 1997.

Chiou, R.Y. and Liang, S.Y., Analysis of acoustic emission in chatter vibration with tool wear effect in

turning, Int. J. Machine Tools Manuf., 40, 927–941, 2000.

Choi, D., Kwon, W.T., and Chu, C.N., Real-time monitoring of tool fracture in turning using sensor

fusion, Int. J. Adv. Manuf. Technol., 15, 305–310, 1999.

Chungchoo, C. and Saini, D., On-line tool wear estimation in CNC turning operations using fuzzy neural

network model, Int. J. Machine Tools Manuf., 42, 29–40, 2002.

Dan, L. and Mathew, J., Tool wear and failure monitoring techniques for turning — a review, Int.

J. Machine Tools Manuf., 30, 579–598, 1990.

Das, S., Chattopadhyay, A.B., and Murthy, A.S.R., Force parameters for on-line tool wear estimation: a

neural network approach, Neural Networks, 9, 1639–1645, 1996.

Dimla, D.E., Sensor signals for tool-wear monitoring in metal cutting operations — a review of methods,

Int. J. Machine Tools Manuf., 40, 1073–1098, 2000.

Dimla, D.E. and Lister, P.M., On-line metal cutting tool condition monitoring I: force and vibration

analyses, Int. J. Machine Tools Manuf., 40, 739–768, 2000a.

Dimla, D.E. and Lister, P.M., On-line metal cutting tool condition monitoring II: tool-state

classification using multi-layer perceptron neural networks, Int. J. Machine Tools Manuf., 40,

769–781, 2000b.


1021

1022

1023

1024

1025

1026

1027

1028

1029

1030

1031

1032

1033

1034

1035

1036

1037

1038

1039

1040

1041

1042

1043

1044

1045

1046

1047

1048

1049

1050

1051

1052

1053

1054

1055

1056

1057

1058

1059

1060

1061

1062

1063

1064

1065

1066

1067

1068

1069

1070

1071

Dimla, D.E., Lister, P.M., and Leighton, N.J. 1996. Investigation of a single-layer perceptron neural

network to tool wear inception in a metal turning process, pp. 3/1–3/4. In Proceedings of the 1997

IEE Colloquium on Modelling and Signal Processing for Fault Diagnosis.

Du, R., Signal understanding and tool condition monitoring, Eng Appl. Artif. Intell., 12, 585–597, 1999.

Du, R., Liu, Y., Xu, Y., Li, X., Wong, Y.S., and Hong, G.S. 2002. Tool condition monitoring using

transition fuzzy probability. In Metal Cutting and High Speed Machining, pp. 375–392, Kluwer

Academic/Plenum Publishers, New York.

Ertunc, H.M. and Loparo, K.A., A decision fusion algorithm for tool wear condition monitoring in

drilling, Int. J. Mach. Tools Manuf., 41, 1347–1362, 2001.

Ertunc, H.M., Loparo, K.A., and Ocak, H., Tool wear condition monitoring in drilling operations using

Hidden Markov Models (HMMs), Int. J. Mach. Tools Manuf., 41, 1363–1384, 2001.

Fu, P., Hope, A.D., and Javed, M.A., Fuzzy classification of milling tool wear, Insight, 39, 553–557,

1997.

Govekar, E. and Grabec, I., Self-organizing neural network application to drill wear classification, Trans.

ASME: J. Eng. Ind., 116, 233–238, 1994.

Govekar, E., Gradisek, J., and Grabec, I., Analysis of acoustic emission signals and monitoring of

machining processes, Ultrasonics, 38, 598–603, 2000.

James, L.C. and Tzeng, T., Multimilling-insert wear assessment using non-linear virtual sensor, time–

frequency distribution and neural networks, Mech. Syst. Signal Process., 14, 945–957, 2000.

Jemielniak, K., Some aspects of AE application in tool condition monitoring, Ultrasonics, 38, 604–608,

2000.

Jennings, A.D. and Drake, P.R., Machine tool condition monitoring using statistical quality control

charts, Int. J. Mach. Tools Manuf., 37, 1243–1249, 1997.

Jiang, C.Y., Zhang, Y.Z., and Xu, H.J., In-process monitoring of tool wear stage by the frequency band

energy method, Ann. CIRP, 36, 45–48, 1987.

Jun, C. and Suh, S., Statistical tool breakage detection schemes based on vibration signals in NC milling,

Int. J. Mach. Tools Manuf., 39, 1733–1746, 1999.

Kapoor, S.G., DeVor, R.E., and Zhu, R. 1998. Development of mechanistic models for the prediction

of machining performance: Model-building methodology, pp. 109–120. In Proceedings of the

International Workshop on Modelling of Machining Operations, Atlanta, GA, May 19.

Kim, J., Kang, M., Ryu, B., and Ji, Y., Development of an on-line tool-life monitoring system using

acoustic emission signals in gear shaping, Int. J. Mach. Tools Manuf., 39, 1761–1777, 1999.

Koizumi, T., Tsujiuchi, N., and Matsumura, Y., Diagnosis with the correlation integral in the time

domain, Mech. Syst. Signal Process., 14, 1003–1010, 2000.

Kopac, J., Influence of cutting material and coating on tool quality and tool life, J. Mater. Process.

Technol., 78, 95–103, 1998.

Kumar, S.A., Ravindra, H.V., and Srinivasa, Y.G., In-process tool wear monitoring through time series

modeling and pattern recognition, Int. J. Prod. Res., 35, 739–751, 1997.

Kuo, R.J. and Cohen, P.H., Intelligent tool wear estimation system through artificial neural networks and

fuzzy modelling, Artif. Intell. Eng., 12, 229–242, 1998.

Lago, L., Olsson, S., Hakansson, L., and Claesson, I. 2002. Design of an efficient chatter control system for

turning and boring applications, pp. 4–7. In Proceedings of the 20th International Modal Analysis

Conference (IMAC XX), Los Angeles, CA, February.

Lee, B.Y. and Tarng, Y.S., Milling cutter breakage detection by the discrete wavelet transform,

Mechatronics, 9, 225–234, 1999.

Lee, L.C., Lee, K.S., and Gan, C.S., On the correlation between dynamic cutting force and tool wear, Int.

J. Mach. Tools Manuf., 29, 295–303, 1989.

Lee, J.H., Kim, D.E., and Lee, S.J., Statistical analysis of cutting force ratios for flank-wear monitoring,

J. Mater. Process. Technol., 74, 104–114, 1998.

Li, X., A brief review: acoustic emission method for tool wear monitoring during turning, Int. J. Mach.

Tools Manuf., 42, 157–165, 2002.


1072

1073

1074

1075

1076

1077

1078

1079

1080

1081

1082

1083

1084

1085

1086

1087

1088

1089

1090

1091

1092

1093

1094

1095

1096

1097

1098

1099

1100

1101

1102

1103

1104

1105

1106

1107

1108

1109

1110

1111

1112

1113

1114

1115

1116

1117

1118

1119

1120

1121

1122

Li, S. and Elbestawi, M.A., Fuzzy clustering for automated tool condition monitoring in machining,

Mech. Syst. Signal Process., 10, 533–550, 1996.

Li, X.Q., Wong, Y.S., and Nee, A.Y.C., Tool wear and chatter detection using the coherence function of

two crossed accelerations, Int. J. Mach. Tools Manuf., 37, 425–435, 1997.

Liu, Q. and Altintas, Y., On-line monitoring of flank wear in turning with multilayerd feed-forward

neural network, Int. J. Mach. Tools Manuf., 39, 1945–1959, 1999.

Lou, K. and Lin, C., An intelligent sensor fusion system for tool monitoring on a machining centre, Int.

J. Adv. Manuf. Technol., 13, 556–565, 1997.

Luo, G., Osypiw, D., and Irle, M. 2002. Tool wear monitoring by on-line vibration analysis with

wavelet algorithm. In Metal Cutting and High Speed Machining, pp. 393–405, Kluwer Academic/

Plenum Publishers, New York.

Mdlazi, L., Marwala, T., Stander, C.J., Scheffer, C., and Heyns, P.S. 2003. The principal component

analysis and automatic relevance determination for fault identification in structures.

In Proceedings of the 21st International Modal Analysis Conference (IMAC), Kissimmee, FL,

Paper 37.

Merchant, M.E., Mechanics of the cutting process, J. Appl. Phys., 16, 318–324, 1945.

Morimoto, Y., Ichida, Y., and Sata, R. 2000. Excitation technique by 2-axes shaker of an CNC lathe,

pp. 1643–1648. In Proceedings of the 18th International Conference on Modal Analysis, San Antonio,

TX.

Pai, P.S., Nagabhushana, T.N., and Rao, P.K.R., Tool wear estimation using resource allocation network,

Int. J. Mach. Tools Manuf., 41, 673–685, 2001.

Ruiz, A., Guinea, D., Barrios, L.J., and Betancourt, F., An empirical multi-sensor estimimation of tool

wear, Mech. Syst. Signal Process., 7, 105–199, 1993.

Scheffer, C. 1999. Monitoring of tool wear in turning operations using vibration measurements, Masters

dissertation (MEng), Department of Mechanical and Aeronautical Engineering, University of

Pretoria, South Africa.

Scheffer, C. 2002. Development of a tool wear monitoring system for turning using artificial intelligence,

Ph.D. thesis, Department of Mechanical and Aeronautical Engineering, University of Pretoria,

South Africa.

Scheffer, C. and Heyns, P.S. 2000a. Synthetic diamond tool wear monitoring using vibration

measurements, pp. 245–251. In Proceedings of the 18th International Modal Analysis Conference,

San Antonio, TX, 7–10 February.

Scheffer, C. and Heyns, P.S. 2000b. Development of an adaptable tool condition monitoring system,

pp. 361–370. In Proceedings of the 13th International Congress on Condition Monitoring and

Diagnostic Engineering Management (COMADEM 2000), Houston, TX, 3–8 December.

Scheffer, C. and Heyns, P.S. 2001a. Tool condition monitoring systems — an overview, pp. 316–323.

International Conference on Competitive Manufacturing (COMA ’01), Stellenbosch, South Africa,

31 January–2 February.

Scheffer, C. and Heyns, P.S., Wear monitoring in turning operations using vibration and strain

measurements, Mech. Syst. Signal Process., 15, 1185–1202, 2001b.

Scheffer, C. and Heyns, P.S. 2002a. A robust and cost-effective system for conducting cutting experiments

in a production environment, pp. 329–334. In Proceedings of 3rd CIRP International Conference

on Intelligent Computation in Manufacturing Engineering (ICME 2002), Ischia (Naples), Italy,

3–5 July.

Scheffer, C. and Heyns, P.S. 2002b. Neural Network approaches for sensor-based tool wear monitoring, In

Proceedings of Metalworking Tools & Fluids, South African Institute of Tribology, Johannesburg,

South Africa, 7 November.

Scheffer, C. and Heyns, P.S., An effective tool condition monitoring system for turning, Mech. Syst. Signal

Process., 2004.Q6

Scheffer, C., Kratz, H., Heyns, P.S., and Klocke, F., Development of a tool wear monitoring system for

hard turning, Int. J. Mach. Tools Manuf., 43, 973–985, 2003.


1123

1124

1125

1126

1127

1128

1129

1130

1131

1132

1133

1134

1135

1136

1137

1138

1139

1140

1141

1142

1143

1144

1145

1146

1147

1148

1149

1150

1151

1152

1153

1154

1155

1156

1157

1158

1159

1160

1161

1162

1163

1164

1165

1166

1167

1168

1169

1170

1171

1172

1173

Sick, B., On-line tool wear monitoring in turning using neural networks, Neural Comput. Appl., 7,

356–366, 1998.

Sick, B., Online and indirect tool wear monitoring in turning with artificial neural networks: a review of

more than a decade of research, Mech. Syst. Signal Process., 16, 487–546, 2002.

Sick, B. and Sicheneder, A. 1997. Time-delay neural networks for on-line tool wear classification and

estimation in turning, pp. 461–466. In Proceedings of the Third Conference on Neural Networks and

Their Applications, Kule, Poland, 14–18 October.

Sick, B., Sicheneder, A., and Lindinger, H. 1998. A comparative evaluation of different neural network

paradigms for tool wear classification in turning, pp. 139–146. In Proceedings of the 3rd

International Workshop Neural Networks in Applications (NN ’98), University of Magdeburg,

Germany, 12–13 February.

Silva, R.J., Rueben, R.L., Baker, K.J., and Wilcox, S.J., Tool wear monitoring of turning operations by

neural network and expert system classification of a feature set generated form multiple sensors,

Mech. Syst. Signal Process., 12, 319–332, 1998.

Silva, R.J., Baker, K.J., Wilcox, S.J., and Reuben, R.L., The adaptability of a tool wear monitoring system

under changing cutting conditions, Mech. Syst. Signal Process., 14, 287–298, 2000.

Tarmal, G.J. and Opavsky, P. 2000. Signal processing in measurement of milling forces, pp. 389–397. In

Proceedings of the 13th International Congress on Condition Monitoring and Diagnostic Engineering

Management (COMADEM 2000), Houston, TX.

Teti, R., A review of tool condition monitoring literature database, Ann. CIRP, 44, 659–667, 1995.

Venkatesh, K., Zhou, M., and Caudill, R.J., Design of artificial neural networks for tool wear monitoring,

J. Intell. Manuf., 8, 215–226, 1997.

Warnecke, G. and Siems, S. 2002. Dynamics in high speed machining. Metal Cutting and High Speed

Machining, pp. 21–30, Kluwer Academic/Plenum Publishers, New York.

Xiaoli, L., Yingxue, Y., and Zhejun, Y., On-line tool condition neural network with improved fuzzy neural

network, High Technol. Lett., 3, 30–38, 1997.

Yao, Y., Fang, X.D., and Arndt, G., Comprehensive tool wear estimation in finish-machining via

multivariate time-series analysis of 3-D cutting forces, Ann. CIRP, 39, 57–60, 1990.

Zawada-Tomkiewicz, A., Classifying the wear of turning tools with neural networks, J. Mater. Process.

Technol., 109, 300–304, 2001.

Zhou, Q., Hong, G.S., and Rahman, M., A new tool life criterion for tool condition monitoring using a

neural network, Eng. Appl. Artif. Intell., 8, 579–588, 1995.


1174

1175

1176

1177

1178

1179

1180

1181

1182

1183

1184

1185

1186

1187

1188

1189

1190

1191

1192

1193

1194

1195

1196

1197

1198

1199

1200

1201

1202

1203

1204

1205

1206

1207

1208

1209

1210

1211

1212

1213

1214

1215

1216

1217

1218

1219

1220

1221

1222

1223

1224

Author QueriesJOB NUMBER: 8568

CHAPTER: Vibration-Based Tool Condition Monitoring Systems

Q1 Kindly note that Dimla (2001) is not present in the reference list, please add or delete.

Q2 Please check: the percentages only total 80%.

Q3 Kindly note that Bonifacio et al. (1994) has been changed to Bonifacio and Diniz, 1994 as per

the reference list, please check.

Q4 Please check change to "AERMS".

Q5 Kindly note that Li and Tzeng (2000) has been changed to James and Tzeng, 2000 to match with

the reference list, please check.

Q6 Kindly update Scheffer and Heyns, 2004.

Q7 Please check the permissions for figures 4, 17, 18, 19, 21, 23 and 24. Please check if permission is

required. If so, then whether the information is correct.

1225

1226

1227

1228

1229

1230

1231

1232

1233

1234

1235

1236

1237

1238

1239

1240

1241

1242

1243

1244

1245

1246

1247

1248

1249

1250

1251

1252

1253

1254

1255

1256

1257

1258

1259

1260

1261

1262

1263

1264

1265

1266

1267

1268

1269

1270

1271

1272

1273

1274

1275

vibration-based tool condition monitoring systems 2005 scheffer

Documents