상세정보

This is an exploration of physical modelling and experimental issues that considers identification of structured models such as continuous-time linear systems, multidimensional systems and nonlinear systems. It avoids a narrow focus on time series analysis and gives a broader perspective on modelling, identification and its applications with many examples of physical modelling. There is also strong emphasis on model validation and a pluralistic approach with model evaluation in a stochastic framework, according to approximation notions.

정보제공 :

CONTENTS
Preface = xi
1. Introduction = 1
 1.1 Why models? = 1
 1.2 Modeling = 3
 1.3 The purpose of identification = 6
 1.4 Systems and model complexity = 7
 1.5 The procedure of identification = 10
 1.6 Historical remarks and bibliography = 13
2. Black box models = 16
 2.1 Introduction = 16
 2.2 Transient response analysis = 17
 2.3 Frequency response analysis = 21
 2.4 Application of frequency response analysis = 29
 2.5 Summary = 30
 2.6 Exercises = 31
3. Signals and systems = 34
 3.1 Introduction = 34
 3.2 Time-domain and frequency-domain transforms = 35
 3.3 Discretized data = 36
 3.4 The z-transform = 38
 3.5 Finite measurement time = 39
 3.6 The transfer function = 43
 3.7 Signal power and energy = 45
 3.8 Spectra and covariance functions = 46
 3.9 Correlation and coherence = 47
 3.10 Statistical characterization of disturbances = 49
 3.11 Exercises = 51
4. Spectrum analysis = 53
 4.1 The discrete Fourier transform = 54
 4.2 Power spectrum estimation = 56
 4.3 Spectral leakage and windowing = 58
 4.4 Transfer function estimation = 66
 4.5 Smoothing of spectra = 69
 4.6 Covariance estimates and 'correlation analysis' = 71
 4.7 Historical remarks = 72
 4.8 Bibliography = 72
 4.9 Exercises = 73
5. Linear regression = 74
 5.1 Introduction = 74
 5.2 Least-squares estimation = 78
 5.3 Optimal linear unbiased estimators = 88
 5.4 Linear regression in the frequency domain = 90
 5.5 Least-squares estimation with linear constraints = 93
 5.6 *A geometrical interpretation = 95
 5.7 *Multivariable system identification = 97
 5.8 Concluding remarks = 100
 5.9 Historical remarks and references = 101
 5.10 Exercises = 102
6. Identification of time-series models = 104
 6.1 Introduction = 104
 6.2 Model structures = 105
 6.3 Maximum-likelihood identification = 113
 6.4 Kalman filter = 120
 6.5 Instrumental variable method = 121
 6.6 Some aspects of application = 125
 6.7 Some remarks on convergence and consistency = 129
 6.8 Concluding remarks = 131
 6.9 Bibliography and references = 131
 6.10 Exercises = 133
7. Modeling = 136
 7.1 Introduction = 136
 7.2 Mechanical systems = 137
 7.3 Thermodynamic modeling = 141
 7.4 Compartment models = 143
 7.5 Principles of modeling = 146
 7.6 Physical parametrizations = 149
 7.7 Network models = 152
 7.8 Historical and bibliographical remarks = 160
 7.9 Exercises = 162
8. The experimental procedure = 167
 8.1 Introduction = 167
 8.2 The experimental condition = 168
 8.3 Identification and closed-loop control = 168
 8.4 Direct or indirect identification? = 172
 8.5 Choice of input = 173
 8.6 Parameter uncertainty and control = 180
 8.7 Planning and operation of experiments = 183
 8.8 Bibliography and references = 186
 8.9 Exercises = 187
9. Model validation = 192
 9.1 Introduction = 192
 9.2 Method prerequisites = 194
 9.3 Model order determination = 200
 9.4 Residual tests = 207
 9.5 Model and parameter accuracy = 217
 9.6 Classification with the Fisher linear discriminant = 221
 9.7 *The concept 'identifiability' = 224
 9.8 Concluding remarks = 226
 9.9 Bibliography and references = 226
 9.10 Exercises = 227
10. Model approximation = 228
 10.1 Introduction = 228
 10.2 Balanced realization and model reduction = 231
 10.3 Continued fraction approximation = 239
 10.4 Moment matching = 243
 10.5 The Pad$$\acute e$$ approximation = 245
 10.6 Describing function analysis = 246
 10.7 Balanced model reduction in identification = 252
 10.8 Bibliography and references = 256
 10.9 Exercises = 258
11. Real-time identification = 260
 11.1 Introduction = 260
 11.2 Recursive least-squares identification = 263
 11.3 Recursive instrumental variable methods = 269
 11.4 Pseudolinear regression = 270
 11.5 Stochastic gradient methods = 272
 11.6 The Levinson-Durbin algorithm = 274
 11.7 Spectral properties = 278
 11.8 Bibliography and references = 278
 11.9 Exercises = 279
12. Continuous-time models = 280
 12.1 Introduction = 280
 12.2 Outline of the method = 281
 12.3 Model transformation = 283
 12.4 A noise model = 291
 12.5 Identification = 293
 12.6 Convergence and consistency = 296
 12.7 *State-space transformations = 299
 12.8 Signal processing filters = 303
 12.9 Concluding remarks = 306
 12.10 Bibliography and references = 308
  Appendix 12.1 - The Cram$$\acute e$$r-Rao lower bound = 309
  Appendix 12.2 - The Hessian Matrix = 310
  Appendix 12.3 - Proof of Theorem 12.1 = 311
 12.11 Exercises = 314
13. Multidimensional identification = 316
 13.1 Introduction = 316
 13.2 Two-dimensional transforms = 318
 13.3 Two-dimensional system analysis = 319
 13.4 Stability = 320
 13.5 Delay-differential systems = 326
 13.6 Two-dimensional spectra = 330
 13.7 Bibliography and references = 335
14. Nonlinear system identification = 336
 14.1 Introduction = 336
 14.2 Wiener models = 338
 14.3 Volterra-Wiener models = 340
 14.4 Power series expansions = 346
 14.5 Discussion and conclusions = 358
 14.6 References = 358
 14.7 Exercises = 360
15. Adaptive systems = 361
 15.1 Introduction = 361
 15.2 Heuristic control methods = 364
 15.3 Aspects on neural networks = 368
 15.4 Extremum control = 371
 15.5 Model-reference adaptive control = 374
 15.6 Multivariable direct adaptive control = 388
 15.7 Discussion and conclusions = 394
 15.8 Bibliography and references = 396
 Appendix 15.1 = 398
A. Appendix : Basic matrix algebra = 402
 A.1 Preliminaries = 402
 A.2 Matrix norms = 409
 A.3 Singular value decomposition = 409
 A.4 QR-factorization = 411
 A.5 Matrix differentiation = 413
 A.6 Polynomials and polynomial matrices = 415
 A.7 Bibliography and references = 417
B. Appendix : Statistical inference = 418
 B.1 Preliminaries = 418
 B.2 Convergence and consistency = 420
 B.3 Some important probability distributions = 424
 B.4 Conditional expectation = 428
 B.5 Statistical hypothesis testing = 429
 B.6 The Cochran theorem = 432
 B.7 References = 434
C. Appendix : Numerical optimization = 435
 C.1 Introduction = 435
 C.2 Descent methods = 436
 C.3 Newton methods = 437
 C.4 Quasi-Newton methods = 437
 C.5 Conjugate gradient methods = 438
 C.6 Direct search methods = 440
 C.7 Parametric optimization = 440
 C.8 Bibliography and references = 445
D. Appendix : Statistical properties of time series = 447
 D.1 Introduction = 447
 D.2 Stochastic processes = 450
 D.3 Difference equations = 453
 D.4 Autoregressive moving average models = 460
 D.5 Sample covariance functions and spectra = 462
 D.6 Nonstationary stochastic models = 464
 D.7 Prediction and reconstruction = 466
 D.8 The Kalman filter = 468
 D.9 Bibliography and references = 470
E. Appendix : A case study = 472
 E.1 Introduction = 472
 E.2 Summary = 473
 E.3 Methods and materials = 474
 E.4 Modeling of posture control = 474
 E.5 Forces on the platform = 478
 E.6 A dynamic response classification = 479
 E.7 Experiments = 479
 E.8 Results of the experiments = 481
 E.9 Discussion = 483
 E.10 Conclusions = 487
  Appendix E.1 - Transfer function = 488
  Appendix E.2 - Force balances = 489
  Appendix E.3 - Calculations and analysis = 491
  Bibliography and references = 497
Index = 501