CONTENTS
PREFACE = 8
ACKNOWlEDGMENTS = 15
OPERATORSANDNOTATIONALCONVENTIONS = 17
1. INTRODUCTION = 1
1.1 systems = 1
1.2 Models = 3
1.3 The system Identification proceduce = 7
1.4 Organization of the Book = 8
1.5 Bibliography = 10
part Ⅰ : systems and models
2. TIME-INVARIAN TLINEAR SYSTEMS = 13
2.1 Impulse Responses, Disturbances and Transfer Functions = 13
2.2 Frequency-domain Expressions = 22
2.3 Signal Spectra = 26
2.4 Single Realization Behavlor and Ergodicity Results = 34
2.5 Multivariable systems = 35
2.6 summary = 36
2.7 Bibliography = 37
2.8 problems = 38
Appendix 2.A : Proof of The orem 2.2 = 43
Appendix 2.B : Proof of The orem 2.3 = 45
Appendix 2.C : Covariance Formulas = 49
3. SIMULATION, PREDICTION, AND CONTROL = 51
3.1 Simulation = 51
3.2 Ptediction = 52
3.3 Observers = 59
3.4 Control = 62
3.5 = 65
3.6 Bibliography = 65
3.7 problems = 66
4. MODELS OF LINE ARTIME-INVARIANT SYSTEMS = 69
4.1 Linear Models and Sets of Linear Models = 69
4.2 AFamily of Transfer-function Models = 71
4.3 State-space Models = 81
4.4 Distributed-Parameter Models = 90
4.5 Model Sets, Model Structures, and Identifiability : Some Formal Aspects = 93
4.6 Identifiability of SomeModel Structures = 101
4.7 summary = 106
4.8 Bibliography = 106
4.9 problems = 108
Appendix 4.A : Identifiability of Black-box Multivariable Model Structures = 115
5. MOOEL SFORTIME-VARYING AND NONLINEAR SYSTEMS = 127
5.1 Linear Time-varying Models = 127
5.2 Nonlinear Modelsas Linear Regressions = 130
5.3 Nonlinear Models = 132
5.4 Formal Characterization of Models = 134
5.5 summary = 137
5.6 Bibliography = 138
5.7 problems = 138
pact Ⅱ : methods
6. NONPARAMETRICTIME AND FREQUENCY-DOMAIN METHODS = 141
6.1 Transient Response Analysis and Correlation Analysis = 141
6.2 Frequency-response Analysis = 143
6.3 Fourier Analysis = 146
6.4 Spectral Analysis = 151
6.5 Estimating the Disturbance Spectrum = 160
6.6 summary = 162
6.7 Bibliography = 162
6.8 problems = 163
Appendix 6.A : Derivation of the Asymptotic Properties of the spectral Analysis Estimate = 167
7. PARAMETERE STIMATION METHODS = 169
7.1 Guiding Principlesbehind Parameter Estimation Methods = 169
7.2 Prediction Errors = 171
7.3 Linear Regressions and the Least-squares Method = 176
7.4 AStatistical Framework for Parameter Estimation and the Maximum Like lihood Method = 181
7.5 Correlating Prediction Error swith Past Data = 190
7.6 Instrumental-variable Methods = 172
7.7 summary = 195
7.8 Bibliography = 196
7.9 problems = 197
Appendix 7.A : Proofofthecramir-Rao Inequality = 206
8. CONVERGENCE AND CONSISTENCY = 208
8.l Introduction = 208
8.2 Conditions on the Data Set = 210
8.3 Prediction-Error Approach = 214
8.4 Consistency and 1 = 218
8.5 Linear Time-Invariant Models : In Frequency domain Description of the Limit Model = 224
8.6 The Correlation Approach = 229
8.7 summary = 233
8.8 Bibliography = 233
8.9 problems = 234
9. ASYMPTOTICDIS TRIBUTION OF PARAMETER ESTIMATES = 239
9.l Introduction = 239
9.2 Theprediction-Error Approach : Basic Theorem = 240
9.3 Expressions for the Asym Ptotic Variance = 242
9.4 Frequency-domaln Expressions for the Asymptotic Variance = 278
9.5 The Correlation Approach = 254
9.6 Use and Relevance of As Mptotic Yariance Expressions = 258
9.7 summary = 262
9.8 Bibliography = 263
9.9 problems = 264
Appendix 9.A : Proofof The orem 9.1 = 266
Appendix 9.B : The Asym Ptoticp Variance = 270
10. COMPUTING THE ESTIMATE = 274
10.1 Linear Regressions and Least Squares = 274
10.2 Numerical Solutionby Iterative Search Methods = 282
10.3 Com Puting Gradients = 285
10.4 Two-Stage and Multistage Methods = 288
10.5 Local Solutions and Initial Values = 292
10.6 summary = 294
10.7 Bibliogra Phy = 294
10.8 problems = 296
11. RECURSIVE ESTFMATION METHODS = 303
11.1 Introduction = 303
11.2 The Recursivees Algorithm = 305
11.3 The Recursive Ⅳ method = 311
11.4 Recursive prediction-Error Methods = 311
11.5 Recursive Pseudolinear Regressions = 316
11.6 The choice of updating' SteP = 318
11.7 Implementation = 322
11.8 summary = 326
11.9 Bibliography = 327
11.10 Problems = 328
Appendix ll.A : Techniques for Asym Ptotic Analysis of Recursive Algorithms = 329
Part Ⅲ : user's choices
12. OPTIONS AND OBJECTIVES = 339
12.1 Options = 339
12.2 Objectives = 341
12.3 Biasandvariance = 345
12.4 summary = 347
12.5 Bibliography = 347
12.6 problems = 347
13. AFFECTING THE BIASDISTRIBUTION OF TRANSFER-FUNCTION ESTIVATES = 349
13.1 Some Basic Expressions = 349
13.2 Heuristic Dlscussion of Transfer-function Fit In Open-loop Operation = 350
13.3 Some Solutions to Formal Design Problems = 354
13.4 summary = 356
13.5 Bibliography = 356
13.6 Problems = 357
14. EXPERIMENTDESIGN = 358
14.1 Some General Considerations = 359
14.2 I : ve Experiments = 361
14.3 Optimal Input Design = 369
14.4 Optimal Experiment Design for High-order Black-box Models = 375
14.5 Choice of Sampling Interval and Presampling Filters = 378
14.6 Pretreatment of Data = 386
14.7 summary = 389
14.8 Bibliography = 390
14.9 Problems = 391
15. CHOICE OF IDENTIFICATION CRITERION = 394
15.1 General Aspects = 394
15.2 Choice of Norm.Robustness = 396
15.3 Variance : Optimallnstruments = 402
15.4 summary = 405
15.5 Bibliography = 406
15.6 Rroblems = 406
16. MODEL STRUCTURE SELECTION AND MODEL VALIDATION = 408
16.1 General As Pects of the Choice of Model Structure = 408
16.2 Aprioriconsiderations = 411
16.3 Model Structure Selection Based on Preliminary Data Analysis = 413
16.4 Model Structures = 416
16.5 Model Validation = 424
16.6 summary = 430
16.7 Bibliogiaphy = 431
16.8 Problems = 431
17. SYSTEM IDENTIFICATION IN PRACTICE = 434
17.1 The Tool : Interactives of Tware = 434
17.2 Alaboratory - scale Application = 440
17.3 Of ship-Steering Dynamics = 449
17.4 What Does System ldentification Have to Offer? = 454
17.5 Bibliography = 456
APPENDIX Ⅰ : Some Concepts from Probability Theory = 457
APPENDIX Ⅱ : Some Statistical Techniques for Linear = 401
Regressions
REVERENCES = 482
AUTHORINDEX = 505
SUBJECTINDEX = 511
