作者简介:
王乐一 (Le Yi Wang),1990年于加拿大蒙特利尔麦吉尔大学电气工程系获得博士学位。自1990起在美国密西根州韦恩州立大学电气与计算机工程系任教,现任教授,IEEE Fellow。主要研究方向为系统复杂性及信息、系统辨识、鲁棒控制、H∞时变系统、自适应系统、混合与非线性系统等。
殷刚(G. George Yin),1987年博士毕业于Brown University数学系。现任美国密西根州韦恩州立大学教授,IEEE Fellow。研究方向包括随机控制、系统辩识与估计、信号过程、通讯网络、应用概率和随机过程、随机近似与优化、奇异摄动、随机系统的数值解和随机系统理论等。
内容简介:
This book presents recently developed methodologies that utilize quantized information in system identification and explores their potential in extending control capabilities for systems with limited sensor information or networked systems. The results of these methodologies can be applied to signal processing and control design of communication and computer networks, sensor networks, mobile agents, coordinated data fusion, remote sensing, telemedicine, and other fields in which noise-corrupted quantized data need to be processed.
Providing a comprehensive coverage of quantized identification, the book treats linear and nonlinear systems, as well as time-invariant and time-varying systems. The authors examine independent and dependent noises, stochastic- and deterministic-bounded noises, and also noises with unknown distribution functions. The key methodologies combine empirical measures and information-theoretic approaches to derive identification algorithms, provide convergence and convergence speed, establish efficiency of estimation, and explore input design, threshold selection and adaptation, and complexity analysis.
System Identification with Quantized Observations is an excellent resource for graduate students, systems theorists, control engineers, applied mathematicians, as well as practitioners who use identification algorithms in their work. Selected material from the book may be used in graduate-level courses on system identification.
英文目录:
Preface
Conventions
Glossary of Symbols
Part I: Overview
1 Introduction
1.1 Motivating Examples
1.2 System Identification with Quantized Observations
1.3 Outline of the Book
2 System Settings
2.1 Basic Systems
2.2 Quantized Output Observations
2.3 Inputs
2.4 System Configurations
2.4.1 Filtering and Feedback Configurations
2.4.2 Systems with Communication Channels
2.5 Uncertainties
2.5.1 System Uncertainties: Unmodeled Dynamics
2.5.2 System Uncertainties: Function Mismatch
2.5.3 Sensor Bias and Drifts
2.5.4 Noise
2.5.5 Unknown Noise Characteristics
2.5.6 Communication Channel Uncertainties
2.6 Notes
Part II: Stochastic Methods for Linear Systems
3 Empirical-Measure-Based Identification
3.1 An Overview of Empirical-Measure-Based Identification
3.2 Empirical Measures and Identification Algorithms
3.3 Strong Convergence
3.4 Asymptotic Distributions
3.5 Mean-Square Convergence
3.6 Convergence under Dependent Noise
3.7 Proofs of Two Propositions
3.8 Notes
4 Estimation Error Bounds: Including Unmodeled Dynamics
4.1 Worst-Case Probabilistic Errors and Time Complexity
4.2 Upper Bounds on Estimation Errors and Time Complexity
4.3 Lower Bounds on Estimation Errors
4.4 Notes
5 Rational Systems
5.1 Preliminaries
5.2 Estimation of xk
5.3 Estimation of Parameter θ
5.3.1 Parameter Identifiability
5.3.2 Identification Algorithms and Convergence Analysis
5.4 Notes
6 Quantized Identification and Asymptotic Efficiency
6.1 Basic Algorithms and Convergence
6.2 Quasi-Convex Combination Estimators (QCCE)
6.3 Alternative Covariance Expressions of Optimal QCCEs
6.4 Cramér-Rao Lower Bounds and Asymptotic Efficiency of the Optimal QCCE
6.5 Notes
7 Input Design for Identification in Connected Systems
7.1 Invariance of Input Periodicity and Rank in Open- and Closed-Loop Configurations
7.2 Periodic Dithers
7.3 Sufficient Richness Conditions under Input Noise
7.4 Actuator Noise
7.5 Notes
8 Identification of Sensor Thresholds and Noise Distribution Functions
8.1 Identification of Unknown Thresholds
8.1.1 Sufficient Richness Conditions
8.1.2 Recursive Algorithms
8.2 Parameterized Distribution Functions
8.3 Joint Identification Problems
8.4 Richness Conditions for Joint Identification
8.5 Algorithms for Identifying System Parameters and Distribution Functions
8.6 Convergence Analysis
8.7 Recursive Algorithms
8.7.1 Recursive Schemes
8.7.2 Asymptotic Properties of Recursive Algorithm (8.14)
8.8 Algorithm Flowcharts
8.9 Illustrative Examples
8.10 Notes
Part III: Deterministic Methods for Linear Systems
9 Worst-Case Identification
9.1 Worst-Case Uncertainty Measures
9.2 Lower Bounds on Identification Errors and Time Complexity
9.3 Upper Bounds on Time Complexity
9.4 Identification of Gains
9.5 Identification Using Combined Deterministic and Stochastic Methods
9.5.1 Identifiability Conditions and Properties under Deterministic and Stochastic Frameworks
9.5.2 Combined Deterministic and Stochastic Identification Methods
9.5.3 Optimal Input Design and Convergence Speed under Typical Distributions
9.6 Notes
10 Worst-Case Identification Using Quantized Observations
10.1 Worst-Case Identification with Quantized Observations
10.2 Input Design for Parameter Decoupling
10.3 Identification of Single-Parameter Systems
10.3.1 General Quantization
10.3.2 Uniform Quantization
10.4 Time Complexity
10.5 Examples
10.6 Notes
Part IV: Identification of Nonlinear and Switching Systems
11 Identification of Wiener Systems
11.1 Wiener Systems
11.2 Basic Input Design and Core Identification Problems
11.3 Properties of Inputs and Systems
11.4 Identification Algorithms
11.5 Asymptotic Efficiency of the Core Identification Algorithms
11.6 Recursive Algorithms and Convergence
11.7 Examples
11.8 Notes
12 Identification of Hammerstein Systems
12.1 Problem Formulation
12.2 Input Design and Strong-Full-Rank Signals
12.3 Estimates of ζ with Individual Thresholds
12.4 Quasi-Convex Combination Estimators of ζ
12.5 Estimation of System Parameters
12.6 Examples
12.7 Notes
13 Systems with Markovian Parameters
13.1 Markov Switching Systems with Binary Observations
13.2 Wonham-Type Filters
13.3 Tracking: Mean-Square Criteria
13.4 Tracking Infrequently Switching Systems: MAP Methods
13.5 Tracking Fast-Switching Systems
13.5.1 Long-Run Average Behavior
13.5.2 Empirical Measure-Based Estimators
13.5.3 Estimation Errors on Empirical Measures: Upper and Lower Bounds
13.6 Notes
Part V: Complexity Analysis
14 Complexities, Threshold Selection, Adaptation
14.1 Space and Time Complexities
14.2 Binary Sensor Threshold Selection and Input Design
14.3 Worst-Case Optimal Threshold Design
14.4 Threshold Adaptation
14.5 Quantized Sensors and Optimal Resource Allocation
14.6 Discussions on Space and Time Complexity
14.7 Notes
15 Impact of Communication Channels
15.1 Identification with Communication Channels
15.2 Monotonicity of Fisher Information
15.3 Fisher Information Ratio of Communication Channels
15.4 Vector-Valued Parameters
15.5 Relationship to Shannon's Mutual Information
15.6 Tradeoff between Time Information and Space Information
15.7 Interconnections of Communication Channels
15.8 Notes
A Background Materials
A.1 Martingales
A.2 Markov Chains
A.3 Weak Convergence
A.4 Miscellany
References
Index