The current research and development in intelligent control and information processing have been driven increasingly by advancements made from fields outside the traditional control areas, into new frontiers of intelligent control and information processing so as to deal with ever more complex systems with ever growing size of data and complexity.As researches in intelligent control and information processing are taking on ever more complex problems, the control system as a nuclear to coordinate the activity within a system increasingly need to be equipped with the capability to analyze, and reason so as to make decision. This requires the support of cognitive components, and communication protocol to synchronize events within the system to operate in unison.In this review volume, we invited several well-known experts and active researchers from adaptive/approximate dynamic programming, reinforcement learning, machine learning, neural optimal control, networked systems, and cyber-physical systems, online concept drift detection, pattern recognition, to contribute their most recent achievements into the development of intelligent control systems, to share with the readers, how these inclusions helps to enhance the cognitive capability of future control systems in handling complex problems. This review volume encapsulates the state-of-art pioneering works in the development of intelligent control systems. Proposition and evocations of each solution is backed up with evidences from applications, could be used as references for the consideration of decision support and communication components required for today intelligent control systems.
Preface v
1 Dynamic Graphical Games: Online Adaptive 1 (48)
Learning Solutions Using Approximate Dynamic
Programming
Mohammed I. Abouheaf
Frank L. Lewis
1.1 Introduction 2 (2)
1.2 Graphs and Synchronization of 4 (3)
Multi-Agent Dynamical Systems
1.2.1 Graphs 4 (1)
1.2.2 Synchronization and tracking error 5 (2)
dynamics
1.3 Multiple Player Cooperative Games on 7 (14)
Graphs
1.3.1 Graphical games 7 (1)
1.3.2 Comparison of graphical games with 8 (1)
standard dynamic games
1.3.3 Nash equilibrium for graphical games 9 (1)
1.3.4 Hamiltonian equation for dynamic 9 (3)
graphical games
1.3.5 Bellman equation for dynamic 12 (2)
graphical games
1.3.6 Discrete Hamilton-Jacobi theory: 14 (5)
Equivalence of Bellman and discrete-time
Hamilton Jacobi equations
1.3.7 Stability and Nash solution of the 19 (2)
graphical games
1.4 Approximate Dynamic Programming for 21 (12)
Graphical Games
1.4.1 Heuristic dynamic programming for 21 (6)
graphical games
1.4.2 Dual heuristic programming for 27 (6)
graphical games
1.5 Coupled Riccati Recursions 33 (2)
1.6 Graphical Game Solutions by 35 (3)
Actor-Critic Learning
1.6.1 Actor-critic networks and tuning 35 (2)
1.6.2 Actor-critic offline tuning with 37 (1)
exploration
1.6.3 Actor-critic online tuning in 38 (1)
real-time
1.7 Graphical Game Example and Simulation 38 (6)
Results
1.7.1 Riccati recursion offline solution 39 (1)
1.7.2 Simulation results using offline 40 (1)
actor-critic tuning
1.7.3 Simulation results using online 41 (3)
actor-critic tuning
1.8 Conclusions 44 (5)
2 Reinforcement-Learning-Based Online Learning 49 (34)
Control for Discrete-Time Unknown Nonaffine
Nonlinear Systems
Xiong Yang
Derong Liu
Qinglai Wei
Ding Wang
2.1 Introduction 49 (4)
2.2 Problem Statement and Preliminaries 53 (2)
2.2.1 Dynamics of nonaffine nonlinear 53 (1)
discrete-time systems
2.2.2 A single-hidden layer neural network 54 (1)
2.3 Controller Design via Reinforcement 55 (8)
Learning
2.3.1 A basic controller design approach 55 (5)
2.3.2 Critic neural network and weight 60 (1)
update law
2.3.3 Action neural network and weight 61 (2)
update law
2.4 Stability Analysis and Performance of 63 (7)
the Closed-Loop System
2.5 Numerical Examples 70 (6)
2.5.1 Example 1 70 (4)
2.5.2 Example 2 74 (2)
2.6 Conclusions 76 (7)
3 Experimental Studies on Data-Driven Heuristic 83 (24)
Dynamic Programming for POMDP
Zhen Ni
Haibo He
Xiangnan Zhong
3.1 Introduction 83 (3)
3.2 Markov Decision Process and Partially 86 (3)
Observable Markov Decision Process
3.2.1 Markov decision process 86 (2)
3.2.2 Partially observable Markov 88 (1)
decision process
3.3 Problem Formulation with the State 89 (2)
Estimator
3.4 Data-Driven HDP Algorithm for POMDP 91 (3)
3.4.1 Learning in the state estimator 92 (1)
network
3.4.2 Learning in the critic and the 93 (1)
action network
3.5 Simulation Study 94 (6)
3.5.1 Case study one 95 (1)
3.5.2 Case study two 96 (2)
3.5.3 Case study three 98 (2)
3.6 Conclusions and Discussion 100(7)
4 Online Reinforcement Learning for 107(26)
Continuous-State Systems
Yuanheng Zhu
Dongbin Zhao
4.1 Introduction 107(2)
4.2 Background of Reinforcement Learning 109(1)
4.3 RLSPI Algorithm 110(4)
4.3.1 Policy iteration 110(1)
4.3.2 RLSPI 111(3)
4.4 Examples of RLSPI 114(6)
4.4.1 Linear discrete-time system 114(4)
4.4.2 Nonlinear discrete-time system 118(2)
4.5 MSEC Algorithm 120(4)
4.5.1 State aggregation 121(1)
4.5.2 MSEC 121(3)
4.6 Examples of MSEC 124(5)
4.6.1 Inverted pendulum 125(1)
4.6.2 Results of MSEC 126(3)
4.7 Conclusions 129(4)
5 Adaptive Iterative Learning Control of Robot 133(18)
Manipulators in the Presence of Environmental
Constraints
Xiongxiong He
Zhenhua Qin
Xianqing Wu
5.1 Introduction 133(2)
5.2 Problem Statement and Preliminaries 135(5)
5.2.1 Dynamics of constrained robot 135(5)
systems
5.3 Adaptive Iterative Learning Control 140(3)
Design
5.3.1 The dynamic model conversion and 140(2)
assumption
5.3.2 Adaptive iterative learning control 142(1)
design
5.4 Stability and Convergence Analysis of 143(2)
the Closed-Loop System
5.5 Numerical Examples 145(4)
5.6 Conclusions 149(2)
6 Neural Network Control of Nonlinear 151(42)
Discrete-Time Systems in Affine Form in the
Presence of Communication Network
Hao Xu
Avimanyu Sahoo
Sarangapani Jagannathan
6.1 Introduction 151(3)
6.2 NNCS Background and Traditional Optimal 154(3)
Control
6.2.1 Nonlinear networked control systems 154(1)
representation
6.2.2 Traditional stochastic optimal 155(2)
control
6.3 Stochastic Optimal Controller Design 157(14)
for NNCS
6.3.1 Online NN-identifier design 157(3)
6.3.2 Stochastic value function setup and 160(3)
critic NN design
6.3.3 Action NN estimation of optimal 163(1)
control input
6.3.4 Closed-loop stability 164(2)
6.3.5 Simulation results 166(5)
6.4 Extension to Infinite Horizon 171(7)
6.4.1 Approximation of the optimal 171(4)
stochastic value function and control
policy using NN
6.4.2 Simulation results 175(3)
6.5 Event-Triggered Control of NNCS 178(7)
6.5.1 Event-triggered control design of 178(4)
uncertain continuous time nonlinear system
6.5.2 Closed-loop system stability and 182(2)
event condition design
6.5.3 Simulation results 184(1)
6.6 Conclusions 185(1)
6.7 Appendix 186(7)
6.7.1 The proof of Theorem 6.3 186(1)
6.7.2 The proof of Theorem 6.4 187(6)
7 Nonlinear and Robust Model Predictive Control 193(28)
of Systems with Unmodeled Dynamics Based on
Supervised Learning and Neurodynamic
Optimization
Zheng Yan
Jun Wang
7.1 Introduction 193(3)
7.2 Problem Formulation 196(7)
7.2.1 Nominal stability 197(1)
7.2.2 Quadratic programming formulation 197(3)
7.2.3 Minimax problem formulation 200(3)
7.3 Neural Network Approaches 203(3)
7.3.1 Extreme learning machine 203(1)
7.3.2 Simplified dual network 204(1)
7.3.3 A two-layer recurrent neural network 205(1)
7.3.4 Overall MPC approach 205(1)
7.4 Simulation Results 206(9)
7.4.1 Example 1 206(5)
7.4.2 Example 2 211(4)
7.5 Conclusions 215(6)
8 Packet-Based Communication and Control 221(24)
Co-Design for Networked Control Systems
Yun-Bo Zhao
Guo-Ping Liu
8.1 Introduction 221(3)
8.2 Packet-Based Transmission in Networked 224(2)
Control Systems
8.3 Packet-Based Control for Networked 226(4)
Control Systems
8.3.1 Packet-based control for NCSs: A 226(2)
unified model
8.3.2 Design of the packet-based control 228(2)
scheme
8.4 Stability of Packet-Based Networked 230(4)
Control Systems
8.4.1 A switched system theory approach 231(1)
8.4.2 A delay dependent analysis approach 232(2)
8.5 Packet-Based Controller Design: A 234(2)
GPC-Based Approach
8.6 Numerical and Experimental Examples 236(4)
8.6.1 Numerical examples 236(2)
8.6.2 Experimental example 238(2)
8.7 Conclusions 240(5)
9 Review of Some Approximate Privacy Measures 245(18)
of Multi-Agent Communication Protocols
Bhaskar DasGupta
Venkatakumar Srinivasan
9.1 Introduction 245(4)
9.1.1 Perfect vs. approximately perfect 248(1)
privacy
9.1.2 Privacy analysis in other 248(1)
environments
9.2 Various Frameworks to Quantify Privacy 249(2)
of Protocols
9.2.1 Communication complexity based 249(1)
approaches
9.2.2 Information-theoretic approaches 249(1)
9.2.3 Cryptographic approaches 250(1)
9.2.4 Two-agent differential privacy 250(1)
framework
9.3 Benchmark Problems and Functions 251(1)
9.4 Examples of Standard Communication 252(2)
Protocols
9.5 A Geometric Approach to Quantify Privacy 254(6)
9.5.1 Tiling functions and dissection 258(2)
protocols
9.5.2 Generalization for d > 2 agents 260(1)
9.6 Conclusions 260(3)
10 Encoding-Decoding Machines for Online 263(20)
Concept-Drift Detection on Datastreams
Cesare Alippi
Giacomo Boracchi
Li Bu
Dongbin Zhao
10.1 Introduction 263(3)
10.2 A CDT Based on the Encoding-Decoding 266(2)
Machine
10.2.1 The observation model 266(1)
10.2.2 The recurrent encoding-decoding 267(1)
machine
10.2.3 The encoding-decoding CDT 267(1)
10.3 An Encoding-Decoding Machine Based on 268(3)
SVD
10.3.1 The SVD-based encoding-decoding 268(1)
machine
10.3.2 The residuals 269(1)
10.3.3 Learning M0 270(1)
10.3.4 Extensions 271(1)
10.4 An ICI-Based Encoding-Decoding CDT 271(3)
10.4.1 Designing the encoding-decoding CDT 271(2)
10.4.2 Adaptation 273(1)
10.4.3 Single vs. multiple sensors 274(1)
10.5 Experiments 274(6)
10.5.1 The "ARMA" dataset 276(1)
10.5.2 The "hairdryer" application 277(1)
10.5.3 The mountain temperature 277(3)
application
10.6 Conclusions 280(3)
11 Recognizing sEMG Patterns for Interacting 283(26)
with Prosthetic Manipulation
Zhaojie Ju
Gaoxiang Ouyang
Marzena Wilamowska-Korsak
Honghai Liu
11.1 Introduction 283(3)
11.2 Recurrence Plot and Quantification 286(2)
Analysis
11.3 Nonlinear Recognition Method 288(6)
11.3.1 Gaussian mixture models 288(3)
11.3.2 Fuzzy Gaussian mixture models 291(2)
11.3.3 Recognizing sEMG signals using 293(1)
FGMMs
11.4 Experimental Results 294(9)
11.4.1 Data collection and description 294(2)
11.4.2 Feature extraction and parameter 296(1)
setting
11.4.3 Recognition with one single feature 297(1)
11.4.4 Recognition with multiple features 298(2)
11.4.5 Recognition with different 300(3)
classifiers
11.5 Conclusions 303(6)
12 Energy Demand Management Through Uncertain 309(26)
Data Forecasting: A Hybrid Approach
Marco Severini
Stefano Squartini
Francesco Piazza
12.1 Introduction 309(2)
12.2 Home Energy Management Problem: The 311(1)
Model
12.3 Modelling the Thermal Optimization 312(2)
Sub-Problem
12.3.1 Notations 312(1)
12.3.2 Heat pump constraints 313(1)
12.4 Neural Networks for Uncertain Data 314(2)
Forecasting
12.5 Optimization Algorithms 316(1)
12.6 Smart Home Energy Management: Case 317(14)
Studies and Simulation Results
12.6.1 Thermal model characterization 317(2)
12.6.2 Task scheduling and energy cost 319(1)
accounting
12.6.3 Solar production 320(1)
12.6.4 Data forecasting 320(1)
12.6.5 Energy management performance 321(10)
12.7 Conclusions 331(4)
13 Many-Objective Evolutionary Algorithms and 335(30)
Hybrid Performance Metrics
Zhenan He
Gary G. Yen
13.1 Introduction 335(5)
13.2 Literature Review on Performance 340(4)
Metrics
13.2.1 Metrics assessing the number of 340(1)
Pareto optimal solutions in the set
13.2.2 Metrics measuring the closeness of 341(1)
the solutions to the true Pareto front
13.2.3 Metrics focusing on distribution 342(1)
of the solutions
13.2.4 Metrics concerning spread of the 343(1)
solutions
13.2.5 Metrics considering both closeness 344(1)
and diversity
13.3 Literature Review on Many-Objective 344(2)
Evolutionary Algorithms
13.4 Performance Metrics Ensemble 346(6)
13.4.1 The proposed framework 346(1)
13.4.2 Double elimination tournament 347(5)
13.5 Experimental Results 352(8)
13.5.1 Selected benchmark test problems 352(2)
13.5.2 Selected performance metrics 354(1)
13.5.3 Parameter setting in experiment 354(1)
13.5.4 Experiment results 354(4)
13.5.5 Observations and insights 358(2)
13.6 Conclusions 360(5)
14 Synchronization Control of Memristive 365(20)
Chaotic Circuits and Their Applications in
Image Encryptions
Shiping Wen
Zhigang Zeng
14.1 Introduction 365(1)
14.2 Synchronization of Memristive Chua's 366(9)
Circuits
14.2.1 Memristive Chua's circuits 367(1)
14.2.2 Adaptive synchronization control 368(3)
of memristive Chua's circuits
14.2.3 Numerical simulations 371(3)
14.2.4 Conclusions 374(1)
14.3 Synchronization of Memristive Lorenz 375(10)
Circuits
14.3.1 Modeling and fuzzy synchronization 375(5)
of memristive Lorenz circuits
14.3.2 Synchronization of fuzzy 380(1)
memristive Lorenz circuits with
memristive Chua's circuits
14.3.3 Simulation results 381(2)
14.3.4 Conclusions 383(2)
15 Graph Embedded Total Margin Twin Support 385(22)
Vector Machine and Its Applications
Xiaobo Chen
Jian Yang
15.1 Introduction 385(3)
15.2 A Brief Review on TSVM 388(2)
15.3 Graph Embedded Total Margin TSVM 390(6)
15.3.1 Model formulation of GTM-TSVM 390(2)
15.3.2 Algorithm derivation 392(2)
15.3.3 Weighting factors for GTM-TSVM 394(2)
15.4 Experimental Results and Analysis 396(8)
15.4.1 Experimental specification 396(1)
15.4.2 A toy example 397(1)
15.4.3 Evaluations on UCI benchmark data 397(5)
sets
15.4.4 Experiments on USPS and MIT CBCL 402(2)
database
15.5 Conclusions 404(3)
16 Regularized Covariance Matrix Estimation 407(24)
Based on MDL Principle
Xiuling Zhou
Ping Guo
C.L. Philip Chen
16.1 Introduction 407(2)
16.2 Theoretical Backgrounds 409(5)
16.2.1 Gaussian mixture model classifier 409(1)
16.2.2 Classical discriminant analysis 410(2)
16.2.3 KLIM method 412(2)
16.3 Covariance Matrix Estimation with 414(8)
Multi-Regularization Parameters
16.3.1 KLIM_L method with 414(1)
multi-regularization parameters
16.3.2 Multi-regularization parameters 415(3)
estimation
16.3.3 Comparison of KLIM_L with 418(1)
regularization methods
16.3.4 Experiments and discussions 418(4)
16.4 Covariance Matrix Estimation with 422(6)
Variable Regularization Parameters
16.4.1 Covariance matrix estimation with 422(1)
variable regularization parameters
16.4.2 Variable regularization matrix 423(2)
estimation
16.4.3 Comparisons 425(1)
16.4.4 Experiments and discussions 426(2)
16.5 Conclusions 428(3)
17 An Evolution of Evolutionary Algorithms with 431(26)
Big Data
Weian Guo
Lei Wang
Qidi Wu
17.1 Introduction 431(3)
17.2 Design of Hyper-Heuristic Framework 434(8)
for Evolutionary Algorithms
17.2.1 Genetic algorithm 434(1)
17.2.2 Particle swarm optimization 435(1)
17.2.3 Biogeography-based optimization 435(3)
17.2.4 Design of framework of 438(1)
hyper-heuristic algorithm
17.2.5 Simulations for the novel 439(3)
framework of evolutionary algorithms
17.3 Performances Analysis of Evolutionary 442(4)
Algorithms for High-Dimensional Benchmarks
17.3.1 Uniform blended migration operator 442(1)
17.3.2 Heuristic migration operator 443(1)
17.3.3 Extended migration operator 443(1)
17.3.4 Numerical simulations for high 444(2)
dimensional benchmarks
17.4 Correlation Analysis for Business 446(8)
Intelligence Based on Evolutionary
Algorithms
17.4.1 Correlation model 451(1)
17.4.2 Simulation for correlation analysis 452(2)
17.5 Conclusions 454(3)
Index 457
新书报道