Sandou Metaheuristic Optimization for the Design of Automatic Control Laws

1. Auflage 2013
ISBN: 978-1-118-79648-1
Verlag: John Wiley & Sons
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)

E-Book, Englisch, 144 Seiten, E-Book

The classic approach in Automatic Control relies on the use ofsimplified models of the systems and reformulations of thespecifications. In this framework, the control law can be computedusing deterministic algorithms. However, this approach fails whenthe system is too complex for its model to be sufficientlysimplified, when the designer has many constraints to take intoaccount, or when the goal is not only to design a control but alsoto optimize it. This book presents a new trend in Automatic Controlwith the use of metaheuristic algorithms. These kinds of algorithmcan optimize any criterion and constraint, and therefore do notneed such simplifications and reformulations.
The first chapter outlines the author's main motivations forthe approach which he proposes, and presents the advantages whichit offers. In Chapter 2, he deals with the problem of systemidentification. The third and fourth chapters are the core of thebook where the design and optimization of control law, using themetaheuristic method (particle swarm optimization), is given. Theproposed approach is presented along with real-life experiments,proving the efficiency of the methodology. Finally, in Chapter 5,the author proposes solving the problem of predictive control ofhybrid systems.
Contents
1. Introduction and Motivations.
2. Symbolic Regression.
3. PID Design Using Particle Swarm Optimization.
4. Tuning and Optimization of H-infinity Control Laws.
5. Predictive Control of Hybrid Systems.
About the Authors
Guillaume Sandou is Professor in the Automatic Department ofSupélec, in Gif Sur Yvette, France. He has had 12 books, 8journal papers and 1 patent published, and has written papers for32 international conferences.His main research interests includemodeling, optimization and control of industrial systems;optimization and metaheuristics for Automatic Control; andconstrained control.

Autoren/Hrsg.

Sandou, Guillaume

Inhaltsverzeichnis

PREFACE ix
CHAPTER 1. INTRODUCTION AND MOTIVATIONS 1
1.1. Introduction: automatic control and optimization 1
1.2. Motivations to use metaheuristic algorithms 3
1.3. Organization of the book 5
CHAPTER 2. SYMBOLIC REGRESSION 7
2.1. Identification problematic and brief state of the art 7
2.2. Problem statement and modeling 10
2.2.1. Problem statement 10
2.2.2. Problem modeling 10
2.3. Ant colony optimization 13
2.3.1. Ant colony social behavior 13
2.3.2. Ant colony optimization 14
2.3.3. Ant colony for the identification of nonlinear functionswith unknown structure 16
2.4. Numerical results 18
2.4.1. Parameter settings 18
2.4.2. Experimental results 19
2.5. Discussion 22
2.5.1. Considering real variables 22
2.5.2. Local minima 22
2.5.3. Identification of nonlinear dynamical systems 23
2.6. A note on genetic algorithms for symbolic regression 23
2.7. Conclusions 25
CHAPTER 3. PID DESIGN USING PARTICLE SWARMOPTIMIZATION 27
3.1. Introduction 27
3.2. Controller tuning: a hard optimization problem 29
3.2.1. Problem framework 29
3.2.2. Expressions of time domain specifications 30
3.2.3. Expressions of frequency domain specifications 32
3.2.4. Analysis of the optimization problem 35
3.3. Particle swarm optimization implementation 35
3.4. PID tuning optimization 37
3.4.1. Case study: magnetic levitation 37
3.4.2. Time response optimization 39
3.4.3. Time response optimization with penalization on thecontrol input 41
3.4.4. Time response optimization with penalization on thecontrol input and constraint on module margin 42
3.5. PID multiobjective optimization 43
3.6. Conclusions 48
CHAPTER 4. TUNING AND OPTIMIZATION OF H infinity CONTROLLAWS 51
4.1. Introduction 51
4.2. H infinity synthesis 54
4.2.1. Full-order H infinity synthesis 54
4.2.2. Tuning the filters as an optimization problem 57
4.2.3. Reduced-order H infinity synthesis 58
4.3. Application to the control of a pendulum in thecart 60
4.3.1. Case study 60
4.3.2. H infinity synthesis schemes 64
4.3.3. Optimization of the parameters of the filters 66
4.3.4. Reduced-order H infinity synthesis: one DOF case 70
4.3.5. Reduced-order H infinity synthesis: three DOF case 71
4.3.6. Conclusions 76
4.4. Static output feedback design 77
4.5. Industrial examples 82
4.5.1. Mold level control in continuous casting 83
4.5.2. Linear parameter varying control of a missile 83
4.5.3. Internal combustion engine air path control 86
4.5.4. Inertial line-of-sight stabilization 86
4.6. Conclusions 87
CHAPTER 5. PREDICTIVE CONTROL OF HYBRIDSYSTEMS 89
5.1. Problematic 89
5.2. Predictive control of power systems 92
5.2.1. Open-loop control and unit commitment 92
5.2.2. Closed-loop control 94
5.3. Optimization procedure 96
5.3.1. Classical optimization methods for unitcommitment 96
5.3.2. General synopsis of the optimizationprocedure 97
5.3.3. Ant colony optimization for the unitcommitment 98
5.3.4. Computation of real variables 100
5.3.5. Feasibility criterion 101
5.3.6. Knowledge-based genetic algorithm 102
5.4. Simulation results 107
5.4.1. Real-time updating of produced powers 107
5.4.2. Case study 107
5.5. Conclusions and discussions 108
CONCLUSION 111
BIBLIOGRAPHY 115
INDEX 127