Mutual Impact of Computing Power and Control Theory

Recent rapid developments in computing power, such as parallel processing and neural networks, have stimulated new trends in control. However a discrepancy exists between available computing power and exploitable algorithms obtained classically from control theory. The aim of this book is to address the discrepancy from both the com­ putational power and control theory viewpoints. Areas such as identification, adaptive control, signal processing and neural networks therefore hold a prominent position in the text presented. The form of the book is such that it should be useful for readers at various levels, particularly those at the research and/or application stage. The book has resulted from the IFAC Workshop on the Mutual Impact of Computing Power and Control Theory, which was held at the Institute of Information Theory and Automation (UTIA), Prague, in September 1992. Organisation of the event was provided jointly by the Department of Adaptive Systems, UTIA, Prague and the School of Engineering and Information Sciences, University of Reading, UK. Selected papers from the Workshop have been chosen to give a good balance across the field, whilst at the same time highlighting important areas for future research. In this way the book represents edited Proceedings from the Workshop. One point, quickly apparent, is the international nature of the presentations themselves, which provide not only a technical appraisal of the field but also inject cultural aspects which are vitally important on the path ahead.

Table of Contents:
Upcoming Numerical Linear Algebra Issues in Systems and Control Theory.- On Design of Approximate Finite-Dimensional Estimators: the Bayesian View.- Sidestepping the Certainty Equivalence in 2-DOF Adaptive Control Via Multiple Implicit Identifiers.- Simulation and Control of Gas Transport and Distribution by Large-Scale Pipeline Networks.- Complexity of Dynamic Models for Intelligent Control.- The Most Simple Systolic Architectures for Parallel Probabilistic Searching Algorithms.- On the Riemannian Interpretation of the Gauss—Newton Algorithm.- Recursive Least Squares Approximation of Bayesian Non-Gaussian/Non-Linear Estimation.- Fundamental Properties of Real Computers.- A Numerical Method for Computing the Approximate Solution of the Infinite-Dimensional Discrete-Time Optimal Linear Filtering Problem.- Smart Matrices.- Branch and Bound Algorithm for Global Optimization in Control Theory.- Mapping LQ Control Design on Fixed-Size Array Processors.- Identification of Data-Compatible Models for Control Applications.- MIMO — A Set of SISO? Multivariate System Adaptively Controlled as a Set of Single-Input Single-Output Models.- Extensions of Standard CADCS Environment for Parallel Computation and Real Time Control.- On the Ordering of Characteristic Input-Output Modes in MIMO Discrete-Time Systems.- Regularised Lattice-Ladder Adaptive Filter.- Neural Network Applications — Case Studies.- A Virtual Software Environment for Artificial Neural Networks on General Purpose Parallel Architectures.- Shape Discrimination and Classification in Robotic Vision Using Scaled Normalized Central Moments.- Autotuned Predictive Control.- Multivariable Constrained Adaptive Predictive Control Based on Pole Placement Design.- The Epsilon - Ritz Method for Solving Optimal Control Problemson Parallel Computers.- Robust PID Control.- Advanced Algorithms Contra Lack of Computing Power.- Personal-Computer-Oriented Systems of Adaptive Identification.- Predictive Control with Constrained Spline Input Signal.- Expert System-Driven Controllers.