Neural Network Engineering in Dynamic Control Systems

The series Advances in Industrial Control aims to report and encourage technology transfer in control engineering. The rapid development of control technology impacts all areas of the control discipline. New theory, new controllers, actuators, sensors, new industrial processes, computer methods, new applications, new philosophies, .... , new challenges. Much of this development work resides in industrial reports, feasibility study papers and the reports of advanced collaborative projects. The series offers an opportunity for researchers to present an extended exposition of such new work in all aspects of industrial control for wider and rapid dissemination. Within the control community there has been much discussion of and interest in the new Emerging Technologies and Methods. Neural networks along with Fuzzy Logic and Expert Systems is an emerging methodology which has the potential to contribute to the development of intelligent control technologies. This volume of some thirteen chapters edited by Kenneth Hunt, George Irwin and Kevin Warwick makes a useful contribution to the literature of neural network methods and applications. The chapters are arranged systematically progressing from theoretical foundations, through the training aspects of neural nets and concluding with four chapters of applications. The applications include problems as diverse as oven tempera ture control, and energy/load forecasting routines. We hope this interesting but balanced mix of material appeals to a wide range of readers from the theoretician to the industrial applications engineer.

Klappentext

This study evaluates the state of the art in the area of neural networks from the engineering perspective. The contributions examine ways of improving the engineering involved in neural network modelling and control, so that the theoretical power of learning systems can be harnessed for practical applications. Neural Network Engineering in Dynamic Control Systems seeks to provide answers to the following questions: Which network architecture for which application? Can constructive learning algorithms capture the underlying dynamics while avoiding overfitting? How can we introduce a priori knowledge or models into neural networks? Can experimental design and active learning be used automatically to create "optimal" training sets? * How can we validate a neural network model?

Inhalt
1 Neural Approximation: A Control Perspective.- 2 Dynamic Systems in Neural Networks.- 3 Adaptive Neurocontrol of a Certain Class of MIMO Discrete-time Processes Based on Stability Theory.- 4 Local Model Architectures for Nonlinear Modelling and Control.- 5 On ASMOD An Algorithm for Empirical Modelling using Spline Functions.- 6 Semi-Empirical Modeling of Non-linear Dynamic Systems through Identification of Operating Regimes and Local Models.- 7 On Interpolating Memories for Learning Control.- 8 Construction and Design of Parsimonious Neurofuzzy Systems.- 9 Fast Gradient Based Off-line Training of Multilayer Perceptrons.- 10 Kohonen Network as a Classifier and Predictor for the Qualification of Metal-Oxide Surfaces.- 11 Analysis and Classification of Energy Requirement Situations Using Kohonen Feature Maps within a Forecasting System.- 12 A Radial Basis Function Network Model for Adaptive Control of Drying Oven Temperature.- 13 Hierarchical Competitive Net Architecture.