Liu G.P. Nonlinear Identification and Control. A Neural Network Approach

Springer, 2001. — 224 p.It is well known that linear models have been widely used in system identification for two major reasons. First, the effects that different and combined input signals have on the output are easily determined. Second, linear systems are homogeneous. However, control systems encountered in practice possess the property of linearity only over a certain range of operation; all physical systems are nonlinear to some degree. In many cases, linear models are not suitable to represent these systems and nonlinear models have to be considered. Since there are nonlinear effects in practical systems, e.g., harmonic generation, intermodulation, desensitisation, gain/expansion and chaos, neither of the above principles for linear models is valid for nonlinear systems. Therefore, nonlinear system identification is much more difficult than linear system identification.
Any attempt to restrict attention strictly to linear control can only lead to severe complications in system design. To operate linearly over a wide range of variation of signal amplitude and frequency would require components of an extremely high quality; such a system would probably be impractical from the viewpoints of cost, space, and weight. In addition, the restriction of linearity severely limits the system characteristics that can be realised.
Recently, neural networks have become an attractive tool that can be used to construct a model of complex nonlinear processes. This is because neural networks have an inherent ability to learn and approximate a nonlinear function arbitrarily well. This therefore provides a possible way of modeling complex nonlinear processes effectively. A large number of identification and control structures have been proposed on the basis of neural networks in recent years.
The purpose of this monograph is to give the broad aspects of nonlinear identification and control using neural networks. Basically, the monograph consists of three parts. The first part gives an introduction to fundamental principIes of neural networks. Then several methods for nonlinear identification using neural networks are presented. In the third part, various techniques for nonlinear control using neural networks are studied. A number of simulated and industrial examples are used throughout the monograph to demonstrate the operation of the techniques of nonlinear identification and control using neural networks. It should be emphasised here that methods for nonlinear control systems have not progressed as rapidly as have techniques for linear control systems. Comparatively speaking, at the present time they are still in the development stage. We believe that the fundamental theory, various design methods and techniques, and many application examples of nonlinear identification and control using neural networks that are presented in this monograph will enable one to analyze and synthesise nonlinear control systems quantitatively. The monograph, which is mostly based on the author's recent research work, is organised as follows.Neural Networks
Sequential Nonlinear Identification
Recursive Nonlinear Identification
Multiobjective Nonlinear Identification
Wavelet Based Nonlinear Identification
Nonlinear Adaptive Neural Control
Nonlinear Predictive Neural Control
Variable Structure Neural Control
Neural Control Application to Combustion Processes