Learning systems in intelligent control:an appraisal of fuzzy, neural and genetic algorithm control applications

D.A. Linkens, Henry O. Nyongesa

IEE Proceedings - Control Theory and Applications · 1996 · 160 citations · 28 references

Concepts

Abstract

In designing controllers for complex dynamical systems there are needs that are not sufficiently addressed by conventional control theory. These relate mainly to the problem of environmental uncertainty and often call for human-like decision making requiring the use of heuristic reasoning and learning from past experience. Learning is required when the complexity of a problem or the uncertainty thereof prevents a priori specification of a satisfactory solution. Such solutious are then only possible through accumulating information about the problem and using this information to dynamically generate an acceptable solution. Such systems can be referred to as intelligent control systems. In recent years, ‘intelligent control’ has come to embrace diverse methodologies combining conventional control theory and emergent techniques based on physiological metaphors, such as neural networks, fuzzy logic, artificial intelligence, genetic algorithms and a wide variety of search and optimisation techniques. The paper reviews aspects of these emergent techniques, in particular, fuzzy logic, neural networks and genetic algorithms that pertain to realisation of intelligent control systems. The fundamental concepts and design techniques of each paradigm are dicussed, providing a compact reference for their application.

References

28