Artificial evolution of neural networks and its application to feedback control

Abstract

This paper develops direct neural control systems with a novel structure inspired by the proportional plus derivative control. A parameter vector based uniform description of the problem of neural network design is presented. Difficulties associated with traditional mathematically-guided design methods are discussed, which lead to the development of a genetic algorithm based evolution method that overcomes these difficulties and makes direct neurocontrollers possible. Techniques are also developed to optimise the architecture in the same process of parameter training leading to a Darwin neural machine. The proposed methods are verified by examples of direct neurocontroller design for a linear and a nonlinear plant.