Identification and Adaptive Control of Nonlinear Processes Using Combined Neural Networks and Genetic Algorithms

This paper presents a new combined neural-genetic method for identifying processes comprising static nonlinearities and linear dynamics, and for adapting the model to linear and nonlinear parameter changes. The approach is based on multilayer and recurrent neural networks. It is shown that the powerful properties of genetic algorithms may be used to advantage in a learning neural network. We solve a number of problems and compare the results and the method with those previously reported by Narendra & Parthasarathy.