Fuzzy neural network for fuzzy modeling and control

Abstract

We propose a fuzzy neural network for fuzzy modeling and then apply the proposed network to control problems. A fuzzy model is constructed by the fuzzy partition of the considered state spaces, and then finely tuned by the proposed five-layer fuzzy neural network. After training by a prior expert knowledge of the target systems, the developed architecture can simultaneously obtain the optimal number of the fuzzy control rules and their corresponding optimal membership functions. In addition, to show its applicability, we have used examples and presented our results.