Design of adaptive neuro-fuzzy controllers

This paper proposes a design of adaptive fuzzy-logic based controllers with neural networks. A detailed discussion of effects of different reasoning methods on fuzzy controls is given and used to illustrate the need for an adaptive implementation of fuzzy control systems. The procedure of decision-making of a fuzzy-logic based control system (FLCS) leads to a structured neuro-fuzzy network consisting of three types of subnets for pattern recognition, fuzzy reasoning, and control synthesis, respectively. The unique knowledge structure embedded in this network enables it to carry out adaptive changes of membership functions for both input signal patterns and output control actions, and of fuzzy conjunction operators, then recover these changes separately later. Gradient methods for optimization have been used to derive off-line training rules and online learning algorithms for the structured neuro-fuzzy network