Fuzzy adaptive control with an associative memory system

Abstract

A fuzzy adaptive control system which uses the fuzzy associative memory organising units system consists of an if-part, an if-then-rule-part and a then-part. The fuzzy rules consist of pairs of a reference model for a controlled process and its controller as knowledge for adapting control. In the if-part, learning vector quantisation (LVQ) networks evaluate process output and show the distances between output response patterns for processes and for the reference models with fuzzy sets. In the fuzzy if-then-rule-part, an associative memory system connects reference models for processes with controllers given in the then-part. The paper presents the control design and shows how to use LVQ to create membership functions of fuzzy rules automatically. It explains fuzzy adaptive control systems which use LVQ, fuzzy models, and associative memory systems. It then presents a nonlinear application example.