A self-organizing fuzzy controller with a fixed maximum number of rules and an adaptive similarity factor

This paper proposes a self-organizing fuzzy controller with a broad generality for minimum phase and stable systems. The controller learns the rules on-line with a minimum knowledge about the process. The rule base is built and permanently actualized from input-output real time data and has a fixed maximum number of rules (FMNR). An (on-line) adaptive similarity factor implements a special efficient inference technique. Feedforward and predictive effect is introduced in fuzzification and defuzzification stages. The defuzzification is carried out in such a way that as the learning process progresses the interval of the control becomes more and more accurate. Results are shown concerning simulations for non-linear SISO, MISO and MIMO systems and a real experimental application using a low-cost microcomputer. (C) 1999 Elsevier Science B.V. All rights reserved.