The simplification of fuzzy control algorithm and hardware implementation

Simplification of Fuzzy Control Algorithm and Hardware Implementation Z. Q. Wu, P. Z. Wang and H. H. Teh Institute of Systems Science National University of Singapore Heng Mui Keng Terrace,Kent Ridge Singapore 0511 The conventional inference composition algorithm of fuzzy controller is very time and memory consuming. As a result, it is difficult to do real time fuzzy inference and most fuzzy controllers are realized by look-up tables. In this paper we derived a simplified algorithm using the defuzzification mean of maximum. This algorithm takes shorter computation time and needs less memory usage, thus making it possible to compute the fuzzy inference on real time and easy to tune the control rules on line. The responsibility of thisalgorithm is proved mathematically In thls paper. Fuzzy controller has been highly developed and come to a new stage of hardware lmplementatlon. Many fuzzy controllers(or so called fuzzy inference machines) in hardware have been available in the market. The conventional fuzzy inference algorithm on which most fuzzy controller based on is too complicated. Further, its hardware implementation is very expensive and of a large volume, and the inference speed is limited. Reducing its cost and volume and improving its inference speed are very important to this technology. In this paper we also describe a hardware implementation based on the above simplified fuzzy inference algorithm. 1. Fuzzy controller algorithm Assume that the fuzzy controller has two inputs and a single output as shown in Figure 1,