Construction Methods of Fuzzy Modeling Based on Learning Algorithms

Abstract

In recent studies about fuzzy modeling, numerous automatic modeling attempts following all qualifications for precision, number of rules and learning times using neural networks have been made. Specifically, a fuzzy modeling using neural netowork deleting rules in learning and a fuzzy modeling with iterative generation mechanism of fuzzy inference rules are wellknown for showing good characters. This paper suggests generalization methods combing them of getting a fuzzy reasoning that have best number of rules after tuning and deleting rules and shows validity to it in numerical examples and in appilcation to the obstacle avoidance.