1996 Volume 8 Issue 4 Pages 695-705
For generating or tuning fuzzy rules Gaussian-type membership functions in a fuzzy system model, it is well known to adopt the Neuro-Fuzzyl learning algorithm based on Gradient-descent method. In the learning algorithm, however, the number of tuning parameters increases quickly, as the number of input variables increases. Moreover, the representatoin of fuzzy rules after learning in the form of fuzzy rule table becomes difficult and the case of weak-firing occurs. In this paper, we propose a new learning approach for generating or tuning fuzzy rules, in order to improve the above problems. Further, we compare it with the conventional method several identified functions, and show that the proposed method is a useful tool for learning a fuzzy system model.
