STRUCTURE/PARAMETER OPTIMIZATION OF FUZZY MODELS BY EVOLUTIONARY ALGORITHM

摘要

In this paper, a new evolutionary algorithm for designing fuzzy models will be proposed. The structure and parameters of fuzzy models are simultaneously optimized by the proposed evolutionary algorithm. The objectives are in twofold. On the one hand, we are try to develop a fast and practical method for modeling high dimensional complex systems with a as possibly simplified fuzzy model. On the other hand, we hope that the resulting fuzzy model is easy to understand, that is, the resulting fuzzy model do not lose its interpretability. Through simulations on modeling problems, it will be shown that the proposed algorithm can successfully find a fuzzy model which approximates the given unknown function accurately with compact number of rules.