Fuzzy Modeling and Fuzzy Rule Generation in Response Surface Based Approximate Optimization

摘要

Among the valuable applications of fuzzy theories, evolutionary fuzzy modeling (EFM) facilitates global approximate optimization characteristic in evolutionary computation. The paper proposes clustering as a method of fuzzy rule generation. Generally fuzzy rules are generated with raw data from experiments by intuition, in this process, human uncertainty could intervene in fuzzy modeling causing errors. In this point of view, fuzzy clustering provides a good way of fuzzy rule generation. Fuzzy c-means (FCM) is used for clustering and some methods of adapting clustering to EFM is presented.