A Method for Constructing Fuzzy Reasoning Models by Variance and Correlation

Abstract

In fuzzy modeling, several techniques for constructing fuzzy reasoning rules have been proposed. In this study, we propose a constructing technique for fuzzy reasoning models considered correlation of rules. This technique will be referred to as correlation type fuzzy reasoning models. In order to show the effectiveness, some numerical simulations were performed. As a result, fuzzy reasoning models were obtained, with a higher precision and a few rules. Further, the problem of overlapping rules was solved by combining correlation type fuzzy models with the deleting type fuzzy models.