A weighted fuzzy production rule evaluation method

Abstract

The knowledge base of a fuzzy expert system is usually represented by fuzzy production rules (FPRs) which can represent fuzzy, vague or imprecise knowledge. However, their knowledge representation power is somewhat limited if they do not allow each proposition in the antecedent part of a given fuzzy production rule to have a different degree of significance. In this paper, each proposition in the antecedent part is assigned a weight, and a fuzzy production rule evaluation method (FPREM) based on point-valued fuzzy sets and weight assignment is proposed. A weighted FPR is preferred over the non-weighted version because it is more flexible. In addition, the advantages of our proposed FPREM are: 1) discrete fuzzy set tables can be used; 2) complex fuzzy relational matrix proposed by Zadeh could be avoided; and 3) the method of calculating the consequent is much easier than Zadeh's compositional rule of inference because there is no need to set up the fuzzy relation between the antecedent and the consequent. >