Methods of Inference for Logical-Type Systems Based on the Fuzzy Degree of Truth

Abstract

The work is devoted to the development of methods of inference (that are used in logicaltype production systems) based on the fuzzy degree of truth for the most commonly used fuzzy implications. For the block of rules at the stage of defuzzification, the discrete version of the method of the center of gravity is applied. The inference of an output value is considered for two cases: where the membership functions of consequents of rules intersect and where these functions are disjoint. The second case also includes models of the singleton type. For both cases, the relations of computing the clear output of the block of system rules are derived; they reduce the computational complexity to the polynomial complexity.