A Tutorial on Fuzzy Rule-Based Expert Systems (FRBES) Models. 2: Models and Computational Techniques

Abstract

Fuzzy rule-based expert systems (FRBES) models, described by a set of fuzzy conditional statements in the canonical form of (If x is A^k then y^k is B^k for k = 1, 2, … , r) or a system of conjunctive or disjunctive fuzzy relational equations in the form of (y^k = x · R^k for k = 1, 2, … , r) were discussed in part I of this tutorial. In Part II provides detailed descriptions of some interesting special cases of FRBES models of dynamic systems along with a number of commonly used formats for FRBES models and their solutions. The fundamental computational processes of FRBES models such as partitioning input and output spaces into fuzzy partitions and assigning them membership functions, fuzzification, inferencing, defuzzification processes are illustrated. A number of interesting graphical techniques for the above-mentioned computational processes are also given. Finally, fuzzy associative memories tables are introduced.