Diagnosis based on fuzzy IF-THEN rules and genetic algorithms

Abstract

This paper proposes an approach for inverse problem solving based on the description of the interconnection between unobserved and observed parameters of an object (causes and effects) with the help of fuzzy IF-THEN rules. The essence of the approach proposed consists of formulating and solving the optimization problems, which, on the one hand, find the roots of fuzzy logical equations, corresponding to IF-THEN rules, and on the other hand, tune the fuzzy model on the readily available experimental data. The genetic algorithms are proposed for the optimization problems solving. The efficiency of the method is illustrated by computer experiment, and also by the example of the inverse diagnosis problem, which requires renewal of the causes (inputs) by the observed effects (outputs).