Failure diagnosis and decision making in industrial processes: A fuzzy set application

The main purpose of failure diagnosis is to analyze any pattern of defects or process malfunctions and to indicate which factors are the most likely causes of those defects. This paper provides a review of the different techniques used to perform this diagnostic analysis. After a brief description of traditional methods, the two most recent approaches, based on production rules and fuzzy set theory, are presented. The two major trends of the fuzzy set based approach are then described. The first one is a fuzzy extension of classical cluster analysis techniques. The second one is the solution of the inverse problem for compositions of fuzzy relations. A different technique, based on the fuzzy relation and the use of fuzzy numbers, is then discussed.