Selection of Heuristic Rules by Fuzzy Inference

摘要

This paper presents a fuzzy inference based procedure for the real-time selection of suitable heuristic rules in dynamic production environments. The proposed procedure is composed of four parts: (a) selection of environmental variables to detect the dynamic changes in the production environment; (b) an information repository containing all static and dynamic information related to the production system; (c) a fuzzy inference system that attributes preference levels to the alternative heuristic rules based on the current status of the environmental variables retrieved from the information repository; and (d) a scheduling decision system that uses the heuristic rule, or combination of rules, more appropriate to the current condition of the production environment The information repository is updated in real time. Each time a machine becomes free or a new production order arrives, the preference levels of the heuristic rules are revised and a scheduling decision is made. Two applications of the proposed system are discussed. The results obtained by the proposed procedure are significantly better than the performance of each of the heuristic rules. The proposed system outperforms also a knowledge-based system using equivalent number of if-then rules, as well as another fuzzy inference based system using a classical defuzzification procedure.