Fuzzy Rules Optimization in Fuzzy Expert System for Machinability Data Selection: Genetic Algorithms Approach

Abstract

Machinability data selection is complex and cannot be easily formulated by any mathematical model to meet design specification. Fuzzy logic is a good approach to solve such problems. Fuzzy rules optimization is always a problems for a complex fuzzy rules from more than 10 thousand combinations. (Wong et aL 1997) developed fuzzy models for machinability data selection. There are more than 2 x 1029 possible sets of rules for each model. Situation would be more complicated if further increase the number of inputs and/or outputs. The fuzzy rules were selected by trial and error and intuition in reference (Wong et aL 1997). Genetic optimization is suggested in this paper to further optimizing the fuzzy rules optimization with genetic algorithms has been developed. Weighted centroid method is used for output defuzzi fication to save processing time. Comparisons between the results of the new models and the previously published literatures are made.