Fuzzy logic models for ranking process effects

When modeling and analyzing manufacturing processes, it may be helpful to know the relative importance of the various process parameters and their interactions. This ranking has traditionally been accomplished through regression modeling and analysis of variance (ANOVA). In this paper, we develop a fuzzy logic modeling technique to rank the importance of process effects. Several different cases are presented using functions that allow the determination of the actual importance of effects. The impact of noisy data on the results is considered for each case. It is shown that in many cases the fuzzy logic model (FLM) ranking methodology is capable of ranking process effects in the exact order or in an order reasonably close to the exact order. For complex processes where regression modeling and ANOVA techniques fail or require significant knowledge of the process to succeed, it is shown that the FLM-based ranking can be performed successfully with little or no knowledge of the process.