Parameter Design in Production by Means of Robust Fuzzed PMOO in Case of Desirable Target

Abstract

A proper parameter design in the production process is critical to ensuring the quality of products and their improvement. The rationality of previous traditional approaches for robust assessment, including the Taguchi method and dual response method, is questionable. In this article, the combination of probabilistic multi-objective optimization (PMOO) with membership approach in fuzzy theory is developed to conduct parameter design of production in case of desirable target with robustness deeply. This method is further applied to two examples of both parametric design of gas metal arc (GMA) welding process and the printing machine’s ability. In the new approach, the mean value of “complement” of the membership value from a set of test data belonging to its desired target of an objective response is taken as one sub-objective response, which is an unbeneficial type of index in the assessment, contributing the first part of the partial preferable probability of the objective. In contrast, the dispersion of a set of test data in terms of membership with respect to the desired target value is taken as the other sub-objective response to contribute the second part of the partial preferable probability of the objective simultaneously, which is an unbeneficial index. Thus, the fuzzed PMOO approach is regulated comprehensively. Besides, the consequences of application examples reflect the reasonability of the approach as an auxiliary measure for PMOO to consistently perform optimal robust design.