Robust Design and Assessment of Product and Production with Multiple Objectives

Each objective can decompose dual sub-objectives, i.e., the mean value and the dispersion of the objective in principle, so multi-objective robust design problem can be seen as the “dual multi-objective optimization” reasonably; the evaluation for the part of partial preferable probability of mean value is conducted according to the type of the attribute, while the evaluation for the part of partial preferable probability of the dispersion is conducted as an unbeneficial type of attribute; the total preferable probability of a candidate scheme is the product of the complete parts of partial preferable probabilities of all possible response attributes, which is the unique indicator of the corresponding alternative in the comprehensive competition, the optimal candidate scheme is with highest total preferable probability.