Abstract
A probability theory-based method for simultaneous optimization of multi-object orthogonal test design is addressed in the present paper, which employs the concept of preferable probability to represent the preferable degree of the candidate alternative in the optimization. The utility indexes of all the performance indicators of alternative are divided into beneficial and unbeneficial types according to the preference in the optimization, and each utility index contributes to a partial preferable probability in positively or negatively correlative manners linearly according to its type; the total preferable probability of a candidate alternative is the product of all partial preferable probabilities, which thus transfers the multi-objective problem into a single objective problem. Finally, all candidate alternatives are ranked upon their total preferable probability to complete the optimization. As to multi-objective orthogonal test design, the optimization is conducted by applying range analysis to the total preferable probabilities of candidate alternatives.

Key words
orthogonal test design, multi-object optimization, probability-based method, overall consideration, preferable probability

Full text (126 KB)