Robust Design Optimization Approach by Combination of Sensitivity Analysis and Sigma Level Estimation

Abstract

This paper proposes a novel robust optimization approach by combination of sensitivity analysis and sigma level estimation in multi-objective problems. In order to obtain robust Pareto-optima within acceptable computational costs, we actively reuse the already calculated data for the sensitivity analysis in multipoint search algorithm. The sigma level estimation is also utilized to select the significant solutions from a large number of Pareto-optima based on both the robustness and the performance. Some numerical results that verify the effectiveness of the proposed methods are presented.