Optimal Weighting Coefficient Assisted Reliability-Based Robust Optimal Design of Electrical Devices

Abstract

In design and manufacturing process of electrical devices, there exist uncertain factors such as uneven material characteristics and manufacturing tolerance of geometric structure. The uncertainty may result in performance deterioration or constraint violation of electrical device obtained in design process. Therefore, it is necessary to consider uncertain factors in the design stage of electrical equipment. In this paper, the optimal weighting coefficient assisted reliability-based robust optimization (ω-RBRDO) is proposed. In this method, the deterministic constraints are replaced by probabilistic ones to ensure reliability. The mean and variance of original objective function against uncertainties are taken as robust indexes. Based on optimal result of deterministic optimization, the mean and variance are normalized, which are combined into one single objective by the optimal weighting coefficients. The ω-RBRDO is investigated by analytic test function and standard electrical problem of superconducting magnetic energy storage system. Finally, the ω-RBRDO is applied to the design of permanent magnet magic ring, which finds the optimal design with maximum magnetic flux density in center airgap, better reliability, and strong robustness.KeywordsOptimization designReliabilityRobustnessUncertain factors