Reliability-Based Robust Optimization of High-Speed PM Synchronous Machine with Local Surrogate Model Strategy

A comprehensive study on reliability-based robust optimization (RBRDO) of high-speed permanent magnet synchronous machine (HSPMSM) is presented in this paper. To ease the computation burden due to robustness analysis in objectives for large scale optimization, this paper considers constraints and objective functions comprehensively. Polynomial chaos Chebyshev interval (PCCI) method is adopted to convert the influence of uncertainties on machine performances into analytical expressions, meanwhile, the weight index Monte Carlo analysis (WMCA) method is used to improve the efficiency of reliability calculation in constraints. Both collocation points in PCCI and sample points in WMCA are obtained by Kriging model with local surrogate strategy to shorten the optimization time. A design example of HSPMSM is investigated and optimization results based on traditional deterministic design optimization (DDO) method and RBRDO approach are obtained. Then, a detailed comparison between DDO and RBRDO is provided. It shows that the proposed local surrogate strategy can greatly reduce the prediction error and the RBRDO approach is effective to maintain high reliability for electric machines, especially for benchmark study.