Hybrid Surrogate Model-Based Multidisciplinary Optimization on High Speed Permanent Magnet Synchronous Motor Facilitating Rotor Design

This article proposes a rapid optimization approach based on a hybrid surrogate model for a high-speed permanent magnet synchronous motor (HSPMSM) to address the difficulties presented by numerous variables, lengthy simulation computation time, and a sizable nonlinear search space in the multidisciplinary optimization of surface-mounted HSPMSM. The proposed model creates a high-precision hybrid surrogate model to decrease fitting error by using the Adaptive Particle Swarm Optimization (APSO) algorithm to calculate the weights of individual submodels. The optimization goals include torque ripple, copper loss, iron loss, and rotor stress. Sobol global sensitivity analysis is employed to ascertain the optimization variables, whereas optimal Latin hypercube sampling is used to extract a representative sample set. The hybrid surrogate model is built using a variety of submodels, and its high fitting precision is confirmed by using the index $\mathrm{R}^{2}$ of determination as the accuracy metric. The Nondominated Sorting Genetic Algorithm II (NSGA-II) is implemented for multidisciplinary optimization, resulting in a Pareto set. Finite element analysis has proven the viability and efficiency of this approach for high-precision multidisciplinary optimization of HSPMSM.