Optimal Design of PMa-SynRM for Electric Vehicles Using a Subregion-Assisted Hybrid Algorithm with Adaptive Nelder–Mead Simplex

The optimal design of a motor can be characterized as a multi-modal optimization problem because numerous performance factors must be considered. Moreover, a time-consuming finite element method (FEM) is essential for precise analysis of motor characteristics. Especially in multi-modal problems, computation time is increased. This article proposes a subregion-assisted hybrid algorithm (SAHA) and its hybridization with the adaptive Nelder-Mead simplex (ANM) algorithm to address the computational burden in multi-modal problems. The SAHA method acts as a global search method to reduce meaningless function calls and to enhance the ability to find undiscovered optima using the subregion and region exclusive space-filling method (RESM). In addition, the ANM acts as a local search method, enabling rapid convergence and reducing function calls using information obtained from the SAHA. The performance of the proposed algorithm was validated through comparison with the niching genetic algorithm (NGA), which is a conventional multi-modal optimization algorithm. Then, the proposed algorithm was applied to the optimal design of a permanent-magnet-assisted synchronous reluctance motor (PMa-SynRM) for electric vehicle traction to reduce torque ripple while considering the average torque and efficiency. Finally, irreversible demagnetization analysis and stress analysis were performed to verify the effectiveness of the selected optimal model.