Conference: 2025 한국자동차공학회 춘계학술대회
Authors: 변제건, 황윤재, 임명섭
DOI:
This paper focuses on the selection of the optimization region in the two-step optimization process. Existing global optimization techniques typically require a large number of experimental points to find the global optimum. However, this leads to an excessive increase in the time required for finite element analysis (FEA) to ensure accuracy. To address this issue, a new criterion for selecting the second-stage optimization region is proposed after first-stage optimization. This criterion considers the trend of the objective function and the local gradients of design variables and performance. The process of selecting experimental points is carried out using the optimal latin hypercube desgin (OLHD) and sequential maximin distance desgin (SMDD) methods, and the Kriging method is applied to generate a surrogate model. The optimization is carried out using a genetic algorithm (GA), and the model's accuracy is validated based on nomalized root mean square error (NRMSE). The results demonstrate that the proposed method better satisfies the objective function and constraints compared to conventional approaches, improving both torque ripple performance and reliability. This research offers a practical approach for high-efficiency motor design and optimization.