Hierarchical Surrogate-Based Optimization Strategy for Optimal Design of Switched Reluctance Motors

摘要

Electrical machines can be realistically analyzedusing numerical methods. However, direct coupling of such tools with search algorithms is computational expensive. Surrogate models are therefore invaluable to reduce the computational cost. In this paper, a hierarchical surrogate-based optimization (SBO) strategy is proposed and applied for optimal design of switched reluctance motors (SRMs). A trust region approach decomposes the problem into optimization subproblems specific to each trust region. The surrogates are applied in a sequence of optimization steps, with the high-fidelity models used to correct the surrogates during the optimization process.