Analysis and Multi-Objective Optimization of the Hybrid Excitation Switched Flux Machine

In this paper, a model-based multi-objective optimization method is proposed for the hybrid excitation switched flux (HESF) machine. In the proposed method, the index space distribution including torque index and flux index, which are highly related with the average torque and flux regulation capability, can be fast obtained by a simple magnetic circuit model, and then only these pareto front designs of the index space are calculated by finite element (FE). Hence, compared with the conventional pure FE-based multi-objective optimization, the optimization time of the proposed model-based multi-objective optimization is much smaller, e.g., 67 hours to 2.3 hours for the studied 12-slot/10-pole HESF machine on a desktop with 64 GB RAM and Intel Core i9-10900F CPU. In addition, in the high dimensional optimization the pure FE based optimization will easily crowd and stuck around many local optimum points due to flux-leakage and local magnetic saturation, whilst the proposed model-based optimization captures the main magnetic path characteristics, which can efficiently find better designs in a reasonable time. In this paper, the pareto front obtained from the proposed model-based optimization method contains better designs than that of the pure FE-based optimization method. A machine prototype is built and tested to validate the proposed model-based multi-objective optimization method in this paper.