Surrogate-Model-Based Multilevel Optimization Design and Analysis of a New Flux Switching Machine With Double-Sided PM Excitation
Yao Meng, Shuhua Fang, Yanzheng Zhu, Hong Chen, Yuxiang Zhong, Ling Qin
Abstract
In this paper, a new flux switching machine with double-sided PM excitation (FSDPMM) is proposed and optimized by using a surrogate model based multi-level optimization (SMMLO) method. The proposed FSDPMM is featured by using PMs on both the stator and rotor sides, which can operate based on the bidirectional flux modulation effect to accommodate abundant air-gap flux density working harmonics to increase torque density. Besides, a new SMMLO method that combining the gradient boosting decision tree (GBDT) based surrogate model and non-dominated sorting genetic algorithm-II (NSGA-II) is proposed to optimize the FSDPMM, which can improve the optimization efficiency effectively. The topology and operation principle of FSDPMM is firstly described. Then, the FSDPMM is optimized by using a SMMLO method. To evaluate the effectiveness of SMMLO method, the electromagnetic performances of initial and optimal FSDPMMs are compared using finite element method. In addition, the advantages of the proposed FSDPMM are evaluated by comparing the electro-magnetic performances of FSDPMM and three existing machines. Finally, a prototype of FSDPMM is manufactured and an experimental validation is carried out.