Multi-Objective Optimization Design Of Permanent Magnet Vernier Motor Based On Surrogate Model

Low speed and high torque characteristics of permanent magnet vernier motors give them advantages in direct drive applications.In order to further improve the torque performance of the motor,it is necessary to optimize the design of the motor.Although the traditional finite element analysis optimization has high accuracy,it has the problems of long calculation time and low optimization efficiency.It is very important to improve the calculation efficiency of the optimization design.Therefore,the optimization based on the efficient and applicable surrogate model is very advantageous.The core of surrogate model optimization is to train the surrogate model with high generalization ability.This thesis studies the multi-objective optimization method based on surrogate model and applies it to the optimization design of permanent magnet vernier motor.The main research contents are as follows:(1)In order to establish a high precision surrogate model with a small number of samples,an improved Latin hypercube sampling method considering the spatial distribution uniformity and orthogonality of the sample points and a Pareto frontier adding method considering the distance between the frontier points and the sample points are used to analyze the global precision and local precision;The validity of the method is verified by 3D power function and ZDT1 problem.(2)Determine the main parameters of the permanent magnet vernier motor based on the motor size equation and design indicators,and analyze its topology and winding structure.Based on finite element analysis,the influence of the main structural parameters of traditional permanent magnet vernier motors and split tooth permanent magnet vernier motors on average torque and torque ripple was analyzed to determine the range of optimization variables.Based on the Sobol model,the sensitivity indices of various structural parameters to average torque and torque ripple were analyzed,providing a basis for variable grouping.(3)In order to improve optimization efficiency and reduce optimization cycles,a multiobjective hierarchical optimization method based on Sobol sensitivity analysis was designed.The structural parameters were grouped and optimized sequentially by the size of the total sensitivity index and cross sensitivity index.The advantages of the proposed method are verified by comparison with direct finite element analysis optimization and surrogate model non-hierarchical optimization.(4)In order to verify the feasibility of the multi-objective optimization design method proposed in this thesis and the correctness of simulation analysis,a permanent magnet vernier motor testing platform was built and tested,mainly including no-load back electromotive force,rated load speed torque waveform,and motor performance under different working conditions.The optimization based on surrogate model has the advantages of wide applicability and fast calculation speed,and has a good application prospect in motor optimization.