Multi-objective Optimization of Linear Induction Motor based on Surrogate Model

Aiming at the problems such as high time cost and low optimization efficiency in the multi-objective optimization of linear induction motor (LIM) using the finite element model (FEM) and optimization algorithm, this paper proposes the optimization method based on RBF neural network surrogate model and non-dominated sorting genetic algorithm-II (NSGA-II). Firstly, the high starting thrust, low thrust fluctuation and primary mass are taken as the optimization objectives, and the sample points are generated by using the optimal latin hypercube sampling (OLHS), and the performances of sample points are calculated by FEM. Then, RBF neural network surrogate model is constructed based on the sample data, and NSGA-II algorithm is used to conduct multi-objective optimization solution of the model. Finally, the accuracy and effectiveness of the optimization method are verified through finite element simulation, and the experimental results show that the optimized thrust has increased by 61%, and the thrust fluctuation and primary mass have decreased by 52.78% and 9.83% respectively.