Eddy Current Analysis and Optimization Design of the Secondary of the Linear Induction Motor With an Approximation and Prediction Method

This article presents an efficient multi-objective optimization model for linear induction motors (LIMs) with a novel approximation and prediction strategy. The proposed method can realize the optimization design with speediness and accuracy, while reducing the calculation complexity of eddy current analysis. The sample data in this model are obtained by 3-D finite-element method (FEM) considering the skin effect and edge effect. Then, the Latin hypercube sampling (LHS) method is adopted to extract dimensional data including every structural parameter of the secondary. Based on these samples, a multiple analysis model is established using eXtreme Gradient Boosting (XGBoost) method. By employing this analysis model, several optimization structures are provided satisfying the limitation and rationality of designing structure and performance requirements. The precisions of the thrust force, vertical force, eddy current, and copper loss of optimal structures are validated by FEM.