Parametric Equivalent Magnetic Network Modeling Approach for Multiobjective Optimization of PM Machine

Abstract

This article proposes a novel parametric equivalent magnetic network (EMN) modeling approach for multiobjective optimization of a surface-mounted permanent magnet (SPM) machine. The key is to apply variable meshes under adjustable topology to establish parametric EMN model. First, a novel adjustable cross-shape mesh method with variable sizes and regular permeance is proposed, which can flexibly vary with parameters. The meshes are combined in EMN model to simultaneously achieve high accuracy, high computing efficiency, and structural parameterization. Second, comprehensive sensitivity analyses are utilized to separate variable EMN parameters with the parametric feature. Significant variables are employed in multiobjective differential evolution algorithm with response surface models. Then, the optimal parameters are balanced and determined from Pareto front with improved torque performance. Finally, a prototype is manufactured, and the effectiveness of the proposed EMN-based optimization design method is verified by finite-element analysis and experimental results.