Optimization of cogging torque in permanent magnet motors based on parameter sensitivity and Kriging model

Abstract

The weakening of cogging torque in permanent-magnet synchronous motors brings about the problem of low efficiency optimization. To solve this problem, an optimization-simulation scheme for the response surface model based on the parameter sensitivity screening is proposed. A surface-mounted motor having 12 poles and 48 slots is used to obtain the law of the influence on cogging torque caused by multiple electromagnetic parameters. Additionally, the Spearman correlation is used to analyze significant factors that affect the cogging torque. The sample is collected by an OSF optimal space filling design. A Kriging model is applied to construct the relationship between the cogging torque and significant factors, and to modify the model accuracy. The multi-island genetic algorithm is used to find an optimal solution to the cogging torque. The results show that the result of the finite element analysis after optimization only has an error of 1.5% compared with the predicted value of the pseudo model, which is 62% lower than the initial value. Therefore, the sharp reduction in cogging torque in the motor verifies the effectiveness of the optimization scheme.