Abstract
Cogging torque has an important effect on torque ripple, as well as the vibration and noise of permanent magnet synchronous motor (PMSM). It is of great significance for the design and optimization of high precision motor. In this paper, the fluctuation of cogging torque in PMSM is analyzed and suppressed, and a Crested Porcupine Optimizer (CPO) combined with Back Propagation Neural Network (BPNN) and Fast Elitist Non-dominated Sorting Genetic Algorithm (CPO-BP-NSGA-II) is proposed to enhance the overall performance of motor. Firstly, the objective cogging torque formula is derived, with polar arc coefficient, permanent magnet thickness, stator slot opening width, and air gap length as structural parameters to be optimized. Secondly, a PMSM model with 10 poles and 36 slots equipped on the electric automobile is designed using ANSYS software. The improved CPO algorithm and BPNN are integrated to predict the cogging torque and average torque. The comparison results demonstrate that the model proposed can achieve higher prediction accuracy than traditional BPNN, with limited cost increase. Finally, to minimize the cogging torque while maintaining the average torque above its initial value, the NSGA-II algorithm is applied to optimize the design parameters. The simulation results indicate that the optimized motor achieves 1.4% increase in average torque and 85% reduction in cogging torque. And the optimized parameters can be acquired with a faster speed and higher accuracy, which provides an effective approach for the vibration performance improvement of PMSM.
Get full access to this article
View all access options for this article.