Genetic Algorithm-Based Current Optimization for Torque Ripple Reduction of Interior PMSMs

Abstract

This paper investigates the torque ripple modeling and minimization for interior permanent magnet synchronous machines (PMSMs). At first, a novel torque ripple model is proposed in which the torque ripples resulted from the spatial harmonics of permanent magnet flux linkage, time harmonics of stator currents and the cogging torque are included. Based on the torque ripple model, a genetic algorithm (GA)-based harmonic current optimization approach is proposed for torque ripple minimization. In this approach, GA is applied to optimize both the magnitude and phase angle of the stator harmonic currents to minimize the peak-to-peak torque ripple, minimize the sum of squares of the harmonic currents, and maximize the average torque component produced by the injected harmonic currents. The results demonstrate that the magnitude of the harmonic current can be significantly reduced by optimizing the phase angles of these harmonic currents. This leads to further suppression of the torque ripple when compared with that of a case where phase angles are not considered in the optimization. Also, an increase of the average torque is achieved when the optimum harmonic currents are injected. The proposed model and approach are evaluated through both numerical and experimental investigations on a laboratory interior PMSM.