Data-driven current harmonic optimization for minimizing torque ripple and injection losses in PMSM drives

Abstract

Torque ripple mitigation is a critical topic in the field of permanent magnet machine drives, and current harmonic injection is regarded as an effective approach to address this issue. However, traditional harmonic injection methods heavily rely on model-based calculations that necessitate various precise motor equivalent parameters. Additionally, they struggle to account for the iron loss effect. Furthermore, due to the nonlinear nature of motor parameters, these approaches frequently result in suboptimal torque ripple mitigation and elevated injection losses. To overcome these limitations, this article proposes a data-driven-based harmonic injection method. In contrast to model-based techniques, the proposed method offers the advantages of independence from motor parameters, unaffected torque ripple reduction by magnetic saturation, and overall minimization of injection copper and iron losses. The key of the proposed method lies in establishing precise correlations between injected current harmonic, torque ripple, and losses through a meta-model. Moreover, a multiobjective optimization process is applied to identify the optimal injection currents, leading to minimizations in torque ripple and injection losses.