Robustness-Oriented Design Optimization Approach for Cogging Torque of Batch Production PMSM

Ubiquitous uncertainty from design, fabrication, and assembly often leads to erratic or infeasible cogging torque solutions, which observably enlarges the difficulty of the robustness-oriented design optimization (RDO) of PMSM. Robustness-oriented design optimization approaches based on FEA (finite element analysis) are still inefficient. Despite adequate surrogate and analysis models, RDO's efficiency and accuracy, considering multi-source uncertainty and repetitive elements, still need to be improved to meet practical needs. Therefore, a physically based surrogate model-assisted RDO approach is proposed in this study and comprises two main contributions. First, we develop an accurate and efficient physical-based surrogate model and the data-based layer model, including an improved vector superposition approach that characterizes both the relationship between rotor permanent magnet uncertainties and the additional harmonics, and the relationship between the rotor permanent magnet design parameters and the inherent harmonics of the cogging torque. The organic combination of these two models realizes the high-efficiency and high-precision calculation of the PMSM cogging torque. Second, we propose an overall RDO strategy for the various uncertainties and high dimensionality. This study presents and discusses the RDO results of the PMSM used in high-precision robots and machine tools. Experimental and simulation results are presented to validate the effectiveness of the proposed surrogate model, strategy, and approach.