Research on a Two-Stage Multi-Objective Optimization Design Method for Permanent Magnet Synchronous Motors Utilizing Partial Correlation Coefficients for Parameter Screening

Permanent Magnet Synchronous Motors have become widely adopted in new energy vehicles, industrial drives, and various other applications, thanks to their remarkable advantages such as high efficiency, high power density, and a broad speed regulation range. However, in the realm of multi-objective optimization design for PMSMs, traditional approaches often utilize the Pearson correlation coefficient for parameter screening, aiming to enhance computational efficiency. Unfortunately, the Pearson correlation coefficient does not adequately account for the interactions among optimization parameters, which can result in deviations in the optimization results. To address this concern, this study presents a multi-objective optimization design method utilizing partial correlation coefficients for parameter screening. By quantifying and comparing these coefficients, we can effectively screen key design parameters. Following this, an optimization design model is developed using a back propagation neural network, with a genetic algorithm employed for global optimization. Simulation results demonstrate that the proposed method exhibits significant advantages in optimization effectiveness compared to traditional methods, with a 78.5% reduction in torque ripple, 27.3% improvement in average torque, and 1% improvement in efficiency, providing more precise and efficient theoretical support for PMSM design.