Application-oriented multiobjective robust parameter design of a permanent magnet synchronous motor based on modified Bayesian optimisation

From the perspective of industrial application, various uncertainties from repetitive unit leading performants fluctuations increase the obstacles to the multiobjective robust parameter design (MRPD) of PMSMs to fulfil multiple operating conditions. Although state-of-the-art methods assume that various uncertainties are simplified and more efficient surrogate models are used, MRPD is still inefficient, even with many pseudo-optimal solutions that are difficult to batch produce. To solve these issues, an application-oriented MRPD method based on modified Bayesian optimisation is proposed in this study. This method treats surrogate modelling and global optimisation under various uncertainties as iterative and evolutionary processes, which increases the diversity of surrogate model samples and optimisation iteration subsets and solves the problems of low efficiency and poor accuracy in traditional methods. Finally, the PMSM for a high-precision CNC machine tool serves as a case study to validate the effectiveness of the method proposed in this study. The consistency of the cogging torque, torque, and torque ripple is considerably improved under all operating conditions.