Torque performance is one of the main factors to consider in motor design. To obtain desirable torque performance, torque ripple must be considered as well as average torque. Therefore, multi-objective optimization is necessary to find good design candidates. In this paper, we suppose cases where there is no previous experience in the development of motors with similar performance, or cases where we try to achieve performance improvement by expanding the design space, and therefore we also target the basic motor structure for optimization. In such a case, the design variables include discrete variables such as the number of slots and coil pitch. Since many design parameters, such as geometric dimensions, are typically continuous variables, our motor design problem is a mixed-variable optimization problem. Our objective functions may exhibit multimodality. In all cases, the finite element analysis (FEA) is required to evaluate objective function values. We propose a solution method to solve the optimization problem which has the above mentioned complex characteristics. By employing surrogate-based sequential approximate optimization (SAO) as the core framework, we aim to reduce computational cost. Furthermore, the proposed method incorporates an alternating direction method of multipliers (ADMM)-based approach to handle mixed variables, and mitigates multimodality by transforming the design variable space. Through a fundamental design problem of a permanent magnet synchronous motor (PMSM), we demonstrate that the proposed approach successfully identifies candidate designs with torque performance tailored to the designer’s preferences, even without prior knowledge of motor characteristics. We verify the obtained results through large-scale multi-case computations.
Asanuma et al. (Thu,) studied this question.