Rotor Multidisciplinary Optimization of High Speed PMSM Based on Multi-Fidelity Surrogate Model and Gradient Sequential Sampling

High-speed permanent magnet synchronous motor (PMSM) is widely used in industrial applications because of the high-efficiency, but the disadvantages which limit its development are still severe, which include the low yield strength of permanent magnet material and the poor rotor dynamics performance, thus the stress and modal analysis of the high-speed rotor are important. The traditional multidisciplinary optimization numerical analysis of the different disciplines is too time-consuming because the optimization process requires multiple calls to the numerical models of the motor. Therefore, a multi-fidelity (MF) surrogate model is proposed to balance the accuracy and efficiency of the multidisciplinary optimization process, which uses the analytical models to establish the large low fidelity (LF) sample space and use numerical models to establish the small high fidelity (HF) sample space; then uses the hybrid sample space to train the surrogate models. And in order to reduce the total number of the sample points, a gradient sequential sampling is proposed. That sampling method can make the sampling process more efficient as well. After that, the multi-fidelity surrogate models are used to optimize the rotor system under the NSGA-II algorithm. Finally, a prototype is manufactured to verify the final optimal point.