Design Optimization of PMaSynRM Using Kriging Surrogate Modeling and NSGA-II Integrated with Taguchi Method

Permanent Magnet Assisted Synchronous Reluctance Motors (PMaSynRMs) are emerging as viable alternative to PMSMs, particularly in applications where cost and material availability are critical. Rotor geometry and magnet placement pose challenges and can be effectively handled through design optimization for improved motor performance. The process of improving motor performance through FEA simulations using multi-objective optimization is complex and time-consuming. This paper proposes a new motor design approach to improve the performance of a PMaSynRM with minimal computational cost by using Kriging surrogate modeling integrated with the Non-Dominated Sorting Genetic Algorithm-II (NSGA-II). Initially, the critical rotor parameters influencing the performance characteristics of PMaSynRM, such as torque, torque ripple, and saliency, are identified. These parameters are sampled within their bounds using a five-level Taguchi orthogonal array, ensuring a balanced exploration of factor levels. A design of experiments (DOE) is carried out using these samples and the resulting responses are regarded as true responses. A Kriging surrogate model is also created with these samples to predict the responses and its accuracy is validated against the true responses. To demonstrate the effectiveness of the proposed optimization approach, this paper presents the design of a 1.5 kW, 36-slot 6-pole PMaSynRM with a fluid flux barrier rotor. The proposed design approach achieves an improvement in rotor saliency by 2.5 times and torque ripple reduction by 6.3 % compared to one of the recent works.