The flux switching permanent magnet (FSPM) motor is relatively a new topology of the permanent magnet (PM) motors, which both PM and armature winding are placed at the stator. This feature leads to more robust design, better heat dissipation, and a proper option for a wide range of industrial applications. The critical part of the development of FSPM motor is the design optimization of the structure to improve the electromagnetic performance of the motor. In this article, first to reduce the computation time and required memory of the optimization procedure, the multiobjective sensitivity analysis based on design of experiment is performed to specify the most effective parameters on the objectives. Then, the initial samples data of the optimization procedure is obtained by 2D finite element method (FEM) model of the FSPM motor, which is validated by the prototype of the motor. Furthermore, based on FEM results the multilayers perceptron artificial neural network for the approximation of relation between design variables and objectives is implemented. Finally, using the nondominated sorting genetic algorithm‐II, the optimization procedure of the FSPM motor is done. The accuracy of the presented optimization procedure is validated by a comparison of the initial prototype and final design of the motor.

中文翻译：

---

基于MLP-ANN建模和NSGA-II算法的磁通切换永磁电动机灵敏度分析和多目标设计优化

磁通切换永磁（FSPM）电动机是永磁（PM）电动机的相对较新的拓扑结构，其中PM和电枢绕组都放置在定子上。这个特性导致了更可靠的设计，更好的散热，并为广泛的工业应用的一个合适的选择。FSPM电机开发的关键部分是结构的设计优化，以提高电机的电磁性能。在本文中，首先为了减少优化过程的计算时间和所需的内存，基于实验设计进行了多目标灵敏度分析，以指定目标上最有效的参数。然后，通过FSPM电动机的2D有限元方法（FEM）模型获得优化程序的初始样本数据，通过电机原型验证。此外，基于有限元结果，实现了多层感知器人工神经网络，用于近似设计变量与目标之间的关系。最后，使用非支配排序遗传算法-II，完成了FSPM电机的优化程序。通过比较电动机的初始原型和最终设计，可以验证所提出的优化程序的准确性。