文章精选 I 山东大学周涵等：同步磁阻电机机械应力及多目标优化分析研究

同步磁阻电机(SYNRM)的机械强度一直是一个很大的挑战。本文提出了一种评估应力等效性和磁桥应力相互作用的分析方法，以及一种多目标优化方法。考虑到桥体处复杂的磁阻结构和不可避免的应力集中，建立了适合SYNRM的有限元模型。初步建立了两输入、一输出、三层的神经网络结构。构造连续函数以提高精度。另外，等效应力可以转化为三维应力图的等值线分布。等值线分布说明了等效应力下磁桥长度的匹配方案。此外，本文还对磁桥相互作用应力的分析进行了探讨。定义了各磁桥长度对应的优化级别，并对各级别进行了有限元分析。采用田口法确定各磁桥上应力源的比重。在此基础上，提出了一种基于多目标粒子群优化(MOPSO)技术的多目标优化方法。以转子磁桥为设计参数，对气隙磁通密度、正弦特性、平均转矩、转矩脉动、机械应力等10个优化目标进行了优化。基于响应面法(RSM)可以得到优化目标与设计参数之间的关系，避免了实验样本过多的问题。将优化后的模型与初始模型进行了比较，验证了优化后的效果。最后，对额定工况下的温度分布进行了分析，为解决热应力问题提供了依据。     

The mechanical strength of the synchronous reluctance motor (SynRM) has always been a great challenge. This paper presents an analysis method for assessing stress equivalence and magnetic bridge stress interaction, along with a multiobjective optimization approach. Considering the complex flux barrier structure and inevitable stress concentration at the bridge, the finite element model suitable for SynRM is established. Initially, a neural network structure with two inputs, one output, and three layers is established. Continuous functions are constructed to enhance accuracy. Additionally, the equivalent stress can be converted into a contour distribution of a three-dimensional stress graph. The contour line distribution illustrates the matching scheme for magnetic bridge lengths under equivalent stress. Moreover, the paper explores the analysis of magnetic bridge interaction stress. The optimization levels corresponding to the length of each magnetic bridge are defined, and each level is analyzed by the finite element method. The Taguchi method is used to determine the specific gravity of the stress source on each magnetic bridge. Based on this, a multiobjective optimization employing the Multiobjective Particle Swarm Optimization (MOPSO) technique is introduced. By taking the rotor magnetic bridge as the design parameter, ten optimization objectives including air-gap flux density, sinusoidal property, average torque, torque ripple, and mechanical stress are optimized. The relationship between the optimization objectives and the design parameters can be obtained based on the response surface method (RSM) to avoid too many experimental samples. The optimized model is compared with the initial model, and the optimized effect is verified. Finally, the temperature distribution of under rated working conditions is analyzed, providing support for addressing thermal stress as mentioned earlier.

周涵，1999年5月出生于中国。他于2021年在山东理工大学获得电气工程及其自动化学士学位。他目前在山东大学电气工程学院攻读博士学位。主要研究方向为永磁电机的计算、分析与控制。

王秀和，1967年7月出生于中国。他分别于1989年和1993年在山东大学获得学士和硕士学位，并于1996年在沈阳工业大学获得博士学位。2001年至2002年，任韩国首尔国立大学博士后。现任山东大学电气工程学院副院长、教授。主要研究方向为永磁电机、节能电机、电磁器件的理论分析计算及其在电机上的应用。他撰写/合作撰写了100多篇关于这些主题的论文。