基于Ansoft的永磁同步电机结构参数优化研究

发布时间：2018-09-05 13:27

【摘要】：对永磁同步电机(Permanent Magnet Synchronous Motor,PMSM)的研究不断加深,促进了电机优化设计的发展,也对优化算法提出了更高的要求。PMSM结构多样,加大了内部磁场的复杂程度,导致等效磁路法等传统方法无法达到需要的精度。电磁场数值分析虽然具备良好的精确性,但计算消耗过大,因此需要一种新型算法来缩短电机设计的周期。本文从PMSM基本尺寸的确定和传统性能分析方法开始,经过二维有限元法相关理论的介绍,过渡到电磁场数值分析方法,在此基础上利用电机设计分析软件建立电机的初始模型,并对电机性能进行分析计算。为改善电机性能,选取磁极厚度、极弧系数、气隙长度以及偏心距为设计变量,齿槽转矩、空载气隙磁密波形正弦畸变率为目标函数进行优化。首先通过仿真实验确定各变量的取值范围,然后设计正交试验来获取回归分析所需的样本空间;接着,分别建立2个目标函数基于支持向量机(Support Vector Machine,SVM)的响应面模型,并通过引入变异操作的微粒群优化算法(Particle Swarm Optimization,PSO)分别对2个目标函数进行单目标优化;将优化结果代入有限元软件,齿槽转矩由初始的4.37N?m降为0.264N?m,空载气隙磁密波形正弦畸变率由29.14%降为17.36%,仿真实验验证了结果的准确性;最后,鉴于实际的电机优化一般均为多目标优化问题,因此将两个目标函数放在一个优化过程中,采用PSO同时对其进行多目标优化,仿真后齿槽转矩为0.31N?m,空载气隙正弦畸变率为22.33%,两者均比较理想。“SVM+PSO”算法优化效果良好,可以保证较高的准确性;同时需要的样本空间小,寻优历经的进化代数少、收敛速度快;两种算法结合后延续了各自的先进性。此外,SVM回归分析采用了“黑箱方法”,大大降低了相关人员对电机知识的依赖,有效简化了电机设计过程;由于算法的通用性,也为电机其他性能参数的优化提供了指导与借鉴。
[Abstract]:The research on PMSM (permanent Magnet synchronous Motor) has been deepened, which has promoted the development of the optimal design of PMSM, and put forward higher requirements for the optimization algorithm. The structure of PMSM is diverse, and the complexity of the internal magnetic field is increased. As a result, traditional methods such as equivalent magnetic circuit method can not achieve the required accuracy. Although the numerical analysis of electromagnetic field has good accuracy, the calculation consumption is too large, so a new algorithm is needed to shorten the period of motor design. This paper begins with the determination of the basic dimensions of PMSM and the traditional performance analysis method, through the introduction of the relevant theory of two-dimensional finite element method to the electromagnetic field numerical analysis method, on this basis, the initial model of the motor is established by using the motor design and analysis software. The performance of the motor is analyzed and calculated. In order to improve the performance of the motor, the magnetic pole thickness, polar arc coefficient, air gap length and eccentricity are selected as design variables, the tooth slot torque and sinusoidal distortion rate of no-load air-gap magnetic density waveform are selected as objective functions to optimize. First, the range of values of each variable is determined by simulation experiments, and then orthogonal test is designed to obtain the sample space required for regression analysis. Then, two response surface models of objective functions based on support vector machine (Support Vector Machine,SVM) are established, respectively. The particle swarm optimization algorithm (Particle Swarm Optimization,PSO) with mutation operation is introduced to optimize the two objective functions, and the optimization results are added to the finite element software. The slotted torque is reduced from the initial 4.37N?m to 0.264 Nm, and the sinusoidal distortion rate of the no-load air-gap magnetic density waveform is reduced from 29.14% to 17.36. The simulation results verify the accuracy of the results. Therefore, the two objective functions are put into one optimization process, and the multi-objective optimization is carried out simultaneously by using PSO. The simulation results show that the slotting torque is 0.31 NM, and the sinusoidal distortion rate of no-load air gap is 22.33. Both of them are ideal. "SVM PSO" algorithm has good optimization effect. It can ensure higher accuracy; at the same time, the sample space is small, the evolution algebra is less, and the convergence speed is fast. The two algorithms combine to continue their advanced nature. In addition, SVM regression analysis adopts "black box method", which greatly reduces the dependence of related personnel on motor knowledge, and simplifies the process of motor design effectively. It also provides guidance and reference for the optimization of other performance parameters of motor.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM341

【参考文献】