基于多支持向量机和粒子群算法的永磁同步直线电机优化设计

Design Optimization of Permanent Magnet Synchronous Linear Motor Based on Multiple Support Vector Machine

引入多支持向量机算法(MSVM),用以解决永磁同步直线电机(PMSLM)优化设计中的快速建模问题。在3D有限元分析的基础上,采用MSVM拟合直线电机结构参数与运行性能参数之间的非线性关系,对电机性能参数(推力、推理波动率、效率和谐波畸变率等)进行回归预测,该方法建立的模型精度可达到93%以上;引入粒子群算法(PSO)对MSVM电机模型进行寻优,得到一组最优的电机结构参数并建立有限元模型。仿真实验结果表明:采用MVSM建模并优化的电机推力大、推力波动小、峰值电流小、效率高,符合电机的优化设计目标。

Multiple support vector machine （MSVM） was introduced to solve the rapid modeling problem of permanent magnet synchronous linear motor design optimization. After the analysis of 3 D - FEA, using MSVM to map the relation of motor structure parameters and motor performances, building the rapid model for optimization calculation which can regress and predict the motor performances （thrust, thrust tipple, efficiency and harmonic distortion rate）, the accuracy of this model built by MSVM can up to 93 % and above;Particle Swarm Optimization （PSO） was introduced to optimize the MSVM motor model to get the motor best combination of structure parameters. After the simulation experiments based on FEA, the results indicate that： the PMSLM motor based on MSVM modeling opti- mization has high thrust, low thrust tipple, low peak current, high efficiency which can satisfy the demands of motor design optimization.