基于BP神经网络遗传优化的电机推力品质研究  

Research on motor thrust quality based onBPNN genetic optimization

作　　者：周杨 赵静[1] 汪伟涛 王婉婉 ZHOU Yang;ZHAO Jing;WANG Weitao;WANG Wanwan(School of Electrical Engineering and Automation,Anhui University,Hefei 230601,China)

机构地区：[1]安徽大学电气工程与自动化学院,安徽合肥230601

出　　处：《传感器与微系统》2022年第1期58-61,共4页Transducer and Microsystem Technologies

基　　金：安徽省自然科学基金资助项目(1808085QE123);国家自然科学基金资助项目(51707002)。

摘　　要：为了提升激光切割机床上的直线电机(LMs)的推力品质,采用了一种基于显著因子筛选和深度学习算法的结构优化方法。首先,根据电机设计准则,分析极距、永磁体宽度、永磁体厚度、气隙长度和单边线圈宽度等参数对电机平均推力及推力波动的影响。其次,利用有限元仿真和试验设计,筛选出电机推力品质的显著影响因子,进而采用反向传播神经网络(BPNN)算法建立其非参数快速计算模型。最后使用遗传算法对关键因子进行多目标优选,得到提升推力品质的优化参数。优化后的结果验证了所提方法的优越性,平均推力提高33.9%,推力波动降低45.5%。To improve the thrust quality of the linear motors(LMs)on the laser cutting machine,a structural optimization method based on significant factor screening and deep learning method is adopted.Firstly,according to the motor design criteria,the influence of pole distance,permanent magnet width,permanent magnet thickness,air gap length and unilateral coil width on the average thrust and thrust fluctuation of the motor is analyzed.Secondly,by using the finite element simulation and experimental design,the significant influence factors of motor thrust quality are selected,and then the nonparametric fast calculation model is established by back propagation neural network(BPNN).Finally,genetic algorithm is used to optimize the key factors and get the optimization parameters to improve the thrust quality.The results after optimization verifies the advantages of the proposed method:the average thrust is increased by 33.9%;the thrust fluctuation is reduced by 45.5%.

关 键 词：直线电机 结构变量 显著因子筛选 反向传播神经网络 推力品质 

分 类 号：TH16[机械工程—机械制造及自动化] TM359.4[电气工程—电机]