电工技术学报  2018, Vol. 33 Issue (zk1): 41-48    DOI: 10.19595/j.cnki.1000-6753.tces.180394
电机与电器 |
永磁同步电机性能分析的典型深度学习模型与训练方法
金亮, 王飞, 杨庆新, 汪冬梅, 寇晓斐
天津工业大学天津市电工电能新技术重点实验室 天津 300387
Typical Deep Learning Model and Training Method for Performance Analysis of Permanent Magnet Synchronous Motor
Jin Liang, Wang Fei, Yang Qingxin, Wang Dongmei, Kou Xiaofei
Tianjin Key Laboratory of Advanced Technology of Electrical Engineering and Energy Tianjin Polytechnic University Tianjin 300387 China
全文: PDF (19702 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 引入人工智能的深度学习算法,建立永磁同步电机齿槽转矩预测分析模型,为解决电机设计、应用特性与系统集成分析间的数据孤岛问题奠定基础。选取永磁同步电机的4个结构参数(极弧系数、气隙长度、永磁体厚度、永磁体宽度)与齿槽转矩的性能关系作为研究对象,使用有限元方法建立8对极、48定子槽的内置式“V”型永磁同步电机模型并进行了仿真分析,得到了结构参数与齿槽转矩的625组数据。人工智能深度学习算法的预测模型为4输入、单输出、4隐层的结构。625组数据中的575组用来训练深度学习预测模型,50组用来测试预测模型的泛化能力。通过比较有限元计算的样本数据与深度学习预测模型的预测结果,验证了人工智能深度学习预测模型的可行性。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
金亮
王飞
杨庆新
汪冬梅
寇晓斐
关键词 永磁同步电机有限元法人工智能深度学习齿槽转矩    
Abstract:This paper introduces the deep learning algorithm of artificial intelligence, and establishes the cogging torque prediction analysis method of permanent magnet synchronous motor (PMSM). This lays a foundation for solving the problem of data isolation between motor design, application characteristics and system integration analysis. The relationship between the four structural parameters of PMSM which are air gap, pole embrace, magnet thickness and magnet width and the performance of the cogging torque is selected as the research object. The PMSM is with eight pole-pairs, 48 stator slots. Then, 625 sets of training samples of deep learning about cogging torque are generated by FEM. We build a prediction model with 4 inputs, one output, and 4 hidden layers, which is trained and optimized by using the artificial intelligence deep learning algorithm. The 575 groups in the 625 sets of data are used to train the deep learning prediction model, and the 50 group is used to test the generalization ability of the prediction model. The feasibility of artificial intelligence deep learning prediction model is verified with the FEM.
Key wordsPermanent magnet synchronous motor    finite element method    artificial intelligence    deep learning    cogging torque   
收稿日期: 2018-03-22      出版日期: 2018-09-26
PACS: TM351  
基金资助:国家自然科学基金面上项目(51577132)和天津市高等学校创新团队培养计划(TD13-5040)资助
通讯作者: 杨庆新 男,1961年生,博士,教授,博士生导师,研究方向为工程电磁场与应用。E-mail:qxyang@tjpu.edu.cn   
作者简介: 金 亮 男,1982年生,博士,副教授,研究方向为工程电磁场与磁技术、电磁场云计算和电磁无损检测等。E-mail:jinliang@tjpu.edu.cn
引用本文:   
金亮, 王飞, 杨庆新, 汪冬梅, 寇晓斐. 永磁同步电机性能分析的典型深度学习模型与训练方法[J]. 电工技术学报, 2018, 33(zk1): 41-48. Jin Liang, Wang Fei, Yang Qingxin, Wang Dongmei, Kou Xiaofei. Typical Deep Learning Model and Training Method for Performance Analysis of Permanent Magnet Synchronous Motor. Transactions of China Electrotechnical Society, 2018, 33(zk1): 41-48.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.180394          https://dgjsxb.ces-transaction.com/CN/Y2018/V33/Izk1/41