基于遗忘因子随机梯度永磁同步电动机参数辨识  

Forgetting Factor Based Stochastic Gradient Identification for Permanent Magnet Synchronous Motor

作　　者：徐鹏[1,2] 肖建[1] 李山[2] 彭小峰[2] 

机构地区：[1]西南交通大学,成都610031 [2]重庆理工大学,重庆400054

出　　处：《微特电机》2014年第4期1-3,7,共4页Small & Special Electrical Machines

基　　金：国家自然科学基金资助项目(51177137);重庆市科委自然科学基金项目(cstc2012jjA40066);重庆市教委科学技术研究项目(KJ130807)

摘　　要：永磁同步电动机具有响应快、精度高、转矩比高等诸多优点。在永磁同步电动机系统数学模型基础上,构建系统回归模型,并采用遗忘因子随机梯度算法(FSG)辨识回归模型参数。仿真实验结果表明FSG算法对永磁同步电动机系统的参数辨识一致收敛,和随机梯度算法(SG)相比,FSG算法对输出非敏感参数值辨识收敛速度和精度方面均有较大优势。Permanent magnet synchronous motor（PMSM） has some excellent features,such as fast response,better accuracy,high torque to current ratio.Based on analysis of PMSM mathematical model,the system regression model was proposed,and forgetting factor based stochastic gradient algorithm was used for the identification of regression model parameters.Simulation results show that FSG algorithm was uniformly convergence for the parameters of PMSM.Comparing to SG algorithm,FSG algorithm has more outstanding performance on convergence speed and precision for non-sensitive parameter.

关 键 词：永磁同步电动机 随机梯度 遗忘因子 一致收敛 

分 类 号：TM351[电气工程—电机] TM341