基于TD-NNLI的永磁同步电机转速辨识  

Speed identification of permanent magnet synchronous motor by using TD-NNLI method

作　　者：蒋彦[1] 刘国海[1] 赵文祥[1] 瞿沥[1] 

机构地区：[1]江苏大学电气信息工程学院,江苏镇江212013

出　　处：《电机与控制学报》2014年第2期62-68,共7页Electric Machines and Control

基　　金：国家自然科学基金(51077066;61273154;51277194);江苏省自然科学基金(SBK201221867);江苏高校优势学科建设工程项目

摘　　要：为获得无位置传感器控制系统中永磁同步电机(permanent magnet synchronous motor,PMSM)的转子速度,提出一种基于跟踪微分器—神经网络左逆(tracking differentiator-neural network left inversion,TD-NNLI)辨识方法。依据逆系统理论,构造一个以dq轴系下的电压电流及其导数为输入,以转速为输出的PMSM系统内含虚拟传感器。采用神经网络逼近该内含传感器的逆系统,将原系统串联在所得逆系统左侧辨识出电机转速。为降低系统噪声及外部扰动的影响,该神经网络的输入采用了由TD快速跟踪并降噪提取的系统状态量及其微分信号。仿真和实验结果表明,TD-NNLI观测器能够抑制系统噪声对于转速辨识的影响,解决了无速度传感器系统闭环控制的转速辨识问题。In order to identify rotor speed of permanent magnet synchronous motor （PMSM） in speed sen-sorless control system, a novel tracking differentiator-neural network left inverse （ TD-NNLI） observer was proposed. According to inversion system theory, as an assumed system, an inherent sensor of PMSM control system was presented, in which the current, the voltage and their derivatives in dq coordinates are the inputs, while the output is the rotor speed. The speed was estimated by taking neural network to ap-proximate inherent sensor inversion and sering original system to the left of inversion system. In order to reduce the impact of system noise and external disturbances, the inputs of neural network were system state variables and their differential signals which were quickly tracked and extracted by TD. The simula-ted and experimental results show that the TD-NNLI observer effectively restrains the impact of system noise for speed identification. The proposed observer offers excellent static and dynamic response and su-perior anti-interference performance, which verifies the validity of the method.

关 键 词：神经网络左逆 永磁同步电机 跟踪微分器 转速观测器 无位置传感器控制 逆系统 

分 类 号：TM301.2[电气工程—电机]