异步电机直接转矩控制定子电阻辨识研究

摘要

在直接转矩控制中异步电机低速运行时,定子电阻易受温度变化及电流频率波动等影响发生变化,进而对系统性能造成不利影响.定子电阻辨识是解决上述难题的有效方法,但由于定子电阻变化未知且与电机其余参量存在严重的耦合关系,所以定子电阻在线辨识很困难,常规辨识方法作用有限.提出利用对角递归神经网络的定子电阻辨识策略.首先推导出一个仅受定子电阻变化影响的速度观测器,当定子电阻发生变化时,观测器测出的转速与实际转速会出现转速误差.构造一个对角递归神经网络,从转速误差中辨识出定子电阻.仿真结果表明,辨识算法效果显著且优于BP网络的作用效果.采用上述算法可有效的解决异步电机直接转矩控制低速性能差的难题.%In the direct torque control of asynchronous motor at low speed,the stator resistance is susceptible to temperature changes and current frequency fluctuations,which can adversely affect the performance of the system.Stator resistance identification is an effective method to solve the problem,but because the stator resistance variation is unknown and the rest parameters of the motor are serious coupling,the stator resistance online identification is difficult and conventional identification methods are limited.Therefore,the stator resistance identification strategy based on diagonal recurrent neural network is proposed.First,a speed observer which is affected by the variation of the stator resistance is designed.When the stator resistance changes,the observer measured speed and actual speed will appear an error.And then,a diagonal recurrent neural network is constructed to identify the variation of the stator resistance from the error.The simulation results show that the proposed algorithm is reliable and better than BP network.The algorithm can effectively solve the problem of low speed performance of direct torque control.