基于人工神经网络的车载异步电机参数辨识

为精确获取车载异步电机在不同运行状态下的参数,将人工神经网络应用到电机的参数辨识中。基于异步电机数学模型建立线性神经网络,神经网络的输入、输出包括电机定子电压、电流和转速,定子电流和转速通过传感器获得,定子电压通过重构占空比获得。使用最小均方差法求取此神经网络的权值矩阵,并由权值矩阵得到电机不同运行状态下的参数。最后将参数表写入控制算法,并利用电驱动系统测试平台进行控制验证,良好的转矩特性证明了算法有效性。

Parameter Identification of Asynchronous Motor for Vehicles Based on Artificial Neural Network

In order to acquire the precise parameters of asynchronous motor used in vehicle under different operating conditions, the artificial neural network is applied to identify the parameter of the motor. Based on the mathematical model of asynchronous motor, a linear neural network is established. The input and output of the neural network include the stator voltage, stator current and speed of the motor. The stator current and speed are obtained by the sensor, and the stator voltage is reconstructed by duty cycle. The Least Mean Square algorithm is used to obtain the weight matrix of the neural network, and all the parameters of the asynchronous motor under different operating states are obtained through the weight matrix. At last, the parameters table is written into the control algorithm, and control effect of motor is verified by the electric drive system test platform, the good torque characteristics prove effectiveness of the algorithm.