无位置传感器开关磁阻电机的转矩波动抑制

Torque Ripple Minimization in a Position Sensorless Switched Reluctance Motor

针对开关磁阻电机驱动系统中位置传感器及转矩脉动的存在限制其应用范围的问题,提出利用网络结构简单、学习效率高的柔性神经网络对其进行建模,进而对开关磁阻电机无位置传感器控制下的转矩波动抑制方法进行了研究.建立了2个柔性神经网络:第1个完成由电机绕组的相电流、相磁链到转子位置之间的非线性映射,将离线训练好的网络用于转子位置的在线估计;第2个完成由电机期望转矩、转子位置到期望的优化电流之间的非线性映射,通过电流跟踪控制,使电机相电流跟随优化电流变化,实现转矩波动抑制.仿真和实验结果显示,柔性神经网络的收敛速度为传统神经网络的6倍,从而实现了转子位置快速的估算,估算误差控制在[-4,°4°]内,同时转矩波动降低1000/,验证了该控制策略较之于传统控制方法的优势.

The application of the switched reluctance motor （SRM） drive is limited by its rotor position sensors and torque ripple. In view of the problem, a method for torque ripple minimization in a position sensorless control of SRM was proposed by modeling based on flexible neural networks （FNN） , which have less nerve cells and quicker learning speed. Two FNNs were built： the first one estimated the rotor position through measurement of the phase flux linkages and phase currents; the second one estimated the reference currents with a desired torque and rotor position, then the real armatures currents were adjusted according to the reference values, therefore the torque ripple was minimized. Simulation and experimental results illustrated that the learning speed of FNN was 6 times faster than the traditional neural network, so the quick estimation of the rotor position was realized with a low error [-4° ,4°] and the torque ripple was reduced by 10%. So this control approach obtained better performance than traditional way.