SRM的神经网络模型参考自适应控制

Neural Networks based Model Reference Adaptive Control for SRM

针对开关磁阻电机的非线性特性,将具有非线性映射能力及自适应能力的BP神经网络应用于开关磁阻电机驱动系统,提出一种基于神经网络的开关磁阻电机在线辨识与自适应控制方法。该方法构造了一个BP网络对系统进行在线辨识,建立其在线参考模型,并为BP神经网络控制器提供了梯度信息,由控制器完成参数的自学习,从而实现参数的在线调整。仿真结果表明该控制系统具有响应迅速、适应性强等优点,具有较高的控制精度和较好的鲁棒性。

Aiming at the nonlinear electromagnetic characteristic of switched reluctance motor （SRM）,a new control solution is presented. This method uses BP neural network which has powerful nonlinear mapping and selfadaptive ability to identify the system online,constructs the on -line reference model,and then give gradient information to BP neural network controller. Through self learning of controller parameters by neural network controller,thus achieve on-line regulation of controller＇s parameters. The simulation result shows that with the method given in this paper,excellent flexibility and adaptability as well as high precision and good robustness are obtained.