永磁同步电机神经网络逆解耦控制研究

针对永磁同步电机的非线性、多变量、强耦合的特点,将神经网络与逆系统解耦方法相结合,并用于永磁同步电机的解耦控制.分析永磁同步电机的数学模型与解析逆模型,完成系统可逆性证明,将永磁同步电机与解析逆系统等效成两个伪线性子系统,构造神经网络逆系统,将永磁同步电机动态解耦为一阶线性磁链子系统与二阶线性转速子系统,利用两个PID控制器对伪线性子系统进行闭环控制器设计,实现系统转速与定子磁链动态解耦控制.利用dSPACE半物理仿真系统完成神经网络训练数据的采集与系统解耦控制实验.结果表明神经网络逆系统方法可以实现永磁同步电机的高新能控制,对负载扰动具有较强的鲁棒性.

Decoupling control of PMSM based on artificial neural network inverse method

A decoupling control method based on artificial neural network(ANN) inverse system theory was applied for the permanent magnet synchronous motor(PMSM),which was a nonlinear and high coupling system.The analytical inverse system of PMSM was obtained by analyzing the reversibility of the mathematical model,which is constituted by two pseudo-linear subsystems,first-order linear flux subsystem and second-order speed subsystem.The dynamic decoupling control between flux and speed of PMSM were realized by using PID algorithm to design closed-loop controller for the two subsystems.The dSPACE platform was built to realized the ANN inverse system control method for a real PMSM,and to obtain the data,which was sampled to train the ANN.The results show that ANN inverse system control strategy was of a high dynamic performance for PMSM,even though there were different ways of load torque disturbance.