基于Adaline神经网络参数辨识的PMSM鲁棒电流预测控制

Robust current predictive control for PMSM based on Adaline neural network parameter identification

针对复杂工况下永磁同步电机存在模型参数失配导致控制系统性能下降的问题,提出一种基于参数在线辨识的鲁棒电流预测控制方法。首先,建立永磁同步电机预测控制模型,详细分析电磁参数失配对电机响应电流及输出转矩和转速的影响。然后,设计了基于Adaline神经网络的参数在线辨识器,并在传统的权值调整算法上,提出一种应用于电机参数辨识系统的新型动态混合最小均方算法。最后,利用在线辨识的参数来实时更新电流预测控制器中的参数,以避免参数失配对控制系统性能的影响。通过仿真和实验验证了所提方法和新型算法的可行性和有效性,其结果表明了该方法不仅能够实现精准在线跟踪电机参数的变化,而且有效抑制了参数失配导致的响应电流偏差。

For the problem that the model parameter mismatch of permanent magnet synchronous motor(PMSM)under complex working conditions causes the performance of the control system to degrade,this paper proposes a robust current predictive control method based on parameter online identification.First,a predictive control model of PMSM was established,and the influence of electromagnetic parameter mismatch on motor response current,output torque and speed was analyzed in detail.Then,an online parameter identifier based on Adaline neural network was designed,and a dynamic hybrid least mean square(NDH-LMS)algorithm was proposed for motor parameter identification system based on the traditional weight adjustment algorithm.Finally,the parameters identified online were used to update the parameters in the current prediction controller in real-time to avoid the impact of parameter mismatch on the performance of the control system.The feasibility and effectiveness of the proposed method and the proposed algorithm were verified by simulation and experimental.The results show that the proposed method can not only accurately track the changes of motor parameters online,but also effectively suppress the response current deviation caused by parameter mismatch.