永磁同步直线电机的小波神经网络控制

摘要

For the tracking performance of permanent magnet linear synchronous motor ( PMLSM ) seriously influenced by ripple force, friction force and other disturbances, a method of control based on wavelet neural network (WNN) is proposed. By theoretical analysis, it was proved that the selected wavelet function approximated nonlinear ripple force with desired accuracy. On the basis of the combined feedforward plus PID control, a wavelet neural network was used to estimate the disturbances voltage. The experimental tracking error of the control based on wavelet neural network is 0. 15 mm, and the precision is 0. 75 percent. The experimental results show the efficiency of the proposed method, compared with combined feedforward plus PID control and neural network adaptive inverse control. The proposed method can not only effectively improves tracking precision and robustness, but also preferably achieves effective disturbance compensation.%针对永磁同步直线电机跟踪性能易受推力波动、摩擦力等干扰严重影响的问题,提出了基于小波神经网络的控制方法.分析证明了所构建的小波神经网络能以任意精度逼近主要干扰:非线性推力波动干扰.并且在复合前馈PID控制方法的基础上,利用小波神经网络实现了对永磁同步直线电机干扰的在线估计补偿.小波神经网络的控制实验的跟踪误差为0.15 mm,精确度为0.75％,实验结果表明,与复合前馈PID控制方法和神经网络自适应逆模型控制方法相比,基于小波神经网络的控制方法有效地提高了系统的跟踪性和鲁棒性,并能有效消除干扰对系统的影响.