永磁直线同步电机递归神经网络补偿器仿真研究

Research on Simulation of Recurrent Neural Network Compensator for Permanent Magnet Linear Synchronous Motor

针对永磁直线同步电机驱动系统没有机械阻尼、抗扰动性能差的缺点,为了抑制参数及负载变化对控制系统影响,采用扰动观测器进行补偿,该观测器能较好地补偿参数及负载在小范围内的变化,但不能有效地处理过大参数变化量.为了加大永磁直线同步电机驱动系统在参数变化和负载干扰时的控制性能,提出以递归神经网络作为补偿器来取代扰动观测器.仿真表明,当系统参数动态变化或受到负载变化的影响时,利用递归神经网络来在线动态地调整网络的参数,系统仍将具有很好的动静态性能.

The driving system of permanent magnet linear synchronous motor has the disadvantage of zero mechanical damping and weak anti-disturbance. In order to restrain disturbance introduced by the parameter variation and external load of the PMLSM control system. Disturbance observer is used for compensation. Compensator is effective under small range of parameter variations and external load disturbance, but not for great parameter variations. To increase the control performance of the PMLSM driving system under the occurrence of parameter variations and external load disturbance,a recurrent neural network compensator is proposed to replace the disturbance observer. The simulation results show good performances of the system by using recurrent neural network to adjust the parameters of neural network on-line dynamically on the condition of variation of system parameters and the impact of external load.