永磁直线伺服系统递归小波Elman神经网络互补滑模控制

Wavelet-based Elman neural network complementary sliding mode control for permanent magnet linear servo system

针对永磁直线同步电机(PMLSM)直接驱动伺服系统易受参数变化、外部扰动和摩擦力等不确定性影响,而降低系统控制性能的问题,提出一种基于递归小波Elman神经网络(RWENN)的互补滑模控制方法。首先,建立含有不确定性的PMLSM动态模型;其次,采用积分滑模面和互补滑模面相结合设计互补滑模控制器。为解决互补滑模控制器参数选取困难的问题并估计系统存在的总不确定性,将互补滑模控制与RWENN相结合。利用RWENN代替互补滑模控制中的切换控制,RWENN可在线训练网络参数并实时调整参数。另外,为进一步提高鲁棒性,设计鲁棒补偿器对RWENN的参数逼近误差进行补偿。实验结果表明,该方法不仅降低了系统的抖振现象,保证了位置跟踪精度,还提高了系统的鲁棒性能。

Permanent magnet linear synchronous motor (PMLSM) direct drive servo system is susceptible to uncertainties such as parameter variations, external disturbances and frictions, and it reduces the control performance of the system. A complementary sliding mode control based on recurrent wavelet-based Elman neural network (RWENN) method is proposed to solve the problems. Firstly, a dynamic model of PMLSM with uncertainties was established. Then, complementary sliding mode controller with the combination of the integral sliding mode surface and the complementary sliding surface was designed. In order to solve the problem that the parameters of complementary sliding mode controller are difficult to be chosen, and estimating the lumped uncertainties in the system, complementary sliding mode control was combined with RWENN. RWENN was used to replace the switching control in complementary sliding mode control. RWENN can train the network parameters and adjust parameters on-line. In addition to further improve the robustness, a robust compensator was designed to compensate the parameter estimation errors of RWENN. The experimental results show that this method not only reduces the chatter of the system and guarantees the position tracking precision, but also improves the robustness of the system.