基于自适应修正拉盖尔递归神经网络的永磁直线同步电机反推控制

Backstepping Control Based on Adaptive Modified Laguerre ecurrent Neural Network for Permanent Magnet Linear Synchronous Motor

针对永磁直线同步电机(PMLSM)伺服系统易受参数变化和非线性外部扰动等不确定性因素影响,提出了一种基于自适应修正拉盖尔递归神经网络(AMLRNN)的反推控制方法。首先,建立了含有不确定性的PMLSM动态模型。然后,采用AMLRNN估计系统中的不确定性,通过基于李雅普诺夫稳定性理论的在线参数训练方法推导出两个最优学习速率来加速参数收敛。该方法可避免传统的自适应反推控制系统中存在的"微分爆炸"问题及抖振现象,使系统具有良好的瞬态性能和鲁棒性能。最后,通过实验证明了所提出的控制方案是有效可行的,与传统的自适应反推控制系统相比,基于AMLRNN的反推控制系统的控制性能更加优越,明显减小了系统的位置跟踪误差。

A backstepping control approach based on adaptive modified Laguerre recurrent neural network（AMLRNN） was proposed for permanent magnet linear synchronous motor（PMLSM） servo system which is vulnerable to influence of the uncertainties, such as parameter variations and nonlinear external disturbances. Firstly, the dynamic model of PMLSM with the uncertainties was established. And then, two optimal learning rates were derived by the on-line parameter training methodology based on the Lyapunov stability theorem to accelerate parameter convergence. This method can avoid the inherent problem of explosion of complexity and chattering phenomenon existed in the general adaptive backstepping control system, and make the system have good transient performance and robust performance. Finally, the experimental results confirm that the proposed scheme is effective and feasible. Compared with the general adaptive backstepping control system, the backstepping control system using AMLRNN has more superior control performance, and the position tracking error of system is obviously reduced.