基于小脑模型的电液位置伺服系统在线学习控制研究

针对非线性电液位置伺服系统的跟踪控制问题，提出了一种采用小脑模型（ＣＭＡＣ）神经网络的在线学习控制方法，与传统的ＣＭＡＣ控制器不同，该控制器采用动态误差作为ＣＭＡＣ神经网络的激励信号，从而使基于ＣＭＡＣ的控制器跟踪连续变化的信号成为可能，给出了具体的控制结构和算法，仿真结果表明，该控制器具有良好的处理非线性及跟踪连续变化信号的能力，并对时变外负载干扰具有明显的抑制作用，而且新型控制器能和较高的学习

On-Line Learning Control for Electrohydraulic Position Servo Systems Based on CMAC Neural Network

An on line learning control method based on the cerebellar model articulation controller (CMAC) neural network is presented for the tracking control problem of the electrohydraulic position servo systems. Unlike conventional CMAC based control methods, the proposed method uses the differences between ideal trajectory and output observations of the system as exciting signals for the CMAC instead of using them directly. This makes it possible for the CMAC based controller to track continuous signals without divergence. The simulation results show that the proposed method is of good robustness and tracking property for the nonlinear electrohydraculic position servo systems with time varying external load disturbance. Moreover, the method can use a high learning rate, which qualifies it to be used for learning on line.