摘要
研究了一种高斯基函数CMAC神经网络用于克服液压伺服系统中摩擦非线性的控制器设计问题。该控制器不需要对摩擦力进行建模 ,而是采用学习的方法克服摩擦非线性 ,因此具有通用性。仿真结果表明 ,该种控制器不仅是有效的 。
A learning controller based on the CMAC (Cerebellar Model Articulation Controller) neural network with Gaussian basis functions is proposed to attenuate the influence of friction in electrohydraulic position servo systems. The learning controller treats the friction by learning, which needs not a priori knowledge of the control objectives. This feature qualifies it to be easy to apply to any servomechanism, particularly when the friction is large and erratic and difficult to model. The performance of the proposed controller is evaluated through simulations in a typical electrohydraulic position servo system subjected to nonlinear friction. The results demonstrate that the proposed controller is not only effective, but also improves the positioning accuracy of the system.
出处
《机床与液压》
北大核心
2000年第3期24-25,共2页
Machine Tool & Hydraulics
