定位平台非线性摩擦的神经网络补偿  

Nonlinear friction compensation based on neural networks for the positioning table

作　　者：王兴松[1] 张正峰[1] 王中华[2] 周香[1] 

机构地区：[1]东南大学机械工程系,南京210096 [2]济南大学信息科学与工程学院,济南250022

出　　处：《东南大学学报（自然科学版）》2002年第4期605-609,共5页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目 (5 9885 0 0 2 );高等学校博士学科点专项科研基金资助项目 (980 2 862 5 )

摘　　要：提出了一种基于BP神经网络的机械伺服系统非线性摩擦的补偿方法 ,根据该方法设计出一种将经典的PD控制与神经网络控制相结合的控制器 .该控制器既有PD控制的优点 ,又有神经网络逼近非线性函数的能力 ,较好地补偿了系统中的非线性摩擦和外部扰动 .应用Lyapunov稳定性定理 ,证明了系统的稳定性 ,并得到系统跟踪误差的边界值 .采用刚毛摩擦动力学模型 ,对X Y定位平台进行仿真和实验 .结果表明该控制器能够补偿系统的非线性因素 ,保证了系统的稳定 ,减小了跟踪误差 .该方案控制效果明显优于PD控制 。A BP neural networks based compensation method for nonlinea friction in mechanical servo systems is presented. According to the proposed method the controller, which is composed of a traditional PD controller and an NN controller, is designed. It has both the merits of the PD controllers and the general approximation property of the neural network, and can be employed to compensate the nonlinear friction and other disturbances. The stability of this method was proved using Lyapunov stability theory, and the boundary of the tracking error was derived as well. By using bristle dynamic friction model, simulations and experiments were conducted on an X-Y positioning table. The experiment results demonstrate that the controller can compensate the nonlinear factors in the system, guarantee the system stability and diminish the tracking error. Therefore it can be applied in industries with better performance.

关 键 词：神经网络 定位平台 非线性摩擦 补偿系统 机械系统 PD控制 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置] TH117[自动化与计算机技术—控制科学与工程]