基于神经网络的电液位置伺服系统自适应滑模控制  

Neural Network Based Adaptive Sliding Mode Control for Electro-Hydraulic Position Servo Systems

作　　者：党选举[1] 袁永全 姜辉[1] 伍锡如[1] 李珊 

机构地区：[1]桂林电子科技大学电子工程与自动化学院,广西桂林541004 [2]桂林电子科技大学教学实践部,广西桂林541004

出　　处：《机床与液压》2018年第1期126-129,共4页Machine Tool & Hydraulics

基　　金：国家自然科学基金资助项目(61263013);广西信息科学实验中心项目(20130110);广西自然科学基金资助项目(2014GXNSFBA118275;2015GXNSFAA139297);智能综合自动化高校重点实验室基金资助项目(2016)

摘　　要：针对电液位置伺服系统(阀控对称缸)模型复杂、参数时变、摩擦影响显著等特点,提出了基于RBF神经网络和基于Lu Gre模型的自适应滑模控制算法。该算法的优点是:(1)利用RBF神经网络逼近控制电流与系统输出压力的关系,将电液位置伺服系统的数学模型简化为二阶,减少了模型参数;(2)采用Lu Gre模型能够准确地描述摩擦过程中复杂的动、静态特性,通过该模型设计摩擦补偿,提高了控制精度;(3)设计自适应滑模控制器,增强了系统的鲁棒性。利用构建的李雅普诺夫函数,证明了闭环系统的稳定性。仿真实验表明:所提算法控制精度较高、响应速度较快、鲁棒性较强。The adaptive sliding mode control algorithm is proposed based on Radial Basis Function （RBF） neural network and LuGre friction mode for the electro-hydranlic position servo system （valve-controlled symmetrical cylinder）, which is model complex, time-varying of parameters and the friction influence significantly in characteristics. The advantages in this algorithm ： （ 1 ） The mathematical model of it could be simplified to the second order and be reduced the model parameters via RBF neural network approximation the relationship between control current and system output pressure. （2） LuGre model could accurately describe the complex of dynamic and static characteristics in the friction process. Designing the friction compensation based on this model could improve the control precision. （3） The design of adaptive sliding mode controller enhanced the robustness of the system. The stability of closed-loop system was proved by constructing Lyapunov function. Simulation experiments show that the proposed algorithm has high control precision, fast response, and strong robustness.

关 键 词：电液位置伺服系统 RBF神经网络 LUGRE模型 自适应 滑模 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TH137[自动化与计算机技术—控制科学与工程]