基于神经网络的电液位置伺服系统自适应滑模控制

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。

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

The adaptive sliding mode control algorithm is proposed based on Radial Basis Function (RBF) neural network and LuGre friction mode for the electro hydraulic 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.