基于神经网络的伺服机械手LuGre摩擦补偿控制

LuGre Friction Compensation Control of Servo Manipulator Based on Neural Network

针对伺服机械手系统的LuGre摩擦模型参数辨识难,难以建立其精确的数学模型,利用径向基函数(RBF)神经网络的万能逼近特性逼近LuGre摩擦,并作为计算转矩控制器的补偿项.通过Lyapunov方法证明了系统的稳定性以及闭环系统跟踪误差的收敛性.仿真结果证明控制算法能对摩擦进行有效补偿,提高了伺服机械手系统的轨迹跟踪控制性能.

To overcome the parameter identification difficulties of the LuGre friction model, and it is noteasy to establish an accurate mathematical model, RBF neural network was used to approximate theLuGre friction model, and was combined with the computed torque controller. The stability of the systemand the convergence of the tracking error of the closedloop system were proved by the Lyapunov method.The simulation results show that the control algorithm can compensate the friction effectively and improvethe tracking control performance.