Robust Control of Robot Manipulator with Uncertain Parameters Based on Neural Network

Aiming at the control problem of space robot with load capture, considering the influence of manipulator joint flexibility and the unknown quality of non-cooperative target, a sliding mode variable structure control method based on least square method quality identification is proposed. Firstly, a non-cooperative target quality observer is designed. Then, the dynamic model of flexible joint space robot with floating base is established by using Lagrange method and momentum moment conservation principle. Based on singular perturbation theory, the system model is approximately decomposed into fast and slow subsystems for control respectively. A hyperbolic tangent sliding mode variable structure controller is designed for the slowly varying subsystem to compensate the motion error and ensure the control accuracy; A feedback compensation control algorithm based on velocity difference is designed for the fast changing subsystem to suppress the residual vibration generated by the flexible joint and improve the robustness of the system. Finally, experiments verify the effectiveness of the method.