Neural Network Control of Space Manipulator Based on Dynamic Model and Disturbance Observer

Abstract

Space flexible manipulators are convenient for performing on-orbit service; however, the vibration of the end effector is becoming increasingly serious because of the excessive length and mass of the arm. To solve this problem, in this paper, neural network control based on a flexible multibody dynamic model and disturbance observer is proposed. The dynamics model is based on the Lagrangian equation and assumed mode method, and also considers the position and attitude constraint equations of the flexible joint. The combination of a neural network controller and adaptive controller is introduced in detail, and a switching mechanism is added to improve the global stability of the system. Considering the joint module as an independent control system, a disturbance observer is added to the current loop of the control system, and a filter is combined to effectively suppress the influence of friction and dynamic coupling on joint control performance. The effectiveness of the proposed dynamic model and control scheme in terms of vibration suppression is verified in experiments on the self-designed space flexible redundant arm.

Keywords

• Disturbance observer
• dynamic model
• neural network control
• space manipulator
• vibration suppression