Rotation tracking control strategy for space flexible structures based on neural network compensation

Abstract

Flexible space structures play an invaluable role in space operation tasks as large attachments to spacecraft and artificial satellites. Due to the characteristics of flexibility and slenderness, flexible space structures will inevitably cause flexible deformation and vibrate during motion, which will affect the rotation control accuracy. To solve this problem, this paper proposes a rotation tracking control strategy based on sliding mode controller with neural network compensation. The control law is designed according to the dynamic model established by the new flexible deformation description method. Firstly, according to the new flexible deformation description method, the dynamic model of space flexible structures is established by the assumed modal method and Lagrange principle. Through a comparison of the accuracy of the dynamics model deformation, it is found that the new flexible deformation description method has high model accuracy. Based on the space flexible structures’ dynamics model, the sigmoid function is proposed as the convergence law for designing the sliding mode controller. Besides, neural networks are proposed to identify and compensate for the disturbing torque caused by ignoring nonlinear factors. Finally, the effectiveness of the sliding mode controller with neural network compensation is validated by simulation and ground control experiments.