Design of Adaptive Neural Network Controllers for LEO Drag-free Satellite

Low-disturbance environment can be achieved by the LEO(Low-Earth Orbit)drag-free satellite,which benefits the validation of relativity,detection of gravitational waves and measurement of gravity field.For drag-free control purpose,most researches proposed controllers with linearized model and ignoring the nonlinear characteristics,which lower the accuracy of controllers.In this paper,by taking into account of the nonlinear characteristics,an adaptive neural network controller is established based on Lyapunov methods and adaptive backstepping control theory.For nonlinear characteristics and unmodeled dynamics,RBF neural network is employed for approximation.At the same time,we introduce the update laws of adaptive neural network weights,which guarantee the stability of the closed-loop system and satisfy requirements of the drag-free satellite control system.The simulation results indicate that the controller is effective and the accuracy of the drag-free satellite can be satisfied.