Neural network compensation of flexible satellite rapid maneuver

Abstract:

Employing the single-gimbal control moment gyroscope( SGCMG) as the main actuator, this paper deals with the rapid maneuver problem of flexible satellite. Firstly, a rigid-flexible coupled dynamics model with variable parameters and model uncertainties is established for the flexible satellite, and the mixed actuator consisting of CMG and RW is configured for agile satellite to meet the requirements of rapid maneuver ability and high precision. Then, a variable structure control law of the integral type is designed with a neural net- work to estimate the uncertainties and eliminate their effects on the system. Finally, the proposed algorithm is validated by simulation. Simulation results show that under external disturbances and model uncertainties, the designed system could suppress vibrations of flexible appendages and achieve the pointing accuracy and high stable accuracy required rapidly without encountering singularities of the actuator