The gimbal servo system of a single gimbal control moment gyroscope (SGCMG) on small satellites is often affected by various factors, including gyro torque, electromechanical coupling effects, and tribological properties. These factors significantly degrade the servo performance of the gimbal, impacting its operational precision and, consequently, the attitude adjustment of the satellite. To address these issues, this paper considers the impact of bearing flexibility and establishes an electromechanical tribo-dynamic model for the SGCMG. Furthermore, an Active Disturbance Rejection Control (ADRC) method based on feedforward compensation is proposed for the gimbal servo system, incorporating dynamic model information as compensation in both the controller and the extended state observer (ESO). Given the limitations of controller hardware, a trade-off exists between speed and precision in solving the dynamic model. To address this, a deep neural network is employed to accurately map the model's input and output, achieving fast and accurate data-driven feedforward compensation within ADRC. This method enhances the system's disturbance rejection capability and servo tracking performance. Finally, through simulation and experimental studies, the effectiveness and superiority of the proposed algorithm are validated, demonstrating its potential for improving the performance and reliability of SGCMG systems on small satellites.
Liu et al. (Wed,) studied this question.