Spacecraft Attitude Control Law Solution Using SDE-NET Network for Time-varying of Inertia

In order to enhance the spacecraft's random maneuvering capability during lunar missions and strive to improve the precision of spacecraft attitude measurement and control, this paper addresses the control challenges arising from unknown or varying moment of inertia during the transfer of materials between Earth and the Moon. A novel integrated attitude control method based on variable deflection angles and the Stochastic Differential Equation Network (SDE-Net) is proposed for uncertainty estimation of the moment of inertia model. Firstly, the ATT-TO-TOR configuration is designed to achieve integration of attitude measurement and control. This approach not only estimates unknown moment of inertia but also obtains boundary solutions for the original attitude maneuvering target after payload changes. Secondly, an incremental sub-transformation is designed, treating the increment as the state evolution of a stochastic dynamical system. A Brownian motion term is introduced to capture cognitive uncertainty, addressing the degradation of precision in traditional pseudoinverse steering laws under high dynamic conditions due to the complexity of matrix pseudoinverse operations. The results demonstrate that the proposed ATT-TO-TOR converges faster than traditional control algorithms under external disturbances. Simulation results reveal the superiority, robustness, and adaptability of the proposed controller in spacecraft attitude control under time-varying moment of inertia. Copyright © 2023 by the International Astronautical Federation (IAF). All rights reserved.