Robust Multiplexed MPC With Constraint Tightening for Satellite Attitude Control and Reaction Wheel Desaturation

Abstract

Since reaction wheels are sensitive and vulnerable devices, satellites often face significant challenges such as reaction wheel failure and angular momentum saturation during long-term operation. To address these issues, this paper proposes a multiplexed model predictive control (MMPC) method for satellites equipped with two reaction wheels and three magnetorquers, which can achieve both precise attitude control and angular momentum desaturation. Moreover, considering the presence of disturbance torques acting on satellites, the proposed MMPC algorithm is endowed with robustness through the constraint tightening approach, where the set of unknown but bounded disturbance torques is represented using zonotopes. Compared with conventional MPC algorithms that update all control variables simultaneously, the MMPC algorithm described in this paper employs asynchronous control moves on each input channel. By dividing the online optimization into smaller problems for each channel, the MMPC scheme reduces computational complexity and supports faster sampling in multivariable systems, thereby enabling quicker responses to unmeasurable disturbances. Finally, simulation results demonstrate the remarkable performance of the proposed method, including reduced computational time and improved control accuracy.