| Popis: |
Satellites propellant mass can reach up to 40\\% of an entire spacecraft. Then, slosh phenomenon can be catastrophic if uncontrolled, resulting in complete system/mission failure. Passive physical solutions, such as baffles, compartmentalization or bladders are commonly employed to prevent high amplitude/low frequency propellant motion [1]. However, physical suppression techniques increase mass, complexity, and cost of the overall system. For these reasons, active control system solutions, capable of both precise attitude control and fuel slosh suppression, are of great interest and remain an active field of research. Novel slosh modeling approaches, including major contribution by ONERA and CNES in the framework of the joint research program COSOR (Control of Orbital \\ Robotic Systems) [2], paved the way towards enhanced attitude control solutions such as LPV-based observers, [3], and $H_{\\infty}$-based control, [4]. While the obtained results so far were very promising, reaction wheels saturations handling remained an unresolved issue for some critical tank fluid initial conditions. Indeed, to reject the sloshing disturbance, the high control action employed often leads the reaction wheels to reach their maximum angular speed. The resulting constant angular momentum causes a sudden drop of the commanded torque to zero and any wheel deceleration would evolve in an opposite direction control action. In this scenario, for large angle variations demands, while a rigid satellite will rotate uncontrolled until tracking error sign inversion, in presence of sloshing perturbations the system will rapidly diverge. This paper proposes a static Reference Governor (RG) based solution to address the actuator saturation constraints problem while ensuring tracking performance specifications in presence of sloshing. In its classic formulation, the RG is an add-on control scheme that filters the reference signal by solving at each time step a constrained linear optimization problem (LP). Employing as baseline the satellite with the slosh model derived in [2], we first highlight the limits of the observer-based control solution developed in [3] and the need of extended robustness to constraints. The benefits of integrating a Reference Governor scheme are later deeply discussed: the range of reachable amplitudes is greatly enlarged while the system failure is prevented. Simulation results and stability analysis are therefore illustrated on different scenarii, ranging from trivial to strong saturation cases.In addition, partly inspired by [5], further investigations are in progress to extend the current Reference Governor version to tackle different types of disturbances, including flexible appendices and manipulator arm perturbations. The lack of accurate disturbance models on the one hand, and the stringent attitude control system requirements on the other, lead to consider advanced RG techniques to be reported in the final version of the paper. The intent is to prevent saturation for a diverse class of disturbances while limiting the system conservatism as much as possible. |
