Optimal reaction wheel speed management strategies

Abstract

Highly precise attitude maneuvers are often required by spacecraft that are imaging moving targets (e.g. space telescopes). However, there are various sources of disturbances that can affect the pointing accuracy and the agility of the satellite. While the environmental disturbances cannot be manipulated, the disturbances from the components inside the satellite could be reduced by applying proper attitude control strategies.

This thesis aims at reducing the reaction wheel (RW) disturbances for reaction wheel assemblies (RWAs) that consists of redundant RWs. The major sources of RW disturbances include the reaction wheel stiction and resonance, and both are activated when the wheel speed approaches certain values (so-called \avoidance speeds" in the rest of this thesis). In other words, the RW stiction and resonance can be eliminated by keeping wheel speeds away from the avoidance wheel speeds. Moreover, the redundant wheels provide an extra degree of freedom for wheel speed control without affecting the attitude maneuver. This thesis explores how to optimally utilize the extra degree of freedom in terms of minimizing RW disturbances.

Three reaction wheel speed management algorithms are proposed by solving the corresponding optimization problems. Stochastic optimization methods such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are implemented. The three algorithms are validated by Monte Carlo tests with the incorporation of a high fidelity spacecraft attitude control simulator. The effect of the reaction wheel assembly confi guration design is also discussed in this thesis. The results from the Monte Carlo test indicate that at low wheel speed level, the average settling time for 5 successive slews can be reduced by 40% by implementing the wheel speed management algorithms. The results also demonstrate that the smaller condition number of the con figuration matrix, the proposed wheel speed management algorithms would have better effectiveness.