File:Closed-loop optimal control implementations for space applications (IA closedloopoptima1094551586).pdf

This thesis explores concepts for a closed-loop optimal control implementation of minimum-time attitude maneuvers of spacecraft. The most common implementation of optimal control solutions is via open-loop commands. However, ignorance of the true system parameters can undermine the open-loop optimal control solution. While traditional closed-loop control methods can compensate for significant levels of uncertainty, this comes at the cost of optimality. This work focuses on optimization of eigenaxis maneuvers, but the concepts are not limited to this constraint. The study begins with an examination of candidate control architectures, weighing the advantages of various closed-loop feedback architectures. A control architecture consisting of a traditional proportional-derivative (or quaternion error) feedback loop and a feed-forward control torque signal is deemed to have the best performance and is then selected for further study. Next, through the analyses of a series of optimal control problems, several real-time optimal control algorithms are developed that continuously adapt to feedback on the system's actual states throughout the maneuver. These algorithms demonstrate significant performance improvements over conventional open-loop implementations, most notably shorter overall maneuver times. The results of this work, therefore, provide an algorithmic enhancement of spacecraft agility.

Subjects: optimal control; closed-loop control; real-time optimal control; eigenaxis; state prediction