Satellite attitude control by adaptive dynamic programming and stable embedding technique

Abstract

In this dissertation, stable embedding technique and adaptive dynamic programming are combined to handle the satellite attitude control problem. The dynamic system defined on manifold is extended to the Euclidean space by stable embedding technique. Adaptive dynamic programming is applied on the extended system to approximate the control policy that stabilizes the system and minimizes some predefined performance index, under the situation that the system knowledge is imprecise. It is shown through numerical analysis that the attitude system is stabilized at the desired state by approximated controllers. Performances of these approximated controllers are quite close to that calculated by ACADO as the approximation order increases.