FAULT-TOLERANT SPACECRAFT ATTITUDE CONTROL VIA JUMP-LINEAR QUADRATIC CONTROL

摘要

This work is concerned with the development of a suboptimal con-rntrol algorithm for Markovian jump-linear systems, and its application tornfault-tolerant spacecraft magnetic attitude control. For completeness,rnthe jump-linear quadratic optimal controller with full state and moderninformation is presented. Relaxing the assumption of perfect mode in-rnformation, a similar optimal control problem is formulated where thernmode is observed via discrete measurements. The elements of the mea-rnsurement matrix, i.e. the probabilities for correct and wrong mode ob-rnservations are assumed known. The optimal controller is developed,rnwhich requires an exponentially growing computational burden, and arnsuboptimal controller is proposed that only requires knowledge of therncurrent mode measurement. This controller is fnite memory and possessrnsome of the classical linear quadratic regulator features such as the linearrnstate feedback structure and a state quadratic optimal cost-to-go. Thernperformances of the suggested algorithm are illustrated through exten-rnsive Monte-Carlo simulations on a simple numerical example. A realisticrnfault-tolerant spacecraft magnetic attitude controller is developed basedrnon the proposed approach. The attitude controller succeeds in mitigat-rning the destabilizing eu000bect of corrupted mode observations while beingrncomputationally eu000ecient.