Attitude-control model identification of on-orbit satellites actuated by reaction wheels

With the long-term running of an on-orbit satellite, parameters in its attitude model are naturally altered due to the wastage of on-board components and the consumption of propellants or pay loads. In order to obtain high-accuracy attitude-control models and improve the on-orbit adaptation of an attitude controller, identification of an on-orbit satellite becomes another available approach. Due to the effect of exogenous disturbing torques and the measurement noises, estimation of the attitude-control models is eventually converted into that of an errors-in-variables model (EIVM). Since any a priori information on disturbing noises is not known in the on-orbit environment, an L2L2-optimal identification algorithm is proposed to estimate a generalized attitude model (GAM) for the satellite and then the corresponding noise model (NM) can be readily given by a model transformation to the GAM. During the parameter optimization for GAM, v-gap metric is employed as a minimization criterion to reduce the conservativeness of the resulting model and the optimization problem can be solved by linear matrix inequalities (LMIs). Finally, the testbed of a micro-satellite simulator is utilized to demonstrate the effectiveness of the proposed identification algorithm.