Two-stage filter for inertia characteristics estimation of high-speed tumbling targets

Abstract

The estimation of the inertia characteristics of a tumbling target before capturing is one of the greatest challenges in on-orbit servicing missions, especially when the time interval between consecutive measurements is large or the rotational speed of the target is very high. In this paper, a two-stage constant state filter (TCSF) is proposed to predict the attitude and estimate the inertia characteristics, including the principal inertia ratios and the principal inertia axes' directions, of a high-speed tumbling target. The man innovation of this paper is to design an Affiliate Filter to estimate the inertia tensor of the target, based on the estimates of angular velocity from an Adaptive Constant State Filter. Compared with the conventional estimation methods like the unscented Kalman filter method, the proposed TCSF shows better stability and accuracy in strongly nonlinear situations, such as that where the time interval between consecutive measurements is 10 s and the angular velocity is more than 14 deg/s. The Monte Carlo simulations have been proposed to demonstrate the performances of the proposed TCSF method.