The Research On Real-time Orbit Determination Of Low Earth Orbit Satellites Based On UKF

This dissertation is mainly dedicated to the research on the nonlinear filtering for real-time orbit determination of low orbit (LEO) satellite. The task of real-time orbit determination is required to provide precise state parameter of satellite orbit in little time. Widely as the classical Extended Kalman Filter (EKF) has been applied, its blemish is clear because of the linearization expression when dealing with nonlinear problems. As a new nonlinear estimation method, the Unscented Kalman Filter (UKF) has attracted more and more attention.Taking the EKF as a basic method, this dissertation mainly researches the application of the UKF in real-time orbit determination for low earth orbit satellite as well as seeks after real-time and high precision technique of orbit determination. Our research emphasizes on the following four aspects:1. The research on the development for nonlinear filteringBased on the comparison between various nonlinear filtering, we summarize the one which is more proper to solve the problem of real-time orbit determination. And we pay more attention to the development of the UKF.2. The research on orbit determination method based on the EKFIn the dissertation, the theory of Bayes filtering is taken as the starting point, and the Extended Kalman Filter, which is an algorithm to approximate the optimal filtering, is studied. By transforming the observation equation, an improved EKF algorithm is obtained which can simplify the Jacobian matrix. And then it is applied in the simulative experiment of real-time orbit determination for LEO satellites.3. The research on orbit determination method based on the UKFBased on the deeply research on the theory of the Unscented Kalman Filter, the scaling UT (Unscented Transform) is applied to the minimal skew simplex sampling strategy, and then an UKF algorithm of minimum computation cost is obtained. Simulation results indicate that it is much more efficient and steady.4. The research of the preferences in the scaled UTThe effect of the scaling factor is analyzed, and then a self-adaptive selecting method for the parameter is proposed and embedded in the UKF algorithm. Simulation results indicate that the method has good applicability, and the relevant precision of the orbit determination approximates to that of the UKF with an optimal fixed scaling factor.The main contradiction of real-time orbit determination exists between evaluation precision and computation cost. How to equipoise the two indexes and get the optimal method to satisfy the requirement of orbit determination is the problem we dedicate to resolve. Simulation results indicate that to the weakly nonlinear orbit determination problems the EKF makes a similar precision with that of UKF; but the former needs less computation cost but more complexity, while the latter is just opposite; to the strongly nonlinear problems the UKF makes higher precision than that of EKF, and our improved algorithm makes the UKF much more stable. Therefore we should choose the better method as the material requirement and condition of the orbit determination task.