Research On Non-linear Filter Methods For High Precision Satellites Orbit Determination

Satellite orbit determination and prediction is a process which extracted the orbit information from the observations data containing the errors. This dissertation is mainly dedicated to the research of the nonlinear filtering on orbit determination of satellites. To determine the precise orbits of satellites is becoming more and more improtant nowadays. It is a key problem that how to choose and improve the nonlinear filters according to the applied background in order to enhance the orbit precision of the satellites.Taking the satellites orbit determination as a basic background, this dissertation mainly research the various nonlinear filters and their improved algorithms. At first the development of nonlinear filters are introduced, which is presented from the filtering algorithm to the technique started with the Bayesian filter. It is pointed that the certain relationship exist in the different filtering methods using the diagram. Moreover the thesis introducedthe basic theories about satellite orbit determination.As a new nonlinear estimation method, the thesis take UKF as the main research algorithm.For the sake of seeing clearly that the changing trendcy of the filter gain matrix and the predicted covariance matrix, the Euclidean norm of matrix was proposed in this paper to describe the trendcy of them.Based on the deeply research on the theory of the UKF, the adjusted and improved UKF are investigated.(1) An adaptive UKF measures is proposed, which can adjust the inaccurate dynamics model and unknown covariance matrix of process niose in real time. The method is applied to orbit determination based on space-borne GPS data of CHAMP. Simulation results show that it is superior in filtering precision.(2) The reasons of the filter instability are discussed and two novel improved algorithms for divergence-suppression are presented in this paper. Moreover simulations are conducted using the satellites orbit determination, and the results are compared with those obtained by normal UKF to illuminate the major reasons of filter instability and demonstrate its effctiveness and advantages over the previous methods.