Abstract:

According to the problem of how to balance between computation efficiency and estimation accuracy of UKF (Unscented Kalman Filter) in satellite orbit determination applications, this paper puts forward a new algorithm by combining UKF and EKF (Extended Kalman Filter). The algorithm has two aspects improvements compared with the standard UKF algorithm. One is the improvement of the UKF sampling strategy, and with the scaled minimal skew simplex sampling strategy instead of the symmetric sampling strategy; the other one is the improvement of the UKF algorithm structure, and the simple UKF algorithm structure is replaced by the UKF-EKF fusion algorithm structure that the strong nonlinear part of the system is processed by UKF, and the weak nonlinear part of the system is processed by EKF. Numerical results show that the estimation accuracy of the new algorithm is similar with that of UKF, while the computation efficiency is effectively improved.

Key words: extended Kalman filter, unscented Kalman filter, sampling strategy, satellite orbit determination, algorithm