Extracting unmodeled systematic errors from BDS orbit determination residuals and application in SPP/PPP

Abstract

To analyze the unmodeled systematic errors in BDS, a filter-assisted Partly Ensemble Empirical Mode Decomposition (PEEMD) method is combined with Hilbert spectrum analysis to extract the feature information from BDS preprocessed orbit determination residuals (PODR). The results show that the feature extraction method can effectively extract a period of about 1 day for GEO/IGSO satellites and a period of about 13 h for MEO satellites. The results of the chi-square test show that the remaining PODR follow a normal distribution. Statistical results from the 3-day experiment indicate that the application of the extracted feature information for BeiDou-only SPP improves positioning accuracy by 20%, 19%, and 23% in the east, north, and upward directions, respectively. BeiDou-only PPP experiments show that the application of the extracted feature information reduces the static PPP convergence time by 34%, 10%, and 21% and the kinematic PPP convergence time by 25%, 11%, and 9% in three coordinate components, respectively.