A method for precisely predicting satellite clock bias based on robust fitting of ARMA models

Autor: Jingchao Zhang, Xia Li, Jun Ye, Kun Jia, Guochao Zhang, Songhui Han, Ruizhe Hao
Rok vydání: 2021
Předmět:
Zdroj: GPS Solutions. 26
ISSN: 1521-1886
1080-5370
DOI: 10.1007/s10291-021-01182-3
Popis: The precise satellite clock bias prediction is critical in improving the positioning, navigation and timing (PNT) service capabilities of the global navigation satellite system (GNSS). Due to the influence of satellite signal path and the observation environment, the satellite clock bias data usually contain outliers that heavily affect the accuracy of satellite clock bias prediction. Based on the time series ARMA model and Bayes statistical theory, we propose a method to precisely predict satellite clock bias and detect outliers in the historical sequence of satellite clock bias. At first, considering the effects of an additive outlier (AO) and innovative outlier (IO), a labeling model for robustly fitting the time series ARMA model and detecting AOs and IOs simultaneously is constructed based on the labeling method of classification variables. Second, the Bayes method for robustly fitting time series ARMA model is proposed based on the Bayes statistical theory. Furthermore, it develops an algorithm to precisely predict satellite clock bias using the Bayes method for robustly fitting the time series ARMA model mentioned above. Finally, in order to illustrate the performance of the method for precisely predicting satellite clock bias that we presented, three examples are designed based on the real GPS data come from the IGS official website, and the prediction results of the method are compared with that of original ARMA model (oARMA), quadratic polynomial model (QP) and gray model (GM). It is found that the method can precisely predict the satellite clock bias as well as accurately detect the outliers in the historical sequence.
Databáze: OpenAIRE
Externí odkaz: