A.S. Pustoshilov1, S.P. Tsarev2, M.M. Valikhanov3, K.S. Firyago4

1–4 Siberian Federal University (Krasnoyarsk, Russia)

1 alphasoft@inbox.ru, 2 sptsarev@mail.ru, 3 mvalikhanov@sfu-kras.ru, 4 kosfiryago@gmail.com

One of the problems of accurate positioning by satellite navigation systems is the lack of high-precision orbits of navigation satellites at the time of measurements, as well as the lack of forecasting methods for long periods of time with high accuracy comparable to the accuracy of the final orbits obtained by analytical centers of global navigation satellite systems. Typically, the final orbits of the navigation satellites are available after about 2 weeks and have an accuracy of about 2–3 cm, while the intermediate results calculated by the same analytical centers have an accuracy by an order of magnitude lower. In this paper we consider a method to include additional (surrogate) accelerations in the motion model to improve the accuracy of matching the motion model of navigation satellites to available orbit data and to improve the accuracy of orbit prediction over a long period of time. Using the model parameter matching procedure developed by us, surrogate accelerations were calculated for various GLONASS and GPS navigation satellites at time intervals from one month to one year according to data from several analytical centers. The differences of the obtained surrogate accelerations between different analytical centers are analyzed. It is shown that the difference between our model with surrogate accelerations and the final orbits in the monthly matching interval is about one meter. The linearity of the model under surrogate acceleration perturbations is proved. The high sensitivity of our model for detection of known physical effects (light pressure on the satellite and changes in this pressure when entering the penumbra from the Earth and the Moon) as well as computational errors in final orbits published by analytical centers (gaps at the junctions of the day, etc.) is demonstrated. Our surrogate modelling methodology with small additional surrogate accelerations in the motion model for long time intervals can be combined with other known results for forecasting of the spacecraft orbits in order to reduce navigation error when using orbits published as rapid and ultra-rapid SP3 data.

Pustoshilov A.S., Tsarev S.P., Valikhanov M.M., Firyago K.S. Surrogate modeling of navigation satellite motion. Achievements of modern radioelectronics. 2026. V. 80. № 2. P. 66–73. DOI: https://doi.org/10.18127/j20700784-202602-07 [in Russian]

1. CODE final product series for the IGS. DOI 10.7892/boris.75876.4
2. Ofitsial′nyy sayt IATS KVNO. URL: https://www.glonass-iac.ru [in Russian]
3. Ofitsial′nyy sayt IGS. URL: http://www.igs.org  [in Russian]
4. Tsarev S. P., Lobanov S. A. Otsenka razryvov i anomal′nykh vybrosov v final′nykh orbitakh obrabatyvayushchikh tsentrov IGS. Reshetnevskiye chteniya. 2013. T. 2. № 17. S. 120–122. [in Russian]
5. Kytmanov A.A., Tsarev S.P. Asymptotics of solutions of special second-order linear recurrencies with polynomial coefficients and boundary effects of polynomial filters. Journal of Symbolic Computation. 2025. T. 128. S. 102386.
6. Forrester A., Sobester A., Keane A. Engineering design via surrogate modelling: a practical guide. John Wiley & Sons. 2008.
7. Montenbruck O., Gill E. Satellite Orbits: Models, Methods and Applications. Springer Berlin Heidelberg. 2000.
8. Ushakov YU.YU. Vychisleniye uskoreniya i parametrov modeli dvizheniya kosmicheskogo apparata na osnove final′nykh orbit sluzhby IGS. Reshetnevskiye chteniya. 2013. T. 2. № 17. S. 75–77. [in Russian]
9. Ushakov YU.YU., Kropacheva M.S. Utochneniye parametrov modeli radiatsionnogo davleniya na kosmicheskiy apparat na osnove yego final′nykh orbit sluzhby IGS. Sovremennyye problemy radioelektroniki: sb. nauch. tr. 2013. [in Russian]
10. Caldas F., Soares C. Machine learning in orbit estimation: A survey. Acta Astronautica. 2024.
11. Peng H., Bai X. Machine learning approach to improve satellite orbit prediction accuracy using publicly available data. Journal of the astronautical sciences. 2020. V. 67. № 2. P. 762–793.
12. Pihlajasalo J., Leppäkoski H., Ali-Löytty S., Piche R. Improvement of GPS and BeiDou extended orbit predictions with CNNs. 2018 European Navigation Conference (ENC). 2018. P. 54–59.
13. Gou J., Rösch C., Shehaj E., Chen K., Kiani S.M., Soja B., Rothacher M. Modeling the differences between ultra-rapid and final orbit products of GPS satellites using machine-learning approaches. Remote Sensing. 2023. V. 15. № 23. P. 5585.
14. Pustoshilov A.S., Tsarev S.P. Vysokotochnoye vosstanovleniye orbit sputnikov global′noy navigatsionnoy sputnikovoy sistemy metodom obucheniya po rasshirennym SP3-dannym. Uspekhi sovremennoy radioelektroniki. 2017. № 12. S. 48–52. [in Russian]