PPP Based Sequential Orbit Determination For Satellites In Low Earth Orbit

Nowadays, many artificial satellites have been launched into Low Earth Orbit(LEO) for ocean surveillance, earth gravity recovery, magnetic field detection, terrestrial surveying and mapping, atmosphere and space environment detection and so on. These scientific missions are always based on precise orbit information of the satellite. During the last 30 years, spaceborne GNSS(Global Navigation Satellite System) technology has been the primary technology for the satellite orbit determination applications due to its continuous monitoring, all-day service, autonomy & independency and high-precision positioning. Batch Least Squares method has been used widely for centimetre-level of precise orbit determination for LEO satellites,however, it is only in post-processing mode. With the increasing demands of orbit determination for space missions, spacecraft high-precision, autonomous and real-time navigation will become the potential technique. In comparison with Least Squares method, Kalman filter is more advantageous in dynamic GNSS data sequential processing with fast convergence, which is necessary for LEO satellite real-time precise orbit determination. This thesis focuses on LEO satellites orbit determination in the sequential Kalman filter framework using spaceborne GNSS measurements, researching into kinematic orbit determination(KOD), reduced-dynamic orbit determination(ROD), integer ambiguity resolution and fixing in precise point positioning(PPP) algorithm and KOD test & validation using combined GPS/Beidou observation in ground testbed. All the work is to achieve the goal of high-precision, real-time and multi-information fused orbit determination, which are specified as follows:In the framework of Kalman filter, a PPP based KOD approach is studied for LEO satellite using GPS observations. A receiver clock modelling aided KOD(KOD-CLK) approach is proposed. This thesis introduces two GPS receiver clock models, namely two-state and threestate models, to describe the deterministic and stochastic property of the receiver clock. Test results indicate that the positioning accuracy using KOD-CLK approach could be improved significantly using one day of GRACE-B real flight data, especially for the radial component. Using two-order receiver clock model, orbit determination RMS errors are achieved with0.088 m, 0.771 m and 0.064 m in the RIC(Radial, In-track and Cross-track) coordinate frame.The similar accuracy could be achieved using three-order receiver clock model. In particular,the error of the radial component is reduced by over 40.0% in the real-time scenario using broadcasting GPS ephemerides.Reduced-dynamic orbit determination(ROD) approach is implemented based on KOD approach by introducing orbital modelling. Orbit determination accuracy of better than 8cm in terms of 3D RMS could be achieved by conventional extended Kalman filter based ROD(EKFROD) approach with a set of GRACE-B data. Aiming at determining the position and velocity of the satellite in real-time, a Consider Kalman Filter(CKF) based ROD(CKF-ROD) approach is derived in this thesis. In the CKF, orbit dynamic model is simplified to meet the space-borne computational limitations. The atmospheric drag and solar radiation pressure coefficients are’considered’ rather than estimated in comparison with EKF-ROD strategy. However, the propagation of the covariance of the consider parameters can absorb the unmodelled and dismodelled perturbations. Therefore, the filter could become convergent with desirable orbit determination performance. Meanwhile the state estimation space is reduced and computational time is saved.Different orders of Earth Gravity Models(EGM) are compared in CKF-ROD strategy, a lower order of EGM can achieve a decimeter-level of orbit determination accuracy in post-processing with the same GRACE-B dataset. The solutions also indicate that this proposed method using5?5 EGM for orbit propagation could achieve satisfactory precision orbit determination with1.5 meter-level of 3D RMS error in real-time scenarios.Integer ambiguity resolution in PPP(PPP-AR) is studied in this thesis. Three PPP-AR approaches are presented in details, namely single-differencing between GPS satellites(SDBS) uncarlibrated phase delays(UPDs) or fractional cycle biases(FCBs) method, decoupled clock method and integer phase clock method. The integer phase clock method is modified as SDBS is formed to eliminate the hardware delay biases at the receiver end. IGS(International GNSS Service) station observations and GRACE-B data are used to test the integer phase clock method with Centre National d′Etudes Spatiales and Collecte Localisation Satellites(CNES-CLS) products. 10-hour GRACE-B data is processed using the reduced-dynamic orbit determination strategy with ambiguity fixing, the orbit determination accuracy could be improved by 15% in terms of 3D RMS.As China’s Bei Dou Navigation Satellite System(BDS) has been formally operational since the end of 2012, standalone Beidou and combined GPS/Beidou positioning techniques tend to be applied in the future space missions. PPP based kinematic orbit determination approach using GPS/Beidou observations is studied and observational model based on InterSystem Biases(ISB) is formulated. GNSS data collected from ground stations are processed in both static and kinematic PPP modes. One decimetre level of positioning accuracy is achieved for the dual-system in kinematic PPP mode. Since LEO satellites can only track Beidou Geostationary Earth Orbit(GEO) satellite in a short arc above the Asia-Pacific region, the performance of the dual-system PPP algorithm is evaluated using the developed GPS/Beidou PPP software for the scenario without Beidou GEO satellites particularly.