Research On Prediction And Characteristics Of GNSS Satellite Clock On Orbit

With GLONASS and GPS satellite clocks aging and modernizing, their characteristicspresent corresponding changes and certain differences. In addition, our BDS has launched16navigation satellites since it launched the first navigation satellite in2007, whichmakes it be able to provide all-weather region navigation service covering the Aisa-Pacficregion officially. Therefore, in order to gurantee the accuracy of navigation andpositioning as well as time-frequency transferring, it is necessary to analyse thecharacteristics varying law of GNSS satellite clocks. Compared with other systems, theclock errors of BDS satellite clocks present new characteristics such as sudden skips,incontinuity, unprediction and so on, which make traditional clock models and algorithmsfail to solve these problems. Therefore, in this research, GNSS precise satellite clockproducts are used to study on characteristics, prediction methods and algorithm of thesatellite clocks on orbit. Meanwhile, considering the close relationship of system timeoffsets and satellite clock errors, this thesis also study on combined real-time predictingmodel of BDS/GPS system time offsets. The main work of this thesis is as follows:(1) For GLONASS/GPS satellite clocks on orbit, the thesis adopts the quadraticpolynomial model to fit the satellite clock error model, calculates the phase, frequency,frequency drift and residual of GLONASS and GPS on-board satellite clocks, choosesModified Total Variance and Hadamard Total Variance to calculate the stability of cesium(Cs) and rubidium (Rb) clock respectively, and then analyses the characterization of theon-board satellite clocks. From examples, finding out that the combined method ofdetecting gross is effective; revealing the variation law of the phase, frequency, drift andresidual for GLONASS and GPS satellite clocks; discovering that the stability of GPSBLOCK IIF satellite clocks are the best, GPS BLOCK IIR-M&IIR Rb clocks and thenewer GLONASS satellite clocks follow, and then there are GPS BLOCK IIA Cs clocksand the older GLONASS satellite clocks, but GPS BLOCK IIA Rb clocks are the worst.Meanwhile, there is a numerical relationship between the stability and residual of thesatellite clocks. (2) For BDS satellite clocks on orbit, the thesis analyses the thousand-second stabilityof all the satellite clocks on orbit and the time-varying tendency and the averagethousand-second stability of GEO/IGSO/MEO and BDS constellation. The results showthat except the thousand-second stability of several satellite clocks happen skips, theothers present stable; the average thousand-second stability of BDS constellation is1.69E-13; the average thousand-second stability of GEO is minimum while MEO ismaximum, and IGSO is middle.(3) Through the analysis of starting point bias in IGU prediction clock, a method ofremoving the systematic error from IGU prediction clock is proposed. Meanwhile,analyzing the precision of the improved IGU prediction clock and the influence of precisepoint positioning, follow conclusions can be drawn. The residual of the improved IGUprediction clock is smaller and more concentrated than IGU prediction clock; The closerto prediction starting point, the more significantly starting point bias is improved; Theclock consistency of GPS constellation improves16.2-23.8％, which improves thestability of GPS time system; The kinematic solution precision of GPS receiver clockerror improves30.5-45.4％,which contributes to the application of precise time service.(4) For the prediction models and algorithm of BDS satellite clock errors, the thesisanalyses characteristics of phase skiping for satellite clocks on orbit, and deduces thedetecting methods of phase skiping, and then based on the polynomial model and thecorrection of starting point bais and periodic terms, BDS clock error prediction model andalgorithm is got. The results indicate that the average prediction precisions of24hoursand6hours are7.91ns and2.27ns respectively.(5) Based on explaining the definition of GNSS system time offsets, building thepolynomial predicting model which is suitable for BDS/GPS system time offsets. Someimproving measurements, such as selecting the quadratic term parameter, determining afew prominent periodic terms and choosing the weight function of observations, areapplied to above model. Simultaneously, on basis of the predicting characteristics of ARmodel, building the combined time offsets predicting method of the polynomial and AR models further. The results of simulations show that the single day predicting precision ofthe new model is superior to0.4ns, significantly better than other traditional methods.