Study On Efficiency Of Real-Time Precise BeiDou/GNSS Satellite Orbit Determination And Property Of Inter-Frequency Clock Bias

With the advantage of providing high precision and high sampling rate observations for users anywhere and anytime,GNSS has been widely used in navigation and positioning,engineering surveying,geoscience research,et al.Since 2010,GNSS has entered a new period that supporting multi-frequency signals has become mainstream with the modernization project of GPS and GLONASS,as well as construction and development of Galileo and BeiDou.Providing real-time multi-system and multi-frequency service with high precision has been a development direction of GNSS application.The real-time precise orbit of BeiDou/GNSS satellites provides dynamic datum for real-time precise service,which is the foundation of real-time GNSS applications including PPP and PPP-RTK.The real-time precise orbit could be determined in two ways,the one is called batch process and the other one is called extended filtering process.Comparing with the batch process,the extended filtering process is more flexible and efficient to operate because a more reliable orbit could be obtained by adjusting process noise of orbit parameters to balance the contribution of dynamical models and observations when dynamical models have less accuracy.The extended filtering process also has the ability to detect maneuver satellites in real time and avoids redundant computation of historical data,which is an important developing orientation for real-time orbit determination with high efficiency.However,with continuous increasement of the number of satellites in orbit,data processing efficiency has become a bottleneck to further improve the performance of filtering process,and the efficiency of the numerical integration of multi-system GNSS satellites should be first settled.In addition,the filtering method has a long convergence time that is up to more than ten hours when using the initial orbit conditions obtained from broadcast ephemeris due to accuracy limitation of initial condition and dynamical models as well as poor geometric structure,leading to a long initialization time for the whole service system.Additionally,in terms of applications of real-time precise orbit,deep intergration of multi-frequency data has been a hot area of research,especially the influencing mechanism of pseudorange and phase intrumental biases on parameter estimated by multi-frequency raw data needs further study.On acount of demands of GNSS real-time applications and trends of deep intergration of multi-frequency data,this dissertation focus on several key issues of the efficiency of real-time POD and applications based on multi-frequency data,aiming at solving exsiting theoretical and technical problems including four aspects:effective numerical integration for multi-system GNSS satellites,fast convergence of satellites' orbit and clock when using filtering process,real-time POD of BeiDou IGSO and MEO satellites in the yaw-fixed regime and the characteristic of BeiDou/GNSS satellites' IFCB and its impact on the fusion processing using BeiDou/GNSS multi-frequency raw data.On the basis of the innovation of these theory and methods,real tracking data is used to evaludate the precision of real-time BeiDou/GNSS orbit and clock.Real-time PPP using multi-frequency data is also conducted.Generally,the main content and contribution of this dissertation includes:1.The satellite orbit variational equation and observation equation used in real-time POD processing are derived,and the most common perturbation force models and error correction models of observations are summarized.The yaw attitude models of current four systems' satellites are investigated,and the principle of Square Root Information Filter(SRIF)algorithm and its implementation in real-time POD processing are studied.2.On the basis of analysing the influence of integration step of different GNSS satellites on the precision of Admas method and time consuming of different perturbation force models,an adaptive step-changed Admas integration method and a synchronous integration algoritm for multi-GNSS satellites are developed to improve the compute efficiency.Results show that,comparing with the traditional step-fixed RKF-Admas orbit integraion method,the compute efficiency of the proposed method is improved significantly without damaging the accuracy:it takes only 0.09s for a single satellite when integrating 24h,which is twice as fast as the traditional step-fixed RKF-Admas orbit integration method.Meanwhile,it takes less than 0.01s for a single satellite when the synchronous integration algoritm for multi-GNSS satellites is applied to integrate an epoch time of 30s.Besides,further improvement can be achieved when the number of satellites is increased.3.The impact of initial orbit parameter on convergence time of real-time POD using filtering process is analyzed,and real-time POD using ultra-rapid ephemeris' constraint algorithm is proposed to improve the orbit accuracy during SRIF convergence period.The constraint equation is derived and the flow chart of the new algorithm is given.Experiment results indicate that,by introducing reasonable variance of the constraint equation,the new method can significantly shortened the convergence time of GPS and BDS satellites' orbit and clock,and RMS of the difference between the orbit estimated by the new method and the old one after convergence is in milimeter range.Furthermore,the impact of orbit arc length on predict orbit of four systems' satellites is analyzed by dynamic fit using WUM precise orbit product.Results show that the best precision of predict orbit is obtained when arc length of 40h-48h and 40h-48h is used for GPS and GLONASS satellites respectively,and 44h-52h is used for BDS and Galileo satellites.The best arc length is related with the accuracy of POD and perturbation force models,but has no significant relation with orbit types.4.The real-time orbit precision of BeiDou IGSO and MEO satellites in the yaw-fixed regime is evaluated,which is derived by batch process and SRIF process respectively.Results show that,for batch process,the observed orbit accuracy of yaw-fixed IGSO and MEO satellites could be effectively improved when estimated with ECOM 5 parameter SRP model and one constant acceleration in along-track direction,which is at the same level with satellites in the yaw-steering regime.However,the predicted 6h orbit has a relative larger precision loss than that of satellites in the yaw-steering regime.For SRIF process,an adaptive adjustment method of noise process of the SRP model parameters is proposed for yaw-fixed satellites,and the real-time orbit is improved by 18%in average from the predicted 6h orbit by using the adaptive adjustment method,which is at the same with satellites in the raw-steering regime.This is because the process noise of orbit parameters and SRP model parameters compensates the error caused by SRP model.In addition,the predict length can be limited in an epoch interval time,which helps to reduce the accumulation of the numerical integration error.5.The characteristic of BeiDou/GNSS satellites' IFCB and its impact on parametersestimated using multi-frequency raw observables is studied.Firstly,the rank deficiency of observation equation using multi-frequency raw measurement is dicussed.By introducing seven constraint datum,the full rank ovservation equation is derived and the relationship of different parameters between two frequency standards of clock is determined.Furthermore,the impact of the variation of phase instrumental bias on parameters derived by uncombined triple-frequency observables is analyzed,and a proportion relation between IFCB derived from DIF(Differential Ionosphere-Free)combination using phase-smoothed range and IFB derived from raw triple-frequency observables is verified.Modeling analysis proves that the IFCB of GPS satellite mainly consists of two kinds of periodical variation which cycles are 2? and ? respectively,and its amplitude can reach as large as 10cm,destroying the stability of GPS IFB parameter.Therefore,triple-frequency measurement is not compatible in observation equation using raw triple-frequency measurements if GPS IFB parameter is regarded as a constant.However,the amplitude of BDS and Galileo satellites' IFCB is at the milimeter level,which has limited influence on parameter estimation using raw trple-frequency measurements.6.An improved satellite multi-frequency DCB estimation algorithm is proposed,in which the ionospheric observable and IFCB derived from both triple-frequency uncombined observation model and Geometry-Free linear combination of Phase-Smoothed Range(GFPSR)is used and weighted according to their different accuracies.Experiment results show that the new method can estimate satellites' DCB precisely even if only a few stations are used,where the RMS difference between GPS DCB derived from the method proposed and CODE product is better than 0.1ns,and the RMS difference between BDS DCB derived from the method proposed and DLR product is better than 0.3ns.Comparing with DCB derived from GFPSR,the STD of GPS satellites' DCB derived by the new method is improved by 30%,and that of BDS satellites' DCB derived by the new method is improved by 20%,in average.7.The precision of real-time orbit and clock,as well as kinematic PPP using different frequency pairs,is evaluated.The impact of IFCB on PPP using signals on the frequency that not define clock datum is studied.10 days' real-tracking data collected from 150 IGS and MGEX stations is used to conduct the experiment.Results show that,the 3D precision of real-time GPS orbit is 7.8cm by comparing with IGS final product,and find a consistency of better than 4m for GEO,20-40cm for IGSO and 11cm for MEO by comparing with WUM and GBM in average.The precision of GPS and BDS real-time orbit is at the same level with current research results.Furthermore,the real-time orbit and clock is used to conduct real-time dynamic PPP using multi-frequency data.Results show that,the accuracy of horizontal direction and vertical directioin is 5cm for GPS only solution using ionosphere-free combination of phase and code measurements on L1W(C)and L2W frequency,while the accuracy for BDS only solution based on L2I and L7I frequency is slightly lower.GPS and BDS combined solution can significantly improve the PPP precision and shorten the convergence time.The results of BDS only dynamic PPP using ionosphere-free combination of phase and code measurements on L7I and L6I frequency show that,the convergence time can be significantly shortened when the satellites' IFCB is corrected,and the positioning accuracy is also improved after the convergence.