Theory And Application Of INS-aided Multi-GNSS Singlefrequency Precise Point Positioning

Precise Point Positioning(PPP)is a positioning technology that emerged in the late 1990 s.It uses a single GNSS(Global Navigation Satellite System)receiver to achieve high-precision absolute positioning on a global scale,and has become a research focus in high-precision positioning.The traditional PPP models usually use a linear combination of dual-frequency carrier phase and pseudorange observations to eliminate the effects of the first-order ionospheric delays.Compared with dualfrequency GNSS receivers,the low-cost single-frequency receivers have been receiving increasing interest by the GNSS community in recent years.However,it's crucial to improve the positioning performance of PPP with single-frequency receivers(SF-PPP)further.Meanwhile,the positioning performance of GNSS is highly dependent on the observation environment nearby,hence it is difficult to provide usable and continuous position information when the GNSS signal is blocked or interrupted.With the aid of autonomous and high-precision Inertial Navigation System(INS)in a short period,it's expected that the positioning performance of SF-PPP could be further improved in an extreme observation environment.This dissertation focuses on SF-PPP with multi-GNSS fusion,the priori constraints of ionosphere delays,and the integration of INS and SF-PPP,and proposes an INS aided multi-GNSS SF-PPP algorithm.The major contributions are as follows:1.A multi-GNSS SF-PPP algorithm is implemented.The related errors of multiGNSS fusion such as inter-system bias(ISB)and inter-frequency bias(IFB)are analyzed,and the SF-PPP observation and state equation based on undifferenced and uncombined model are deduced.Furthermore,the stochastic models of observations and state vectors are presented.The experiment results show that,compared with single GPS system,the precision of SF-PPP is significantly improved in the case of GPS,GLONASS and BDS fusion.Taking the dynamic vehicle experiment as an example,the positioning accuracy in the east,north,and vertical components is improved by 69.9%,85.1%,and 44.6%,respectively.After convergence,the position accuracy in east and north components is approximately 10-20 cm,and the vertical accuracy is better than 50 cm.2.In order to speed up the convergence of SF-PPP,a SF-PPP model with additional external ionosphere constraints is proposed.The introduction of virtual ionospheric observations(obtained from the GIM products provided by IGS)can increase the geometric strength of the parameter solutions of SF-PPP and reduce the convergence time of filtering.In this paper,three methods,such as constant constraint,spatiotemporal information constraint and gradual relaxation constraint,are used to determine the variance of virtual ionospheric observations.The comparison results show that the gradual relaxation constraint is the better choice.Dynamic vehicle experiment results show that compared with standard SF-PPP,the convergence speed and the positioning accuracy is obviously improved.In the case of a combination of GPS,GLONASS and BDS,the RMS(Root Mean Square)of the position error in east,north,and vertical components is reduced by 87.2%,48.9% and 55.6%,respectively,and the position accuracy in east and north components after convergence is approximately 20 cm and 30 cm,and the vertical position accuracy is better than 50 cm.3.The INS aided SF-PPP integration algorithm is implemented based on the loosecombination mode.The solving algorithm and error model of INS are presented,and the observation equations and state equations of INS and SF-PPP loose-combination are deduced.Afterwards,the initialization of INS and the error correction of lever arm are discussed.Dynamic vehicle experiment results show that the position accuracy of the INS aided SF-PPP is mainly determined by the SF-PPP in the loose-combination mode,and the position results is smoothed by INS.In terms of velocity,the eastward and northward velocity accuracy is about 5 cm/s and 7 cm/s,and the velocity accuracy of vertical is better than 10 cm/s.For the attitude,the pitch and roll angle accuracy is about 0.1,and the heading angle error is about 0.5.The simulated GNSS signal interruption results indicate that within 10 s,the position accuracy could be maintained at sub-meter level,the velocity error in east,north and vertical components is about 0.1 m/s,while the error of pitch and roll angle drifts to about 0.1-0.2,and the accuracy of heading angle is better than 1.