Research On Model And Method For GPS/Galileo Tightly Combined Real-Time Relative Positioning

With the development of four global navigation satellite systems and local navigation satellite systems,satellite navigation and positioning technology has entered the multi-frequency multi-system GNSS era.It has gradually become a research hotspot in recent years to make full use of multi-frequency multi-system resources to further improve positioning accuracy,availability and reliability.Multi-GNSS mixed processing include loosely combined processing and tightly combined processing.The loosely combined relative positioning chooses pivot satellite individually for each system and combine double-differenced(DD)observations from every system to perform positioning.Thus,only intra-system DD observations can be created.The tightly combined relative positioning chooses one pivot satellite for all the observations from every system.Consequently,both inter-system and intra-system DD observations can be created.Compared with classical loosely combined model,the tightly combined model can reduce the number of pivot satellites,increase additional observations,improve the model strength and further improve the positioning accuracy and reliability.At present,although many scholars have carried out research on GNSS tightly combined relative positioning,there are still some problems that not solved properly.For example,most of current research focuses on tightly combined relative positioning of overlapping frequency,and there are few studies on the non-overlapping frequency and the mixed processing of overlapping and non-overlapping frequencies.Researches on medium-long baseline tightly combined relative positioning are still not mature.Besides,analyses on positioning performance of tightly combined model are not sufficient yet.Among the current navigation satellite systems,GPS is currently the most complete and mature Global Navigation Satellite System,and began to implement modernization since 2005.The European Galileo navigation satellite system has been developing rapidly in recent years.It has also provided initial operational capabilities in December 2016.There are currently 17 working satellites in orbit and it will complete construction of the whole constellation in 2020.There are two overlapping frequencies(GPS L1/Galileo E1,GPS L5/Galileo E5a)and one closing frequency(GPS L2/Galileo E5b)between GPS and Galileo,which provides the foundation for GPS/Galileo mixed processing.Taken those into account,the thesis focuses on the models,approaches and key issues of mixed processing GPS/Galileo overlapping and non-overlapping frequencies tightly combined relative positioning from short to medium-long baselines,proposes a real-time DISB estimation method,and comprehensively evaluates the positioning performance of tightly combined model.The main works and conclusions of this thesis are as follows:(1)The research statuses of GNSS tightly combined relative positioning at home and abroad are summarized and the development,signal design and time and coordinate reference system of GPS/Galileo are introduced.The function model and stochastic model of classical loosely combined relative positioning are derived from original observation.The Kalman parameter estimation method is described.(2)Systematic and deep researches on the model and method of mixed processing GPS/Galileo overlapping and non-overlapping frequencies tightly combined RTK for short baselines are carried out.Aims at the problems in non-overlapping frequency,the correction method of single differenced bias is proposed.DISBs are extracted and analyzed by single-epoch estimation using Curtin static short-baseline measured data.The result shows,for baselines with identical receivers(including the same firmware version),the overlapping frequency DISBs are ignorable,while the non-overlapping frequency DISBs and DISBs from different receivers are usually unneglectable.Code DISB and overlapping phase DISB doesn't influence by the receiver's reboot,however,non-overlapping phase DISB would change after receiver restarts.The short-baseline DISBs could achieve convergence in minutes.Meanwhile,DISBs process methods that can meet real-time applications are proposed.(3)Systematic and deep researches on the model and method of mixed processing GPS/Galileo overlapping and non-overlapping frequencies tightly combined RTK for medium-long baselines are carried out.Aims at the linear correlation of DISB parameters and DD ionosphere parameters in medium-long baseline tightly combined RTK,the thesis proposes two reparametrization strategies(eliminating one DISB parameter or one DD ionosphere parameter),and proves theirs equality and deductive relationship.The mixed processing of GPS/Galileo overlapping and non-overlapping frequencies tightly combined RTK are carried out using measured car data and US PBO data.Owning to the component of DD ionospheric delay,the medium-long baseline DISBs are not suitable in short baseline.On the contrary,DISBs derived in short baseline could be corrected to medium-long baseline.(4)The performance of mixed processing GPS/Galileo overlapping and non-overlapping frequencies tightly combined RTK are fully evaluated using measured data.A large amount of measured data are adapted to verify the proposed methods from the aspect of Ambiguity Dilution of Precision(ADOP),ratio test value,ambiguity conditional number,decorrelation number,search times and positioning accuracy.The results show that:when the satellite elevation cut-off is 10 degree,for short baseline RTK,comparing with loosely combined model,the tightly combined model can significantly increase decorrelation number,reduce the ambiguity ADOP,conditional number,and search times in LAMBDA by 10%,and can increase the ambiguity ratio test value by 8%to 12%,and improve the positioning accuracy to some degree.For medium-long baseline RTK,the tightly combined model can significantly increase the ambiguity decorrelation number,ncrease the ratio test value by about 40%,and can reduce the ambiguity conditional number and search times by 50%and 20%respectively.ADOP and positioning accuracy are also improved.When the elevation cut-off is higher and the signal is more severely blocked,the improvement of tightly combined model would be more obvious.These evaluationsdemonstrate that the tightly combined model truly improves the model strength,the float ambiguity accuracy and positioning accuracy.The models and methods established in this thesis are correct.