| The global navigation satellite system(GNSS)has the advantages of providing realtime and high-precision observation.It is a great system engineering in the 20 th century,which contains a lot of innovation in science and technology and engineering application.Since the 1970 s,GNSS navigation and positioning theory and technology have experienced rapid development.On the one hand,the global navigation system has gradually developed from a single GPS system to multi-systems,such as GLONASS,Galileo and Beidou.The signal has also developing from the dual frequency signal to the three or multi frequency signal.The infrastructure is increasingly improved,and the hardware configuration is gradually diversified.On the other hand,the data processing mode is also developing in the direction of real-time and high-precision.This lays a foundation for the wide application of GNSS technology.The reliability of satellite navigation and positioning is one of the inseparable factors for the stable application of this technology.The real and accurate products and positioning results are the premise and foundation of all applications.In the process of GNSS real-time data processing,if the gross error can be found timely and accurately,it can effectively avoid the influence of gross error on the accuracy of adjustment results.Reasonable quality control can improve the reliability.Therefore,it is of great significance to study the real-time quality control from observation data to parameter estimation in real-time GNSS data processing.The essence of quality control is to divide data into "good" and "bad",which is consistent with the idea of clustering in mathematics.Traditional quality control methods think that the data is either good or bad,but sometimes it is difficult to define the data accurately in the process of actual data processing.Fuzzy clustering method quantitatively expresses the fuzzy relationship between samples.It establishes the uncertainty description of sample category,and avoid the traditional "either or that" classification.According to the characteristics that fuzzy clustering analysis method can effectively realize the uncertainty description and classification of samples,fuzzy clustering can be applied to GNSS data processing quality control.According to the requirements of real-time and reliability in application of GNSS and the research hotspots of GNSS real-time quality control,this paper focuses on two aspects of quanlity control.One is the real-time observation data quality control based on fuzzy clustering analysis and the other is the real-time ambiguity resolution and its reliability test.These methods have been applied to real-time precise orbit determination and positioning.Studies on observation preprocess and ambiguity fix reliability test are conducted based on the real GNSS observation.The main contributions of this paper are as follows:(1)The development trend and application requirements of satellite navigation technology are analyzed and summarized.The necessity of real-time quality control is demonstrated.Combined with the essence and characteristics of quality control and fuzzy clustering,this paper proposes that the fuzzy clustering analysis method can be applied to real-time quality control of GNSS satellite navigation precision data processing.This method can be effectively used in two kinds of quality control problems: observation data preprocessing and ambiguity resolution reliability test.It can solve the problems of multiple gross errors/cycle slip detection and the decline of positioning and orbit determination accuracy caused by wrong fixed ambiguity.(2)A method of single / multi frequency gross error identification and cycle slip detection based on fuzzy clustering analysis is proposed.Based on the coefficient matrix and residual vector of raw observation equations or single difference observation equation between epochs,a series of sample is constructed and membership values are also calculated.Through membership analysis,gross error identification and cycle slip detection are realized.The method makes full use of the geometry and residual distribution of the observed data,and the mathematical model is feasible.The single frequency experiment results show that,compared with the traditional robust method,the detection success rate of this method is improved by nearly 30% in the case of multiple gross errors and cycle slips.The effectiveness of the proposed method is further verified by the results of dual frequency experiment.(3)This paper analyzes the effect of PPP ambiguity resolution based on lambda and ratio test,and further puts forward a reliability test method for PPP single-differenced ambiguity fix based on fuzzy clustering analysis.In this method,the coefficient matrix of time-varying parameters and the residual vector of observation after fixing ambiguity are used to construct samples and the membership values are calculated.Through membership analysis,wrong fixed single difference ambiguities are identified.The effectiveness of the proposed method is verified by static and dynamic positioning experiments with 140 IGS stations around the world.The results show that,compared with the single ratio test method,the rate of accuracy recognition for fixed ambiguities in static positioning is improved from90% to 97%,which is improved by 17% in the dynamic experiment.The rate of correct recognition for fixed ambiguities is improved by 12% on average.In addition,the problem that the positioning accuracy is reduced due to the wrong fixed ambiguity is effectively avoided.(4)The real-time ambiguity fixing in orbit determination based on root mean square information filtering is studied and realized.Furthermore,a double-differenced fixed ambiguity reliability test method based on fuzzy clustering analysis is proposed.In this method,the efficiency of filtering is fully considered.The coefficient matrix of time-varying parameters and the residual vector of observation after fixing ambiguity are used to construct samples and the membership values are calculated.Through membership analysis,the wrong fixed double-differenced ambiguities are identified.Simulated real-time orbit determination test results show that the accuracy of real-time filtering orbit is improved by 24%,27% and6% in along-track direction,cross-track direction and radial direction respectively,reaching4.4cm,2.6cm and 2.2cm.The feasibility of avoiding the problem of filter divergence caused by wrong fixed ambiguity is analyzed. |
PDF Full Text Request