Research On Low-cost And High Precision Positioning Algorithm For Mass Navigation

With the continuous improvement of GNSS chip and antenna performance,the capability of signal acquisition is gradually enhanced.The low-power,high-performance,modular GNSS receiver has gradually entered the mass consumer market,achieved the transition of GNSS from the professional field to the civilian field,broke through the bottleneck of "low precision" and "poor reliability" in the traditional mass navigation(WIFI positioning,ultra-wideband positioning technology,etc.),reduced the user's positioning cost,and promoted the industrial application of GNSS in the field of location services.Compared with the measuring receiver,the data quality of the low-cost receiver is slightly worse.How to use the innovation of the algorithm to compensate for the difference in data quality,so as to ensure the accuracy,robustness and real-time of the positioning,and further develop the GNSS application market,It is great significance.In view of this,the key issues of low-cost receivers in the relative positioning process are focused on,mainly involving low-cost receiver observation stochastic model refinement,low-cost receiver robust-RTD algorithm based on the constraints of velocity and robust RTK algorithm based on partial-fixing strategy.The positioning accuracy,real-time,robustness can been improved dramatically by the innovation or improvement of the algorithms and models.The main work and achievements of the paper are as follows:1.The basic observation model of GNSS data processing is introduced in detail.The error sources in GNSS positioning are analyzed,and the specific elimination methods are given.The stochastic model of observations in GNSS positioning is described,and the advantages and disadvantages of each stochastic model are presented.The two parameter estimation strategies commonly used in GNSS parameter estimation,namely least squares estimation and Kalman filter estimation,are discussed,and their characteristics are elaborated.2.Under the premise that the station coordinates and the double-difference ambiguity are all precisely known,the single difference observation model between stations is deduced by re-parameterization.Two formulas for calculating the residual value of the observations are given,and the observed residuals of the single satellites are stripped out.For different types of satellites of different systems of BDS and GPS,the variation of carrier and pseudorange single-difference observation noise with elevation/SNR is analyzed.According to the elevation /SNR interval,the observation accuracy statistics of each type of satellite is given.The nonlinear function ‘lsqnonlin' in matlab is used to refine the stochastic model with the exponential function model and the SNR weighted model.The fitting coefficient of the stochastic model is given.3.The structural formula of the standardized residual is deduced in detail.Based on this,the correlation between the observations is analyzed,and a robust adaptive algorithm of partial gross error is proposed.In order to make full use of redundant observation information to solve the "flying point" problem in urban canyon environment,two kinds of RTD positioning models based on the constraints of velocity are presented,and the advantages and disadvantages of the two models are analyzed.Four sets of RTD positioning experiments in different scene were designed to verify the robustness of the algorithm.4.In order to shorten the accurately fixed time of the ambiguity,and improve the fixed success rate of the a priori ambiguity,the partial-fixing ambiguity resolution strategy is studied.The satellite cutoff elevation and continuous tracking times are used as the screening conditions to select the ambiguity subset and its variance-covariance,a priori ambiguity fixed success rate and the Ratio value are used as check indicators to determine whether the ambiguity is correctly fixed.For the problem of half-cycle ambiguity deviation in low-cost receivers,the influence of half-cycle ambiguity deviation on the positioning results is analyzed by measured data,and the deviation is processed by the robust Kalman filtering algorithm.Three sets of experiments were designed to verify the positioning performance of the algorithm under ultra-short baseline,short baseline conditions and enhanced information.