Research On Optimization Method And Application Of BDS/GNSS Multi-frequency And Multi-system PPP Model

As the main source of energy in China,coal resources have made great contributions to economic development.In the 14th Five-Year Plan,it is pointed out that the spatial information and location services in the coal industry are particularly important,especially in the deformation monitoring of mine areas.Global Navigation Satellite System（GNSS）has been widely used in mine monitoring because of its all-weather,high-precision and high-frequency observation capability.Precise Point Positioning（PPP）technology,as one of the GNSS deformation monitoring techniques,has a broad application prospect and market due to its independence from base stations and distance limitations.With the rapid development of various navigation systems,further research is needed on how to fully utilize multi-system and multi-frequency observations to improve PPP accuracy.China’s independently developed Bei Dou Navigation Satellite System（BDS）has gone through a three-step development strategy and officially launched global services on July 31,2020.The research on PPP technology with BDS as the core and its application in deformation monitoring in mining areas will also play an important role in the innovative development of BDS.This paper focuses on the theory of multi-frequency observation linear combination,BDS-3 satellite pseudorange multipath deviation（MP）correction strategy,construction of full-frequency multi-system mixed PPP model and its application in deformation monitoring of surface movement in Banji mining area.The main research contents and results are as follows:（1）To address the issue of large epoch intervals in precision orbit that cannot meet usage needs and different characteristics of BDS three types of orbit satellites,a comparative analysis was conducted on the precision orbit interpolation accuracy of BDS three types of satellites based on sliding and non-sliding Lagrange interpolation methods.The influence of interpolation methods and interpolation orders on three types of satellites was explored,and the optimal interpolation orders of the two interpolation methods were compared.（2）Aiming at the combination application of BDS-3 multi-frequency observations,six commonly used frequency combinations are listed,and the special integer combinations of different frequency combinations which are suitable for ambiguity resolution,cycle slip detection and precision positioning are given.The characteristics and applicable scenarios of different combinations are analyzed,and the search results are compared with the triple-frequency combination of BDS-2.The search results show that the increase of extra frequency cannot only increase the number of optional combinations,but also reduce the errors such as observation noise and ionospheric delay.（3）With the constraint that the noise of the multi-frequency observation combination is minimized,a real coefficient search method of geometry-based and ionosphere-free（GBIF）combination based on least square principle is proposed.The coefficient acquisition method of GBIF combination considering first-order(I₁₁)and second-order(I₁₂)ionospheric delay is given in detail,and the analytical expression of multi-frequency GBIF combination considering only I₁₁is derived.Based on the Lagrange multiplier method,the analytical expression of multi-frequency geometry-free and ionosphere-free（GFIF）combination under the condition ofη₁（28）1is derived when only the I₁₁is considered,and the derivation method when the I₁₂is considered.On this basis,the characteristics of BDS with different frequency combinations and the differences in obtaining GFIF combination sequences using different methods were compared and analyzed.（4）The multi-frequency optimal MP combination coefficient search method is derived with the constraint of minimum combination noise,and the analytical expression of multi-frequency MP combination considering only I₁₁and the derivation method taking I₁₂into account are given.The influence of different frequency combinations on MP sequence extraction is analyzed,and based on the optimal combination coefficients of multi-frequency MP,the characteristics of MP for BDS three types of satellites are discussed.An MP denoising algorithm based on elastic nets（EN）regularization is proposed,and its denoising performance is compared with L2regularization.A QP+AR model considering inter-satellite correlation for IGSO and MEO satellite MP sequences is constructed,and the impact of MP correction on BDS-3/BDS-2 single-frequency static and kinematic PPP performance is analyzed by correcting pseudorange observations.（5）An adaptive BDS-3/GNSS multi-frequency multi-system mixed PPP positioning model that fully utilizes the full frequency observations of each system is constructed,and the performance of the model is evaluated using data collected in different scenarios.Taking BDS-3 and GPS multi-frequency observations as an example,the derivation process of the adaptive multi-frequency multi-system mixed PPP positioning model is described in detail.Firstly,the observation data quality of each satellite navigation system is analyzed based on MGEX station observations,and the correctness of the constructed PPP model is verified.Then,the improvement of accuracy due to the introduction of additional frequencies is discussed,as well as the improvement of PPP when introducing dual-frequency observations of other GNSS systems based on BDS-3 full-frequency observations.Finally,the application performance of the constructed adaptive multi-frequency and multi-system mixed PPP model in complex environments is evaluated by the the dynamic vehicle data in the urban area.（6）The MDMPPP software was developed and designed,and the 3D deformation trend and deformation rate were analyzed based on the long-term observations from the automated deformation monitoring station of Banji coal mine.On this basis,the overall 3D deformation of the surface above the working face was discussed in combination with the actual observation data of the surface movement observation stations and the simulated In SAR data.A settlement prediction model based on Kalman filter and BP adaboost neural network is constructed,and improved using genetic algorithm and phase space reconstruction.The accuracy of the model prediction and its influence by the length of the original modeling sequence are discussed.Figure[88]Table[50]Reference[225]...