A method and system for fusion and enhancement positioning based on Beidou 5G

Abstract

The application discloses a kind of based on Beidou 5G fusion enhanced positioning method and system, it is related to fusion positioning technical field, including: step one: obtain Beidou observation data, 5G observation data and terminal motion state data;Step two: generate Beidou pseudo-range direction constraint line bundle;Step three: generate 5G ground phase constraint chord segment;Step four: generate star ground initial network;Step five: based on star ground initial network extraction network key node, execute sparse chord bundle recursion algorithm, extract sparse network node and generate star ground sparse chord bundle graph;Step six: generate sparse network contraction potential field;Step seven: extraction sparse network contraction center point, and carry out local interpolation correction, generate fusion enhanced positioning coordinate;Step eight: obtain fusion enhanced positioning trajectory and output fusion enhanced positioning result.The application utilizes sparse chord bundle recursion algorithm and network contraction potential field, realizes Beidou 5G high-precision fusion enhanced real-time positioning.

Description

Technical Field

[0001] This invention relates to the field of fusion positioning technology, and in particular to a fusion enhanced positioning method and system based on BeiDou 5G. Background Technology

[0002] With the increasing demand for intelligent transportation, low-altitude logistics, emergency communication, and high-precision navigation for mobile terminals, the fusion enhancement positioning technology combining BeiDou satellite navigation and 5G communication networks has attracted widespread attention. Existing positioning systems typically rely on multi-source observation data such as BeiDou pseudorange, satellite positions, 5G base station phase measurements, and 5G angle of arrival measurements. They use fusion methods such as extended Kalman filtering, weighted least squares, and factor graph optimization to jointly estimate terminal coordinates. However, in practical applications, the following problems commonly exist: BeiDou observation data mainly comes from satellites in the airspace, while 5G observation data mainly comes from ground base stations. The two types of observations differ significantly in spatial distribution, error morphology, and constraint direction. Existing filtering and fusion methods often treat them as independent measurement items and weight them together, lacking a spatial geometric fusion process that unifies the directional structure of BeiDou pseudorange residuals with the continuous structure of 5G ground phase. This makes the fusion results susceptible to anomalies from a single observation source. In environments such as urban canyons, obstructed roads, and underground entrances and exits, BeiDou pseudorange is prone to directional deviations, while 5G phase measurements may experience local abrupt changes due to integer jumps, uneven geometric distribution of base stations, and angle of arrival errors. Traditional filtering methods usually rely on noise matrices or residual thresholds for correction, making it difficult to maintain the continuity of the satellite-ground constraint structure. At the same time, existing methods often require global searches or complex iterations of a large number of candidate coordinates when performing high-precision positioning, resulting in high computational costs and making it difficult to generate stable fused enhanced positioning coordinates and fused enhanced positioning trajectories in real time at the terminal.

[0003] Therefore, how to provide a fusion-enhanced positioning method and system based on BeiDou and 5G is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose a fusion-enhanced positioning method and system based on BeiDou and 5G. This invention utilizes a sparse string beam recursive algorithm and a mesh shrinking potential field. It expresses the spatial directional relationship of satellite observations through BeiDou pseudorange direction constraint beams and the phase continuity relationship of base stations through 5G ground phase constraint chords, constructing an initial satellite-ground mesh under a satellite-ground fusion constraint coordinate system. The sparse string beam recursive algorithm generates a satellite-ground sparse string beam diagram by extracting sparse mesh nodes and pruning redundant intermediate nodes, thereby reducing the computational load for positioning. Furthermore, it extracts the sparse mesh shrinking center point using the sparse mesh shrinking potential field and weakens the mesh discretization error through local interpolation correction, improving the accuracy, stability, continuity, and real-time calculation capability of fusion-enhanced positioning coordinates in complex environments.

[0005] According to an embodiment of the present invention, a fusion-enhanced positioning method based on BeiDou 5G includes the following steps: Step 1: Acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch; Step 2: Calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. Project the pseudorange difference of the single satellite along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint bundle. Step 3: Calculate the phase difference of a single base station based on 5G observation data, and perform integer loopback on the phase difference of a single base station in continuous positioning epochs. Combine the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. Step 4: Establish a satellite-ground fusion constraint coordinate system based on terminal motion status data, and map the BeiDou pseudorange direction constraint bundle and 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground network; Step 5: Extract key nodes of the initial web weaving based on the satellite-ground network, execute the sparse chord recursion algorithm on the key nodes of the web weaving, extract sparse web nodes and generate a satellite-ground sparse chord diagram; Step 6: Generate local candidate coordinate sampling points centered on the sparse web nodes in the star-ground sparse chord diagram, and generate a sparse web contraction potential field based on the local web contraction amount corresponding to the local candidate coordinate sampling points. Step 7: Extract the sparse web contraction center point from the sparse web contraction potential field, and perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch; Step 8: Based on the fusion-enhanced positioning coordinates, obtain the fusion-enhanced positioning trajectory and output the fusion-enhanced positioning result.

[0006] Optionally, the BeiDou observation data includes BeiDou satellite number, BeiDou satellite position, and BeiDou pseudorange; the 5G observation data includes 5G base station number, 5G base station position, 5G phase measurement value, and 5G angle of arrival measurement value; the terminal motion status data includes the terminal coordinates of the previous positioning epoch and the terminal predicted displacement. By mapping the BeiDou observation data, 5G observation data, and terminal motion status data according to the positioning epoch, we can obtain the BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch.

[0007] Optionally, step two specifically involves: Based on BeiDou observation data, the BeiDou satellite positions and corresponding BeiDou pseudoranges of each BeiDou satellite are read, and the terminal coordinates of the previous positioning epoch are read. Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the positions of each BeiDou satellite, and use it as the single-satellite prediction distance; Subtract the corresponding single-satellite predicted distance from the BeiDou pseudorange to obtain the single-satellite pseudorange difference; Starting from the coordinates of the previous positioning epoch terminal, a spatial line-of-sight vector is constructed pointing to the positions of each Beidou satellite, and the spatial line-of-sight vector is normalized to obtain the spatial direction vector of a single satellite. Projecting the pseudorange difference of a single star along the corresponding spatial direction vector of the single star yields the pseudorange direction component of the single star. Using the spatial direction vector of a single star as the skeleton direction, and combining the pseudorange direction components of each single star, a directed constraint ray in the local coordinate system is constructed, and pseudorange constraint feature points are determined on the directed constraint ray based on the pseudorange direction components of the single star. According to the azimuth order of each BeiDou satellite relative to the terminal, the pseudorange constraint feature points corresponding to adjacent directed constraint rays are spatially topologically connected to generate BeiDou pseudorange direction constraint bundles.

[0008] Optionally, step three specifically includes: Based on 5G observation data, the 5G base station location, 5G phase measurement value and 5G angle of arrival measurement value of each 5G base station are read, and the terminal coordinates of the previous positioning epoch are read. Convert the 5G phase measurement values ​​into single base station phase distances according to the corresponding 5G signal wavelengths; Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the location of each 5G base station, and use it as the predicted distance to the base station. Subtract the corresponding predicted distance of the base station from the phase distance of the single base station to obtain the phase difference of the single base station; Extract the single base station phase difference value in the continuous positioning epoch according to the 5G base station number, and perform integer cycle back-call. Calculate the change in phase difference value between the single base station phase difference value in the current positioning epoch and the single base station phase difference value in the previous positioning epoch. Divide the change in phase difference value by the corresponding 5G signal wavelength and round down to obtain the integer compensation number. Multiply the integer compensation number by the corresponding 5G signal wavelength to obtain the integer compensation amount. Subtract the integer compensation amount from the single base station phase difference value in the current positioning epoch to obtain the continuous back-call phase difference value. The ground incidence vector of each 5G base station relative to the terminal is determined based on the 5G angle of arrival measurement value; In the local coordinate system, starting from the coordinates of the terminal of the previous positioning epoch, the geometric scale corresponding to the length of the continuous back-connection phase difference is intercepted along the incident direction vector of each ground to generate the single base station phase constraint endpoints corresponding to each 5G base station. Based on the topological adjacency relationship of 5G base stations, the single base station phase constraint endpoints corresponding to adjacent 5G base stations are connected by line segments to generate 5G ground phase constraint chords.

[0009] Optionally, step four specifically involves: Read the terminal coordinates and predicted displacement of the previous positioning epoch from the terminal motion status data; Using the previous epoch terminal coordinates as the origin, the projection direction of the terminal predicted displacement direction on the horizontal plane is taken as the longitudinal axis, a direction perpendicular to the longitudinal axis is constructed in the horizontal plane as the transverse axis, and a direction perpendicular to the horizontal plane and pointing upwards is taken as the elevation axis to establish a star-ground fusion constraint coordinate system. The pseudorange constraint feature points and their corresponding topological connections in the BeiDou pseudorange directional constraint bundle are mapped to the satellite-ground fusion constraint coordinate system through spatial coordinate rotation and translation transformation to generate a satellite domain constraint feature network. Map the single base station phase constraint endpoints and corresponding line segment connections in the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate a ground constraint chord network. Under the space-ground fusion constrained coordinate system, the initial space-ground mesh is generated based on the space-domain constrained feature mesh and the ground-domain constrained chord mesh.

[0010] Optionally, step five specifically includes: Read the star-domain constraint feature network and the ground constraint chord network from the initial star-ground mesh; In the constrained feature network, pseudo-range constraint feature points are read sequentially according to the topological connection relationship. The knot turning angle between two adjacent topological connection edges is calculated. The pseudo-range constraint feature points with knot turning angles greater than or equal to the set knot turning threshold are extracted as key nodes of the network. In the ground-constrained chord network, the single base station phase constraint endpoints are read sequentially according to the line segment connection relationship. The chord turning angle between two adjacent connecting edges is calculated. The corresponding single base station phase constraint endpoints with chord turning angles greater than or equal to the set chord turning threshold are extracted as key nodes for weaving the network. The sparse chord recursion algorithm is executed on the key nodes of the web. Taking any key node of the web as the starting point of the recursion, the adjacent key nodes of the web are read along the topological connection relationship of the star domain constraint feature network or the line segment connection relationship of the ground constraint chord network, and the connection direction between the starting point of the recursion and the adjacent key nodes of the web is taken as the current recursion direction. Continue reading the next key node of the web weaving along the current recursive direction, and calculate the directional offset and spatial offset of the next key node of the web weaving relative to the current recursive direction; When the directional offset does not exceed the set recursive directional threshold and the spatial offset does not exceed the set recursive spatial threshold, the next key node of the weaving network will be incorporated into the current recursive chain. When the directional offset exceeds the set recursive direction threshold or the spatial offset exceeds the set recursive spatial threshold, the current recursive chain ends, and the next key node of the weaving network is used as the new recursive starting point to re-establish the current recursive direction. Sparse pruning is performed on the intermediate key nodes of each recursive chain, and the first and last key nodes of the recursive chain are retained as sparse nodes. Sparse topological edges are directly constructed between the first and last key nodes. Sparse topological edges from the constrained feature network are taken as constrained sparse edges, and sparse topological edges from the constrained string segment network are taken as ground sparse edges. By combining sparse mesh nodes, star-domain sparse edges, and ground sparse edges, a star-ground sparse chord diagram is generated.

[0011] Optionally, step six specifically includes: Read the sparse mesh nodes, star-field sparse edges, and ground sparse edges in the star-ground sparse chord graph; With each sparse mesh node as the center, a sampling radius is extended along the positive and negative directions of the horizontal axis, the positive and negative directions of the vertical axis, and the positive and negative directions of the elevation axis of the star-ground fusion constraint coordinate system to form a local sampling cube; Within the local sampling cube, sampling positions are arranged along the horizontal axis, vertical axis, and elevation axis according to a set sampling interval, and the spatial points corresponding to each sampling position are used as local candidate coordinate sampling points. Read the sparse edges of the star domain connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling point to each sparse edge of the star domain, and take the minimum shortest distance as the shrinkage amount of the star domain direction. Read the ground sparse edges connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling points to each ground sparse edge, and take the minimum shortest distance as the ground chord shrinkage amount. When the corresponding sparse mesh node is not connected to the sparse edge of the star field or the sparse edge of the ground, only the shrinkage amount corresponding to the connected sparse topology edge is retained. The weighted sum of the shrinkage in the star region direction and the shrinkage in the ground chord segment is used to obtain the local shrinkage of the mesh corresponding to the local candidate coordinate sampling point. A three-dimensional numerical matrix covering the local sampling cube is constructed, and the local shrinkage of each web is filled into the corresponding coordinate grid points in the three-dimensional numerical matrix to generate a sparse web shrinkage potential field.

[0012] Optionally, step seven specifically includes: Read the three-dimensional numerical matrix in the sparse web shrinkage potential field, and read the local web shrinkage corresponding to each coordinate grid point in the three-dimensional numerical matrix; In the three-dimensional numerical matrix, the local shrinkage of the mesh corresponding to the current coordinate grid point is compared with that of the adjacent coordinate grid points. The current coordinate grid point whose local shrinkage is less than that of the adjacent coordinate grid points is taken as the candidate shrinkage center point. When there are multiple candidate shrinkage center points, the candidate shrinkage center point with the smallest local shrinkage of the web is selected as the sparse web shrinkage center point. Read the sparse web contraction center point and the surrounding adjacent coordinate grid points, and construct a local interpolation window; Within the local interpolation window, calculate the difference in shrinkage between the local shrinkage of the web corresponding to each adjacent coordinate grid point and the local shrinkage of the web corresponding to the shrinkage center point of the sparse web. The reciprocal of the sum of the shrinkage difference and the preset non-zero perturbation is used as the initial interpolation weight for the corresponding adjacent coordinate grid points; Divide the initial interpolation weight of each adjacent coordinate grid point by the sum of all initial interpolation weights to obtain the normalized interpolation weight of the corresponding adjacent coordinate grid point. Calculate the coordinate offset of each adjacent coordinate grid point relative to the sparse mesh contraction center point, and then sum the coordinate offsets according to the corresponding normalized interpolation weights to obtain the center correction offset. The center correction offset is superimposed on the spatial coordinates of the sparse web shrinkage center point to obtain the corrected shrinkage center point. The corrected shrinkage center point is transformed from the space-ground fusion constraint coordinate system to the spatial coordinate system corresponding to the current positioning epoch, generating the fusion-enhanced positioning coordinates for the current positioning epoch.

[0013] Optionally, step eight specifically includes: Read the fused and enhanced positioning coordinates corresponding to each positioning epoch; According to the positioning epoch sequence, the fused enhanced positioning coordinates corresponding to adjacent positioning epochs are connected to obtain the fused enhanced positioning trajectory; The fusion-enhanced positioning coordinates and fusion-enhanced positioning trajectory corresponding to each positioning epoch are correlated and output to obtain the fusion-enhanced positioning result.

[0014] According to an embodiment of the present invention, a fusion-enhanced positioning system based on BeiDou 5G includes the following modules: The observation data acquisition module is used to acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch. The BeiDou harness generation module is used to calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. The pseudorange difference of the single satellite is projected along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint harness. The 5G chord generation module is used to calculate the phase difference of a single base station based on 5G observation data, and to perform integer loopback on the phase difference of a single base station in continuous positioning epochs. It combines the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. The initial satellite-ground web generation module is used to establish a satellite-ground fusion constraint coordinate system based on terminal motion state data, and to map the BeiDou pseudorange direction constraint bundle and the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground web. The sparse string bundle recursion module is used to extract key nodes of the initial star-ground mesh based on the initial mesh, execute the sparse string bundle recursion algorithm on the key nodes of the mesh, extract sparse mesh nodes, and generate a star-ground sparse string bundle diagram. The contraction potential field generation module is used to generate local candidate coordinate sampling points centered on the sparse mesh nodes in the star-ground sparse chord diagram, and generate a sparse mesh contraction potential field based on the local shrinkage amount of the mesh corresponding to the local candidate coordinate sampling points. The positioning coordinate generation module is used to extract the sparse web contraction center point from the sparse web contraction potential field, and to perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch. The positioning result output module is used to obtain the fusion-enhanced positioning trajectory based on the fusion-enhanced positioning coordinates and output the fusion-enhanced positioning result.

[0015] The beneficial effects of this invention are: This invention addresses the problem that existing fusion methods such as Kalman filtering and weighted least squares mainly rely on the superposition of measurement residuals and have difficulty expressing the structural relationship between BeiDou spatial directional constraints and 5G ground phase continuous constraints by reconstructing the BeiDou pseudorange direction constraint bundle and 5G ground phase constraint chords. It transforms the BeiDou pseudorange error from a simple distance residual into a directional constraint skeleton with spatial orientation order and pseudorange deviation scale, and transforms the 5G phase measurement from a single base station phase value into a ground phase chord with base station topological adjacency relationship and arrival angle directional traction relationship. This allows BeiDou observation data and 5G observation data to no longer participate in fusion as isolated measurement items, but to form an overlapping, recursive, and shrinkable initial satellite-ground network under the satellite-ground fusion constraint coordinate system.

[0016] By applying a sparse chord recursive algorithm to the initial satellite-to-ground network, this invention can sparsely prune redundant intermediate network key nodes while preserving the main transition structures of the satellite-constrained feature network and the ground-constrained chord network. This reduces the computational load of invalid nodes in the positioning solution and avoids the end-side computing burden caused by traditional global candidate point search or multiple rounds of filtering iteration. By constructing a sparse network contraction potential field, this invention transforms the BeiDou directional contraction trend and the 5G chord network contraction trend into the local contraction amount of the network corresponding to the local candidate coordinate sampling points. This makes the generation of fused enhanced positioning coordinates no longer dependent on a single noise matrix or fixed weights, but based on the sparse network between satellite and ground. The geometric contraction relationship of the chord diagram in local space is realized; by extracting the contraction center point of the sparse mesh from the contraction potential field of the sparse mesh and performing local interpolation correction, this invention can reduce the coordinate quantization error caused by the discrete grid resolution and improve the accuracy of the fusion-enhanced positioning coordinates of the current positioning epoch; finally, by connecting the fusion-enhanced positioning coordinates in the order of positioning epochs, the fusion-enhanced positioning trajectory and fusion-enhanced positioning results are output, so that the system can still maintain good spatial structure continuity, positioning result stability and real-time solution capability at the end side in complex environments such as urban canyons, uneven base station distribution, BeiDou pseudorange deviation and 5G phase cycle slip. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a BeiDou 5G-based fusion-enhanced positioning method proposed in this invention; Figure 2 This is a schematic diagram of the structure of a BeiDou 5G-based fusion enhanced positioning system proposed in this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0019] refer to Figure 1 A fusion-enhanced positioning method based on BeiDou and 5G includes the following steps: Step 1: Acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch; Step 2: Calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. Project the pseudorange difference of the single satellite along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint bundle. Step 3: Calculate the phase difference of a single base station based on 5G observation data, and perform integer loopback on the phase difference of a single base station in continuous positioning epochs. Combine the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. Step 4: Establish a satellite-ground fusion constraint coordinate system based on terminal motion status data, and map the BeiDou pseudorange direction constraint bundle and 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground network; Step 5: Extract key nodes of the initial web weaving based on the satellite-ground network, execute the sparse chord recursion algorithm on the key nodes of the web weaving, extract sparse web nodes and generate a satellite-ground sparse chord diagram; Step 6: Generate local candidate coordinate sampling points centered on the sparse web nodes in the star-ground sparse chord diagram, and generate a sparse web contraction potential field based on the local web contraction amount corresponding to the local candidate coordinate sampling points. Step 7: Extract the sparse web contraction center point from the sparse web contraction potential field, and perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch; Step 8: Based on the fusion-enhanced positioning coordinates, obtain the fusion-enhanced positioning trajectory and output the fusion-enhanced positioning result.

[0020] In this embodiment, the BeiDou observation data includes the BeiDou satellite number, BeiDou satellite position, and BeiDou pseudorange; the 5G observation data includes the 5G base station number, 5G base station position, 5G phase measurement value, and 5G angle of arrival measurement value; the terminal motion status data includes the terminal coordinates of the previous positioning epoch and the terminal predicted displacement. The BeiDou observation data, 5G observation data, and terminal motion status data are mapped according to the positioning epoch to obtain the BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch. In the specific implementation process, the positioning epoch is set to 1 second. The positions of the BeiDou satellite, the 5G base station, and the terminal coordinates of the previous positioning epoch are uniformly transformed to the same three-dimensional spatial coordinate system, with the coordinate unit being meters. The BeiDou pseudorange unit is meters, the 5G phase measurement value is recorded according to the corresponding 5G carrier wavelength, and the 5G angle of arrival measurement value is recorded in degrees. The terminal predicted displacement is obtained by integrating the terminal velocity, with the integration time set to 1 second, and the terminal coordinates of the previous positioning epoch are used as the displacement starting point. By unifying the time scale, coordinate units, and spatial reference, BeiDou observation data, 5G observation data, and terminal motion status data can form a stable correspondence under the same positioning epoch. This avoids fusion errors caused by time misalignment and inconsistent coordinate references of data from different sources, providing a stable data foundation for generating BeiDou pseudorange direction constraint bundles and 5G ground phase constraint chords.

[0021] In this embodiment, step two specifically involves: Based on BeiDou observation data, the BeiDou satellite positions and corresponding BeiDou pseudoranges of each BeiDou satellite are read, and the terminal coordinates of the previous positioning epoch are read. Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the positions of each BeiDou satellite, and use it as the single-satellite prediction distance; Subtract the corresponding single-satellite predicted distance from the BeiDou pseudorange to obtain the single-satellite pseudorange difference; Starting from the coordinates of the previous positioning epoch terminal, a spatial line-of-sight vector is constructed pointing to the positions of each Beidou satellite, and the spatial line-of-sight vector is normalized to obtain the spatial direction vector of a single satellite. Projecting the pseudorange difference of a single star along the corresponding spatial direction vector of the single star yields the pseudorange direction component of the single star. Using the spatial direction vector of a single star as the skeleton direction, and combining the pseudorange direction components of each single star, a directed constraint ray in the local coordinate system is constructed, and pseudorange constraint feature points are determined on the directed constraint ray based on the pseudorange direction components of the single star. According to the azimuth order of each Beidou satellite relative to the terminal, the pseudorange constraint feature points corresponding to adjacent directed constraint rays are spatially topologically connected to generate Beidou pseudorange direction constraint bundles. In the specific implementation process, the coordinates of the previous positioning epoch are used as the origin of the local coordinate system. The position of each Beidou satellite is subtracted from the coordinates of the previous positioning epoch to obtain the corresponding spatial line-of-sight vector. The spatial line-of-sight vector consists of a horizontal component, a vertical component, and an elevation component. The magnitude of the spatial line-of-sight vector is calculated by adding the squares of the horizontal component, the vertical component, and the elevation component, and then taking the square root to obtain the length of the spatial line-of-sight vector. Then, the horizontal component, the vertical component, and the elevation component are divided by the length of the spatial line-of-sight vector to obtain a single-satellite spatial direction vector with a length of 1.

[0022] When constructing directed constraint rays, the starting point of the ray is the coordinates of the previous positioning epoch, and the spatial direction vector of a single star is used as the extension direction of the ray. The pseudorange direction component of the single star is used as the length scale attached to the corresponding ray, thereby determining a corresponding pseudorange constraint feature point on each ray. The coordinates of the pseudorange constraint feature point in the local coordinate system are determined by the horizontal projection component, the vertical projection component, and the elevation projection component. When performing spatial topology connection, each directed constraint ray is projected onto the horizontal plane of the local coordinate system, and the azimuth angle of each directed constraint ray relative to the vertical axis of the local coordinate system is calculated. The directed constraint rays are then sorted and read in order of increasing azimuth angle. For each group of adjacent directed constraint rays, the corresponding pseudorange constraint feature points on the two rays are connected by line segments until all adjacent directed constraint rays are connected. This generates a BeiDou pseudorange direction constraint bundle that preserves the spatial distortion morphology of the pseudorange difference of each single star.

[0023] In this embodiment, step three specifically includes: Based on 5G observation data, the 5G base station location, 5G phase measurement value and 5G angle of arrival measurement value of each 5G base station are read, and the terminal coordinates of the previous positioning epoch are read. Convert the 5G phase measurement values ​​into single base station phase distances according to the corresponding 5G signal wavelengths; Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the location of each 5G base station, and use it as the predicted distance to the base station. Subtract the corresponding predicted distance of the base station from the phase distance of the single base station to obtain the phase difference of the single base station; Extract the single base station phase difference value in the continuous positioning epoch according to the 5G base station number, and perform integer cycle back-call. Calculate the change in phase difference value between the single base station phase difference value in the current positioning epoch and the single base station phase difference value in the previous positioning epoch. Divide the change in phase difference value by the corresponding 5G signal wavelength and round down to obtain the integer compensation number. Multiply the integer compensation number by the corresponding 5G signal wavelength to obtain the integer compensation amount. Subtract the integer compensation amount from the single base station phase difference value in the current positioning epoch to obtain the continuous back-call phase difference value. The ground incidence vector of each 5G base station relative to the terminal is determined based on the 5G angle of arrival measurement value; In the local coordinate system, starting from the coordinates of the terminal of the previous positioning epoch, the geometric scale corresponding to the length of the continuous back-connection phase difference is intercepted along the incident direction vector of each ground to generate the single base station phase constraint endpoints corresponding to each 5G base station. Based on the topological adjacency relationship of 5G base stations, the single base station phase constraint endpoints corresponding to adjacent 5G base stations are connected by line segments to generate 5G ground phase constraint chord segments. In the specific implementation process, the 5G signal wavelength is determined according to the carrier frequency used by the 5G base station. When the carrier frequency is set to 3.5GHz, the corresponding 5G signal wavelength is 0.0857 meters. The 5G phase measurement value is recorded in weeks and multiplied by 0.0857 meters to obtain the single base station phase distance. When performing integer cycle backhaul, the single base station phase difference value in the continuous positioning epoch is processed one by one according to the 5G base station number. The change in phase difference value is the single base station phase difference value of the current positioning epoch minus the single base station phase difference value of the previous positioning epoch. The rounding method is used to convert the jump variable in the phase difference value change that exceeds the full wavelength into an integer cycle compensation value. Then, the integer cycle compensation value is deducted from the single base station phase difference value of the current positioning epoch, so that the phase difference value change between adjacent positioning epochs remains continuous. This eliminates the integer wavelength jump of the single base station phase difference value between continuous positioning epochs, and makes the continuous backhaul phase difference value retain the true continuous change. This ensures that the generated single base station phase constraint endpoint and 5G ground phase constraint chord will not be abruptly distorted due to phase cycle jump.

[0024] 5G angle-of-arrival measurements include horizontal azimuth and elevation angles. The horizontal azimuth angle is used to determine the lateral and longitudinal components of the ground incident direction vector in the local coordinate system's horizontal plane, while the elevation angle is used to determine the elevation component. The lateral, longitudinal, and elevation components are then normalized to obtain a ground incident direction vector of length 1. When determining the single-base station phase constraint endpoints, the origin is taken as the point corresponding to the terminal coordinates of the previous positioning epoch in the local coordinate system. The continuous reconnection phase difference is used as a signed length scalar, which is multiplied by the lateral, longitudinal, and elevation components of the ground incident direction vector to obtain the corresponding... The horizontal, vertical, and elevation coordinate increments are calculated. The three-dimensional coordinates of the origin are then added with the corresponding horizontal, vertical, and elevation coordinate increments to generate the single-base station phase constraint endpoints for each 5G base station. The topological adjacency relationship of 5G base stations is established according to the horizontal distance between base stations, and two 5G base stations with a distance of less than or equal to 150 meters are determined as adjacent 5G base stations. According to the topological adjacency relationship of 5G base stations, the single-base station phase constraint endpoints corresponding to adjacent 5G base stations are connected by line segments in the local coordinate system to generate 5G ground phase constraint chords that can express the continuous change of local phase.

[0025] In this embodiment, step four specifically includes: Read the terminal coordinates and predicted displacement of the previous positioning epoch from the terminal motion status data; Using the previous epoch terminal coordinates as the origin, the projection direction of the terminal predicted displacement direction on the horizontal plane is taken as the longitudinal axis, a direction perpendicular to the longitudinal axis is constructed in the horizontal plane as the transverse axis, and a direction perpendicular to the horizontal plane and pointing upwards is taken as the elevation axis to establish a star-ground fusion constraint coordinate system. The pseudorange constraint feature points and their corresponding topological connections in the BeiDou pseudorange directional constraint bundle are mapped to the satellite-ground fusion constraint coordinate system through spatial coordinate rotation and translation transformation to generate a satellite domain constraint feature network. Map the single base station phase constraint endpoints and corresponding line segment connections in the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate a ground constraint chord network. Under the space-ground fusion constrained coordinate system, an initial space-ground mesh is generated based on the space domain constrained feature mesh and the ground constrained chord mesh; In practical implementation, the purpose of establishing a satellite-ground fusion constrained coordinate system is to unify the large-scale geometric skeleton of the satellite expressed by the BeiDou pseudorange direction constraint bundle and the local high-frequency phase field expressed by the 5G ground phase constraint string segment into a local space dominated by the terminal's motion trend. The satellite-ground fusion constrained coordinate system takes the terminal coordinates of the previous positioning epoch as the origin, the projection direction of the terminal's predicted displacement direction on the horizontal plane as the vertical axis, constructs a direction perpendicular to the vertical axis in the horizontal plane as the horizontal axis, and uses a direction perpendicular to the horizontal plane pointing upwards as the elevation axis, thereby ensuring that the vertical axis, horizontal axis, and elevation axis constitute a three-dimensional orthogonal coordinate system.

[0026] In the specific transformation process, the transformation relationship between the local coordinate system and the satellite-ground fusion constrained coordinate system is obtained, and a three-dimensional rotation matrix is ​​constructed based on the azimuth angle of the terminal's predicted displacement direction on the horizontal plane. ; Where R represents the three-dimensional rotation matrix, and θ is the azimuth angle of the terminal's predicted displacement direction in the horizontal plane relative to the longitudinal reference direction of the local coordinate system. When the projected length of the terminal's predicted displacement in the horizontal plane is less than 0.05 meters, θ is determined using the longitudinal axis direction of the previous positioning epoch; when the projected length of the terminal's predicted displacement in the horizontal plane is greater than or equal to 0.05 meters, θ is determined based on the terminal's predicted displacement direction in the current positioning epoch, thereby avoiding frequent axial swings of the satellite-ground fusion constrained coordinate system caused by short-term stationary periods or minor jitter of the terminal.

[0027] When mapping the star-domain constraint feature network, the three-dimensional coordinates of the pseudorange constraint feature points corresponding to the BeiDou pseudorange direction constraint bundle in step two in the local coordinate system are subtracted from the terminal coordinates of the previous positioning epoch to obtain the local coordinates relative to the origin. Then, the local coordinates are multiplied by the three-dimensional rotation matrix R to convert them into spatial coordinates in the star-ground fusion constraint coordinate system. At the same time, the topological connection relationship formed in the BeiDou pseudorange direction constraint bundle according to the azimuth angle order of each BeiDou satellite relative to the terminal remains unchanged, thereby preserving the spatial distortion shape of the BeiDou pseudorange residual in the star-ground fusion constraint coordinate system and generating the star-domain constraint feature network.

[0028] When mapping the ground-constrained chord network, the same three-dimensional rotation matrix R is used to map the three-dimensional coordinates of the single-base station phase constraint endpoints corresponding to each 5G base station in step three to the satellite-ground fusion constraint coordinate system, while maintaining the line segment connection relationship formed by the topological adjacency relationship of 5G base stations, thereby generating the ground-constrained chord network. Since both the satellite-domain constraint feature network and the ground-constrained chord network are mapped to the same coordinate origin, the same axis, and the same scale in the satellite-ground fusion constraint coordinate system, they can overlap topologically in a unified space, avoiding fusion deviations caused by inconsistencies in coordinate references between BeiDou observations and 5G observations.

[0029] When generating the initial satellite-to-ground network, the topological connections in the satellite-constrained feature network and the line segment connections in the ground-constrained chord network are written into the same adjacency list. Each node in the adjacency list stores its three-dimensional coordinates, source type, and connected objects. The source type includes BeiDou pseudorange directional constraint chord sources and 5G ground phase constraint chord sources. In this way, the initial satellite-to-ground network simultaneously includes the BeiDou directional residual structure and the 5G phase continuity structure, providing a unified data foundation for performing sparse chord recursive algorithms on the initial satellite-to-ground network, extracting key nodes of the network, and generating a satellite-to-ground sparse chord graph.

[0030] In this embodiment, step five specifically includes: Read the star-domain constraint feature network and the ground constraint chord network from the initial star-ground mesh; In the constrained feature network, pseudo-range constraint feature points are read sequentially according to the topological connection relationship. The knot turning angle between two adjacent topological connection edges is calculated. The pseudo-range constraint feature points with knot turning angles greater than or equal to the set knot turning threshold are extracted as key nodes of the network. In the ground-constrained chord network, the single base station phase constraint endpoints are read sequentially according to the line segment connection relationship. The chord turning angle between two adjacent connecting edges is calculated. The corresponding single base station phase constraint endpoints with chord turning angles greater than or equal to the set chord turning threshold are extracted as key nodes for weaving the network. The sparse chord recursion algorithm is executed on the key nodes of the web. Taking any key node of the web as the starting point of the recursion, the adjacent key nodes of the web are read along the topological connection relationship of the star domain constraint feature network or the line segment connection relationship of the ground constraint chord network, and the connection direction between the starting point of the recursion and the adjacent key nodes of the web is taken as the current recursion direction. Continue reading the next key node of the web weaving along the current recursive direction, and calculate the directional offset and spatial offset of the next key node of the web weaving relative to the current recursive direction; When the directional offset does not exceed the set recursive directional threshold and the spatial offset does not exceed the set recursive spatial threshold, the next key node of the weaving network will be incorporated into the current recursive chain. When the directional offset exceeds the set recursive direction threshold or the spatial offset exceeds the set recursive spatial threshold, the current recursive chain ends, and the next key node of the weaving network is used as the new recursive starting point to re-establish the current recursive direction. Sparse pruning is performed on the intermediate key nodes of each recursive chain, and the first and last key nodes of the recursive chain are retained as sparse nodes. Sparse topological edges are directly constructed between the first and last key nodes. Sparse topological edges from the constrained feature network are taken as constrained sparse edges, and sparse topological edges from the constrained string segment network are taken as ground sparse edges. By combining sparse mesh nodes, star-domain sparse edges, and ground sparse edges, a star-ground sparse chord diagram is generated. In the specific implementation process, the beam turning threshold is set to 25 degrees, and the chord turning threshold is set to 20 degrees. In the satellite constrained feature network, taking the current pseudorange constrained feature point as the common endpoint, its preceding and following topological connection edges are read respectively. The angle between the direction vectors of the two topological connection edges is calculated to obtain the beam turning angle. If the beam turning angle is greater than or equal to 25 degrees, it indicates that the BeiDou pseudorange directional constraint beam has undergone a significant geometric turning, and the corresponding pseudorange constrained feature point is retained as a key node in the network. In the ground constrained chord network, taking the current single base station phase constraint endpoint as the common endpoint, its preceding and following connection edges are read respectively. The angle between the direction vectors of the two connection edges is calculated to obtain the chord turning angle. If the chord turning angle is greater than or equal to 20 degrees, it indicates that the 5G ground phase at that location has undergone a significant geometric turning. When a local phase geometric bend occurs in the sparse chord segment, the corresponding single-base station phase constraint endpoint is retained as a key node in the sparse chord recursion algorithm. When executing the sparse chord recursion algorithm, the recursion direction threshold is set to 12 degrees, and the recursion space threshold is set to 0.5 meters. The directional offset is the angle between the current recursion direction and the connection direction of the next key node in the sparse chord, and the spatial offset is the vertical distance from the next key node in the sparse chord to the extension line of the current recursion direction. If the directional offset does not exceed 12 degrees and the spatial offset does not exceed 0.5 meters, the key node in the sparse chord is considered to still be within the same approximately straight constraint segment and is incorporated into the current recursion chain. If the directional offset exceeds 12 degrees or the spatial offset exceeds 0.5 meters, the current recursion chain is terminated, and the next key node in the sparse chord is taken as the new recursion starting point. The current recursion direction is re-established to start the next round of recursion. By sparsely pruning the key intermediate nodes in each recursive chain and replacing the original multi-segment connections with sparse topological edges, the number of computational nodes for the sparsely woven shrinkage potential field can be reduced while preserving the main turning structures of the star-domain constrained feature network and the ground-constrained chord network, thereby improving the real-time performance of the fusion-enhanced positioning.

[0031] In this embodiment, step six specifically includes: Read the sparse mesh nodes, star-field sparse edges, and ground sparse edges in the star-ground sparse chord graph; With each sparse mesh node as the center, a sampling radius is extended along the positive and negative directions of the horizontal axis, the positive and negative directions of the vertical axis, and the positive and negative directions of the elevation axis of the star-ground fusion constraint coordinate system to form a local sampling cube; Within the local sampling cube, sampling positions are arranged along the horizontal axis, vertical axis, and elevation axis according to a set sampling interval, and the spatial points corresponding to each sampling position are used as local candidate coordinate sampling points. Read the sparse edges of the star domain connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling point to each sparse edge of the star domain, and take the minimum shortest distance as the shrinkage amount of the star domain direction. Read the ground sparse edges connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling points to each ground sparse edge, and take the minimum shortest distance as the ground chord shrinkage amount. When the corresponding sparse mesh node is not connected to the sparse edge of the star field or the sparse edge of the ground, only the shrinkage amount corresponding to the connected sparse topology edge is retained. The weighted sum of the shrinkage in the star region direction and the shrinkage in the ground chord segment is used to obtain the local shrinkage of the mesh corresponding to the local candidate coordinate sampling point. Construct a three-dimensional numerical matrix covering the local sampling cube, and fill the local shrinkage of each web into the corresponding coordinate grid points in the three-dimensional numerical matrix to generate a sparse web shrinkage potential field; In the specific implementation process, the sampling radius is set to 2.0 meters, and the sampling interval is set to 0.2 meters. With the sparse mesh node as the center, sampling ranges from -2.0 meters to +2.0 meters are taken on the horizontal, vertical, and elevation axes, respectively, and the sampling positions are arranged at 0.2-meter intervals, thus forming local candidate coordinate sampling points covering the local space surrounding the sparse mesh node. When calculating the shortest distance, both the star-field sparse edge and the ground sparse edge are treated as spatially finite line segments. For any local candidate coordinate sampling point, it is first vertically projected onto the straight line direction of the corresponding sparse edge to obtain the projection point. When the projection point falls... When the two endpoints of a finite line segment are within the same space, the Euclidean distance between the local candidate coordinate sampling point and the projection point is taken as the shortest distance. When the projection point exceeds the endpoint range of the finite line segment, the Euclidean distance between the local candidate coordinate sampling point and the nearest endpoint of the line segment is taken as the shortest distance. For the same local candidate coordinate sampling point, if it is connected to multiple sparse star field edges, the minimum value among the shortest distances corresponding to all sparse star field edges is selected as the star field direction contraction amount. If it is connected to multiple sparse ground edges, the minimum value among the shortest distances corresponding to all sparse ground edges is selected as the ground chord contraction amount.

[0032] When performing weighted summation to synthesize the local shrinkage of the sparse mesh, if the current sparse mesh node is connected to both a sparse edge in the star region and a sparse edge on the ground, the weight of the shrinkage in the star region direction is set to 0.5, and the weight of the shrinkage in the ground chord segment is also set to 0.5. If the current sparse mesh node is not connected to either a sparse edge in the star region or a sparse edge on the ground, only the shrinkage corresponding to the connected sparse topological edge is retained, and the preset weight of this retained shrinkage is set to 1. A three-dimensional numerical matrix is ​​established using a voxel grid with a grid resolution of 0.2 meters. Each coordinate grid point stores a corresponding local shrinkage of the mesh, thereby reducing the computational load and memory throughput of high-dimensional matrices while preserving the centripetal shrinkage trend of sparse edges in the star region and on the ground, and improving the real-time performance of the fusion-enhanced positioning.

[0033] In this embodiment, step seven specifically includes: Read the three-dimensional numerical matrix in the sparse web shrinkage potential field, and read the local web shrinkage corresponding to each coordinate grid point in the three-dimensional numerical matrix; In the three-dimensional numerical matrix, the local shrinkage of the mesh corresponding to the current coordinate grid point is compared with that of the adjacent coordinate grid points. The current coordinate grid point whose local shrinkage is less than that of the adjacent coordinate grid points is taken as the candidate shrinkage center point. When there are multiple candidate shrinkage center points, the candidate shrinkage center point with the smallest local shrinkage of the web is selected as the sparse web shrinkage center point. Read the sparse web contraction center point and the surrounding adjacent coordinate grid points, and construct a local interpolation window; Within the local interpolation window, calculate the difference in shrinkage between the local shrinkage of the web corresponding to each adjacent coordinate grid point and the local shrinkage of the web corresponding to the shrinkage center point of the sparse web. The reciprocal of the sum of the shrinkage difference and the preset non-zero perturbation is used as the initial interpolation weight for the corresponding adjacent coordinate grid points; Divide the initial interpolation weight of each adjacent coordinate grid point by the sum of all initial interpolation weights to obtain the normalized interpolation weight of the corresponding adjacent coordinate grid point. Calculate the coordinate offset of each adjacent coordinate grid point relative to the sparse mesh contraction center point, and then sum the coordinate offsets according to the corresponding normalized interpolation weights to obtain the center correction offset. The center correction offset is superimposed on the spatial coordinates of the sparse web shrinkage center point to obtain the corrected shrinkage center point. The corrected shrinkage center point is transformed from the star-ground fusion constraint coordinate system to the spatial coordinate system corresponding to the current positioning epoch, generating the fusion-enhanced positioning coordinates for the current positioning epoch. In the specific implementation process, the three-dimensional numerical matrix adopts a grid resolution of 0.2 meters, and the adjacent grid points of the current coordinate grid point are taken from its surrounding 26 three-dimensional neighboring grid points. When the local shrinkage of the current coordinate grid point is less than the local shrinkage of the corresponding 26 adjacent coordinate grid points, it is determined as a candidate shrinkage center point. The local interpolation window is set to a 3×3×3 voxel window, with the sparse mesh shrinkage center point as the center, and the surrounding 26 adjacent coordinate grid points are read to participate in the correction. The preset non-zero perturbation is set to 1×10. -6 This is used to avoid reciprocal calculation anomalies when the shrinkage difference is zero. When calculating the initial interpolation weights, the smaller the shrinkage difference, the closer the corresponding adjacent coordinate grid point is to the low potential field region, and the larger its initial interpolation weight. Then, each initial interpolation weight is divided by the sum of all initial interpolation weights, so that the sum of all normalized interpolation weights equals 1. The coordinate offset is calculated separately for the horizontal axis, vertical axis, and elevation axis, that is, the three-dimensional coordinates of the adjacent coordinate grid points are subtracted from the three-dimensional coordinates of the sparse mesh shrinkage center point to obtain the horizontal offset, vertical offset, and elevation offset. The offsets in each direction are weighted and summed according to the normalized interpolation weights to obtain the center correction offset. By superimposing the center correction offset onto the spatial coordinates of the sparse mesh shrinkage center point, the corrected shrinkage center point is obtained, thus performing sub-grid-level correction by utilizing the shrinkage trend of the surrounding adjacent coordinate grid points without changing the dominant role of the lowest potential field point.

[0034] When transforming the corrected contraction center point from the satellite-ground fusion constraint coordinate system to the spatial coordinate system corresponding to the current positioning epoch, the origin, horizontal axis unit direction vector, vertical axis unit direction vector, and elevation axis unit direction vector of the satellite-ground fusion constraint coordinate system are read. The origin is the terminal coordinate of the previous positioning epoch, the vertical axis unit direction vector is the vector obtained by normalizing the projection direction of the terminal predicted displacement direction on the horizontal plane, the horizontal axis unit direction vector is the unit vector in the horizontal plane that is perpendicular to the vertical axis unit direction vector, and the elevation axis unit direction vector is the unit vector that is perpendicular to the horizontal plane and pointing upwards. Let the lateral, longitudinal, and elevation coordinates of the corrected contraction center point in the satellite-to-ground fusion constraint coordinate system be Δx, Δy, and Δz, respectively. Multiplying Δx by the unit direction vector of the lateral axis yields the lateral coordinate increment vector; multiplying Δy by the unit direction vector of the longitudinal axis yields the longitudinal coordinate increment vector; and multiplying Δz by the unit direction vector of the elevation axis yields the elevation coordinate increment vector. These lateral, longitudinal, and elevation coordinate increment vectors are then superimposed on the origin to obtain the spatial coordinates corresponding to the current positioning epoch. These spatial coordinates serve as the fusion-enhanced positioning coordinates for the current positioning epoch. Through this reverse coordinate transformation, the corrected contraction center point no longer remains within the local satellite-to-ground fusion constraint coordinate system but is restored to the actual spatial coordinate representation of the current positioning epoch, allowing it to be directly used for fusion-enhanced positioning trajectory connection.

[0035] In this embodiment, step eight specifically includes: Read the fused and enhanced positioning coordinates corresponding to each positioning epoch; According to the positioning epoch sequence, the fused enhanced positioning coordinates corresponding to adjacent positioning epochs are connected to obtain the fused enhanced positioning trajectory; The fusion-enhanced positioning coordinates and fusion-enhanced positioning trajectory corresponding to each positioning epoch are correlated and output to obtain the fusion-enhanced positioning result.

[0036] refer to Figure 2 A BeiDou / 5G-based fusion enhanced positioning system includes the following modules: The observation data acquisition module is used to acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch. The BeiDou harness generation module is used to calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. The pseudorange difference of the single satellite is projected along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint harness. The 5G chord generation module is used to calculate the phase difference of a single base station based on 5G observation data, and to perform integer loopback on the phase difference of a single base station in continuous positioning epochs. It combines the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. The initial satellite-ground web generation module is used to establish a satellite-ground fusion constraint coordinate system based on terminal motion state data, and to map the BeiDou pseudorange direction constraint bundle and the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground web. The sparse string bundle recursion module is used to extract key nodes of the initial star-ground mesh based on the initial mesh, execute the sparse string bundle recursion algorithm on the key nodes of the mesh, extract sparse mesh nodes, and generate a star-ground sparse string bundle diagram. The contraction potential field generation module is used to generate local candidate coordinate sampling points centered on the sparse mesh nodes in the star-ground sparse chord diagram, and generate a sparse mesh contraction potential field based on the local shrinkage amount of the mesh corresponding to the local candidate coordinate sampling points. The positioning coordinate generation module is used to extract the sparse web contraction center point from the sparse web contraction potential field, and to perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch. The positioning result output module is used to obtain the fusion-enhanced positioning trajectory based on the fusion-enhanced positioning coordinates and output the fusion-enhanced positioning result.

[0037] Example 1: To verify the feasibility of this invention in practice, it was applied to a continuous positioning test of a mobile terminal on a comprehensive urban road network. The test area included densely built-up roads, underground entrances / exits, obstructed sections under overpasses, and open roads. The terminal was installed in a test vehicle, and the total test duration was 46 minutes. The positioning epoch was set to 1 second, and a total of 2760 positioning epochs were collected. In this test scenario, the densely built-up roads exhibited issues such as variations in the number of visible BeiDou satellites and pseudorange fluctuations at low elevation angles; the underground entrances / exits showed issues such as 5G base station angle of arrival offsets and short-term jumps in phase measurements; and the sections under overpasses exhibited simultaneous instability in both BeiDou and 5G observation data. Existing BeiDou / 5G fusion methods often employ weighted least squares, extended Kalman filtering, or factor graph optimization for numerical fusion, which are prone to positioning point offsets, local trajectory breaks, and short-term jumps in the aforementioned scenarios.

[0038] During testing, the terminal acquires BeiDou observation data, 5G observation data, and terminal motion status data every second for the same positioning epoch. BeiDou observation data includes BeiDou satellite IDs, satellite positions, and pseudoranges; 5G observation data includes 5G base station IDs, base station positions, 5G phase measurements, and angle of arrival measurements; and terminal motion status data includes the terminal coordinates and predicted displacement from the previous positioning epoch. The system first calculates the difference between the predicted distance and pseudorange of a single satellite based on the BeiDou observation data. Then, it obtains the spatial direction vector of each satellite based on its position relative to the terminal. The pseudorange difference is projected along the corresponding spatial direction vector to form a BeiDou pseudorange direction constraint bundle. This processing transforms BeiDou observations from mere distance residuals into geometric constraints with directional relationships within the satellite domain.

[0039] For 5G observation data, the system converts 5G phase measurements into single-base station phase distances according to the 0.0857-meter wavelength corresponding to the 3.5GHz carrier, and calculates the single-base station phase difference. During continuous positioning epochs, when the phase difference change occurs with an integer wavelength jump, the system performs integer-cycle backtracking to obtain continuous backtracking phase difference values. Subsequently, the system determines the ground incident direction vector by combining the 5G angle of arrival measurement value, and connects the single-base station phase constraint endpoints corresponding to adjacent base stations according to the 5G base station topological adjacency relationship, generating 5G ground phase constraint chords. In this way, 5G observation data is expressed as a ground constraint structure with local continuity, rather than isolated base station measurements.

[0040] In subsequent processing, the system establishes a satellite-to-ground fusion constraint coordinate system based on the terminal's predicted displacement, mapping the BeiDou pseudorange direction constraint bundle and the 5G ground phase constraint chord to the same coordinate space to generate an initial satellite-to-ground web. For redundant nodes in the initial web, the system extracts key web nodes and executes a sparse chord recursive algorithm, incorporating approximately unidirectional key web nodes with small spatial offsets into the same recursive chain. The intermediate key web nodes in the recursive chain are then pruned to generate a satellite-to-ground sparse chord diagram. Subsequently, a local sampling cube is constructed centered on the sparse web nodes, with a sampling radius of 2.0 meters and a sampling interval of 0.2 meters. The local web shrinkage corresponding to the local candidate coordinate sampling points is calculated, forming a sparse web shrinkage potential field. The sparse web shrinkage center point is extracted from the sparse web shrinkage potential field, and local interpolation correction is performed using a 3×3×3 voxel window. Finally, the fusion-enhanced positioning coordinates for the current positioning epoch are generated, and then connected in the positioning epoch sequence to form the fusion-enhanced positioning trajectory.

[0041] Table 1 Comparison of Positioning Accuracy and Trajectory Continuity

[0042] The root mean square error of horizontal positioning in Table 1 above is used to reflect the overall deviation between the terminal plane position and the reference trajectory. The root mean square error of elevation positioning is used to reflect the positioning stability in the height direction. 95% positioning error means that the error of 95% of the positioning points does not exceed this value. The number of trajectory jumps is used to measure the number of abnormal abrupt changes in the positioning trajectory. As shown in Table 1, the root mean square error (RMSE) of horizontal positioning using traditional BeiDou single-source positioning is 3.86m, the 95% positioning error reaches 8.95m, and the number of trajectory jumps is 37, indicating that it is easily affected by the error of a single satellite observation in environments with obstruction or multipath. BeiDou / 5G weighted least squares fusion and extended Kalman filter fusion, after introducing 5G observation data, have reduced both the error and the number of jumps, but still mainly rely on numerical residual fusion, which does not fully utilize the constraints of the space-ground spatial structure. BeiDou / 5G factor map optimization fusion further reduces the error, but the number of trajectory jumps is still 11. The method of this invention reduces the RMSE of horizontal positioning to 0.82m, the RMSE of vertical positioning to 1.26m, the 95% positioning error to 1.91m, and the number of trajectory jumps to only 4, indicating that this invention can significantly improve the accuracy, stability, and trajectory continuity of fusion-enhanced positioning in complex environments.

[0043] Furthermore, in order to verify the effectiveness of the present invention in terms of real-time calculation efficiency and adaptability to occluded road sections, the calculation time, memory usage and effective output of different fusion positioning schemes under end-side operating conditions were statistically analyzed, and the results are shown in Table 2.

[0044] Table 2 Comparison of Real-time Solving Efficiency and Adaptability to Complex Scenarios

[0045] The average single-epoch calculation time in Table 2 above reflects the time required for each positioning epoch to complete the positioning calculation. The peak memory usage reflects the storage resource requirements of the algorithm when running on the edge. The effective positioning rate of obstructed road sections measures the proportion of effective positioning results that can still be output in obstructed environments such as elevated roads, dense buildings, or underground entrances and exits. As shown in Table 2, the average single-epoch calculation time for BeiDou single-source positioning is the lowest, at only 18.6ms, with a peak memory usage of 42MB. However, its effective positioning rate for obstructed road sections is only 81.3%, indicating that although it is computationally simple, its stability in complex environments is insufficient. The calculation times for weighted least squares fusion and extended Kalman filter fusion are 34.8ms and 41.5ms, respectively, with effective positioning rates for obstructed road sections increasing to 88.5% and 91.2%, but they still suffer from insufficient adaptability to complex environments. The effective positioning rate for obstructed road sections using factor graph optimization fusion reaches 94.6%, but the calculation time increases to 96.3ms, with a peak memory usage of 168MB, resulting in significant real-time operational pressure on the edge. The average single-epoch calculation time for the method of this invention is 53.7ms, with a peak memory usage of 91MB, and an effective positioning rate for obstructed road sections reaching 97.8%. This indicates that the invention maintains high positioning effectiveness while keeping computational resources within a reasonable range, balancing positioning stability and real-time calculation capabilities.

[0046] This embodiment verifies the effectiveness of the present invention in complex obstruction and multi-source observation unstable scenarios. The present invention, through the coordinated processing of BeiDou pseudorange direction constraint bundles, 5G ground phase constraint chord segments, satellite-ground sparse chord diagrams, and sparse mesh contraction potential fields, can reduce the problems of location point jumps and trajectory drift in traditional filtering fusion, and improve the accuracy and continuity of fused enhanced positioning coordinates. At the same time, the sparse chord recursive algorithm reduces redundant calculation nodes, enabling the present invention to maintain a high effective positioning rate in obstructed road sections while taking into account the real-time solution requirements of the end side, proving that the method of the present invention has good engineering feasibility and application value.

[0047] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A fusion-enhanced positioning method based on BeiDou and 5G, characterized in that, Includes the following steps: Step 1: Acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch; Step 2: Calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. Project the pseudorange difference of the single satellite along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint bundle. Step 3: Calculate the phase difference of a single base station based on 5G observation data, and perform integer loopback on the phase difference of a single base station in continuous positioning epochs. Combine the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. Step 4: Establish a satellite-ground fusion constraint coordinate system based on terminal motion status data, and map the BeiDou pseudorange direction constraint bundle and 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground network; Step 5: Extract key nodes of the initial web weaving based on the satellite-ground network, execute the sparse chord recursion algorithm on the key nodes of the web weaving, extract sparse web nodes and generate a satellite-ground sparse chord diagram; Step 6: Generate local candidate coordinate sampling points centered on the sparse web nodes in the star-ground sparse chord diagram, and generate a sparse web contraction potential field based on the local web contraction amount corresponding to the local candidate coordinate sampling points. Step 7: Extract the sparse web contraction center point from the sparse web contraction potential field, and perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch; Step 8: Based on the fusion-enhanced positioning coordinates, obtain the fusion-enhanced positioning trajectory and output the fusion-enhanced positioning result.

2. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, The BeiDou observation data includes BeiDou satellite number, BeiDou satellite position, and BeiDou pseudorange; the 5G observation data includes 5G base station number, 5G base station position, 5G phase measurement value, and 5G angle of arrival measurement value; the terminal motion status data includes the terminal coordinates of the previous positioning epoch and the terminal predicted displacement. By mapping the BeiDou observation data, 5G observation data, and terminal motion status data according to the positioning epoch, we can obtain the BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch.

3. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step two specifically involves: Based on BeiDou observation data, the BeiDou satellite positions and corresponding BeiDou pseudoranges of each BeiDou satellite are read, and the terminal coordinates of the previous positioning epoch are read. Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the positions of each BeiDou satellite, and use it as the single-satellite prediction distance; Subtract the corresponding single-satellite predicted distance from the BeiDou pseudorange to obtain the single-satellite pseudorange difference; Starting from the coordinates of the previous positioning epoch terminal, a spatial line-of-sight vector is constructed pointing to the positions of each Beidou satellite, and the spatial line-of-sight vector is normalized to obtain the spatial direction vector of a single satellite. Projecting the pseudorange difference of a single star along the corresponding spatial direction vector of the single star yields the pseudorange direction component of the single star. Using the spatial direction vector of a single star as the skeleton direction, and combining the pseudorange direction components of each single star, a directed constraint ray in the local coordinate system is constructed, and pseudorange constraint feature points are determined on the directed constraint ray based on the pseudorange direction components of the single star. According to the azimuth order of each BeiDou satellite relative to the terminal, the pseudorange constraint feature points corresponding to adjacent directed constraint rays are spatially topologically connected to generate BeiDou pseudorange direction constraint bundles.

4. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step three specifically involves: Based on 5G observation data, the 5G base station location, 5G phase measurement value and 5G angle of arrival measurement value of each 5G base station are read, and the terminal coordinates of the previous positioning epoch are read. Convert the 5G phase measurement values ​​into single base station phase distances according to the corresponding 5G signal wavelengths; Calculate the straight-line distance between the terminal coordinates of the previous positioning epoch and the location of each 5G base station, and use it as the predicted distance to the base station. Subtract the corresponding predicted distance of the base station from the phase distance of the single base station to obtain the phase difference of the single base station; Extract the single base station phase difference value in the continuous positioning epoch according to the 5G base station number, and perform integer cycle back-call. Calculate the change in phase difference value between the single base station phase difference value in the current positioning epoch and the single base station phase difference value in the previous positioning epoch. Divide the change in phase difference value by the corresponding 5G signal wavelength and round down to obtain the integer compensation number. Multiply the integer compensation number by the corresponding 5G signal wavelength to obtain the integer compensation amount. Subtract the integer compensation amount from the single base station phase difference value in the current positioning epoch to obtain the continuous back-call phase difference value. The ground incidence vector of each 5G base station relative to the terminal is determined based on the 5G angle of arrival measurement value; In the local coordinate system, starting from the coordinates of the terminal of the previous positioning epoch, the geometric scale corresponding to the length of the continuous back-connection phase difference is intercepted along the incident direction vector of each ground to generate the single base station phase constraint endpoints corresponding to each 5G base station. Based on the topological adjacency relationship of 5G base stations, the single base station phase constraint endpoints corresponding to adjacent 5G base stations are connected by line segments to generate 5G ground phase constraint chord segments.

5. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step four specifically involves: Read the terminal coordinates and predicted displacement of the previous positioning epoch from the terminal motion status data; Using the previous epoch terminal coordinates as the origin, the projection direction of the terminal predicted displacement direction on the horizontal plane is taken as the longitudinal axis, a direction perpendicular to the longitudinal axis is constructed in the horizontal plane as the transverse axis, and a direction perpendicular to the horizontal plane and pointing upwards is taken as the elevation axis to establish a star-ground fusion constraint coordinate system. The pseudorange constraint feature points and their corresponding topological connections in the BeiDou pseudorange directional constraint bundle are mapped to the satellite-ground fusion constraint coordinate system through spatial coordinate rotation and translation transformation to generate a satellite domain constraint feature network. Map the single base station phase constraint endpoints and corresponding line segment connections in the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate a ground constraint chord network. Under the space-ground fusion constrained coordinate system, the initial space-ground mesh is generated based on the space-domain constrained feature mesh and the ground-domain constrained chord mesh.

6. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step five specifically involves: Read the star-domain constraint feature network and the ground constraint chord network from the initial star-ground mesh; In the constrained feature network, pseudo-range constraint feature points are read sequentially according to the topological connection relationship. The knot turning angle between two adjacent topological connection edges is calculated. The pseudo-range constraint feature points with knot turning angles greater than or equal to the set knot turning threshold are extracted as key nodes of the network. In the ground-constrained chord network, the single base station phase constraint endpoints are read sequentially according to the line segment connection relationship. The chord turning angle between two adjacent connecting edges is calculated. The corresponding single base station phase constraint endpoints with chord turning angles greater than or equal to the set chord turning threshold are extracted as key nodes for weaving the network. The sparse chord recursion algorithm is executed on the key nodes of the web. Taking any key node of the web as the starting point of the recursion, the adjacent key nodes of the web are read along the topological connection relationship of the star domain constraint feature network or the line segment connection relationship of the ground constraint chord network, and the connection direction between the starting point of the recursion and the adjacent key nodes of the web is taken as the current recursion direction. Continue reading the next key node of the web weaving along the current recursive direction, and calculate the directional offset and spatial offset of the next key node of the web weaving relative to the current recursive direction; When the directional offset does not exceed the set recursive directional threshold and the spatial offset does not exceed the set recursive spatial threshold, the next key node of the weaving network will be incorporated into the current recursive chain. When the directional offset exceeds the set recursive direction threshold or the spatial offset exceeds the set recursive spatial threshold, the current recursive chain ends, and the next key node of the weaving network is used as the new recursive starting point to re-establish the current recursive direction. Sparse pruning is performed on the intermediate key nodes of each recursive chain, and the first and last key nodes of the recursive chain are retained as sparse nodes. Sparse topological edges are directly constructed between the first and last key nodes. Sparse topological edges from the constrained feature network are taken as constrained sparse edges, and sparse topological edges from the constrained string segment network are taken as ground sparse edges. By combining sparse mesh nodes, star-domain sparse edges, and ground sparse edges, a star-ground sparse chord diagram is generated.

7. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step six specifically involves: Read the sparse mesh nodes, star-field sparse edges, and ground sparse edges in the star-ground sparse chord graph; With each sparse mesh node as the center, a sampling radius is extended along the positive and negative directions of the horizontal axis, the positive and negative directions of the vertical axis, and the positive and negative directions of the elevation axis of the star-ground fusion constraint coordinate system to form a local sampling cube; Within the local sampling cube, sampling positions are arranged along the horizontal axis, vertical axis, and elevation axis according to a set sampling interval, and the spatial points corresponding to each sampling position are used as local candidate coordinate sampling points. Read the sparse edges of the star domain connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling point to each sparse edge of the star domain, and take the minimum shortest distance as the shrinkage amount of the star domain direction. Read the ground sparse edges connected to the corresponding sparse mesh nodes, calculate the shortest distance from the local candidate coordinate sampling points to each ground sparse edge, and take the minimum shortest distance as the ground chord shrinkage amount. When the corresponding sparse mesh node is not connected to the sparse edge of the star field or the sparse edge of the ground, only the shrinkage amount corresponding to the connected sparse topology edge is retained. The weighted sum of the shrinkage in the star region direction and the shrinkage in the ground chord segment is used to obtain the local shrinkage of the mesh corresponding to the local candidate coordinate sampling point. A three-dimensional numerical matrix covering the local sampling cube is constructed, and the local shrinkage of each web is filled into the corresponding coordinate grid points in the three-dimensional numerical matrix to generate a sparse web shrinkage potential field.

8. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step seven specifically involves: Read the three-dimensional numerical matrix in the sparse web shrinkage potential field, and read the local web shrinkage corresponding to each coordinate grid point in the three-dimensional numerical matrix; In the three-dimensional numerical matrix, the local shrinkage of the mesh corresponding to the current coordinate grid point is compared with that of the adjacent coordinate grid points. The current coordinate grid point whose local shrinkage is less than that of the adjacent coordinate grid points is taken as the candidate shrinkage center point. When there are multiple candidate shrinkage center points, the candidate shrinkage center point with the smallest local shrinkage of the web is selected as the sparse web shrinkage center point. Read the sparse web contraction center point and the surrounding adjacent coordinate grid points, and construct a local interpolation window; Within the local interpolation window, calculate the difference in shrinkage between the local shrinkage of the web corresponding to each adjacent coordinate grid point and the local shrinkage of the web corresponding to the shrinkage center point of the sparse web. The reciprocal of the sum of the shrinkage difference and the preset non-zero perturbation is used as the initial interpolation weight for the corresponding adjacent coordinate grid points; Divide the initial interpolation weight of each adjacent coordinate grid point by the sum of all initial interpolation weights to obtain the normalized interpolation weight of the corresponding adjacent coordinate grid point. Calculate the coordinate offset of each adjacent coordinate grid point relative to the sparse mesh contraction center point, and then sum the coordinate offsets according to the corresponding normalized interpolation weights to obtain the center correction offset. The center correction offset is superimposed on the spatial coordinates of the sparse web shrinkage center point to obtain the corrected shrinkage center point. The corrected shrinkage center point is transformed from the space-ground fusion constraint coordinate system to the spatial coordinate system corresponding to the current positioning epoch, generating the fusion-enhanced positioning coordinates for the current positioning epoch.

9. The fusion-enhanced positioning method based on BeiDou 5G according to claim 1, characterized in that, Step eight specifically involves: Read the fused and enhanced positioning coordinates corresponding to each positioning epoch; According to the positioning epoch sequence, the fused enhanced positioning coordinates corresponding to adjacent positioning epochs are connected to obtain the fused enhanced positioning trajectory; The fusion-enhanced positioning coordinates and fusion-enhanced positioning trajectory corresponding to each positioning epoch are correlated and output to obtain the fusion-enhanced positioning result.

10. A BeiDou-5G-based fusion enhanced positioning system, comprising the BeiDou-5G-based fusion enhanced positioning method according to any one of claims 1 to 9, characterized in that, Includes the following modules: The observation data acquisition module is used to acquire BeiDou observation data, 5G observation data, and terminal motion status data under the same positioning epoch. The BeiDou harness generation module is used to calculate the pseudorange difference of a single satellite based on BeiDou observation data and obtain the spatial direction vector of the single satellite. The pseudorange difference of the single satellite is projected along the corresponding spatial direction vector of the single satellite to generate the BeiDou pseudorange direction constraint harness. The 5G chord generation module is used to calculate the phase difference of a single base station based on 5G observation data, and to perform integer loopback on the phase difference of a single base station in continuous positioning epochs. It combines the 5G angle of arrival measurement value and the spatial adjacency relationship of 5G base stations to generate 5G ground phase constraint chords. The initial satellite-ground web generation module is used to establish a satellite-ground fusion constraint coordinate system based on terminal motion state data, and to map the BeiDou pseudorange direction constraint bundle and the 5G ground phase constraint chord to the satellite-ground fusion constraint coordinate system to generate the initial satellite-ground web. The sparse string bundle recursion module is used to extract key nodes of the initial star-ground mesh based on the initial mesh, execute the sparse string bundle recursion algorithm on the key nodes of the mesh, extract sparse mesh nodes, and generate a star-ground sparse string bundle diagram. The contraction potential field generation module is used to generate local candidate coordinate sampling points centered on the sparse mesh nodes in the star-ground sparse chord diagram, and generate a sparse mesh contraction potential field based on the local shrinkage amount of the mesh corresponding to the local candidate coordinate sampling points. The positioning coordinate generation module is used to extract the sparse web contraction center point from the sparse web contraction potential field, and to perform local interpolation correction on the sparse web contraction center point to generate the fused enhanced positioning coordinates of the current positioning epoch. The positioning result output module is used to obtain the fusion-enhanced positioning trajectory based on the fusion-enhanced positioning coordinates and output the fusion-enhanced positioning result.

Images