Roadside parking space management method and system based on Beidou positioning system

Abstract

The invention discloses a roadside parking space management method and system based on a Beidou positioning system, and relates to the technical field of roadside parking space management, and the method comprises the following steps: collecting the current position coordinates of a vehicle through the Beidou positioning system, and combining a parking space position information database in an urban road parking management platform, identifying the current standard parking space number of the vehicle, and generating an analysis area with the parking space center as the circle center according to the center coordinate of the parking space and a preset analysis radius; and calling the accessed urban three-dimensional geographic information platform, and retrieving and analyzing all building boundary objects in the area. According to the invention, by introducing a building density intelligent prediction and self-adaptive fuzzy buffer mechanism and combining secondary verification of the vehicle body orientation and the inertial navigation trajectory, the positioning drift error of the Beidou in the building dense area is effectively offset, the parking space judgment fault tolerance is improved, the misjudgment and error control risk is significantly reduced, and the accuracy and stability of roadside parking management are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of roadside parking space management, and particularly relates to a roadside parking space management method and system based on the Beidou positioning system. Background Art

[0002] Managing roadside parking spaces based on the Beidou positioning system refers to a management method that uses the high-precision positioning ability provided by the Beidou satellite navigation system to monitor and intelligently dispatch the parking spaces on both sides of urban roads in real time. This method installs Beidou positioning terminals on vehicles or parking spaces to collect the position, parking status, and duration information of vehicles in real time, and transmits the data to the urban traffic management platform. The platform can combine electronic maps and parking rules to dynamically judge whether the vehicle is parked regularly, whether it occupies the space beyond the time limit, etc., thereby realizing intelligent supervision of parking behaviors. Compared with the traditional methods that rely on manual inspections or geomagnetic induction, the Beidou positioning system has the advantages of all-weather, high precision, wide coverage, etc., improving the utilization efficiency of urban parking resources and providing important support for intelligent transportation and refined urban management.

[0003] The prior art has the following deficiencies: In the existing roadside parking space management technology based on Beidou positioning, a static rasterized position information judgment mechanism is often used to determine whether a vehicle is parked regularly. However, in densely built-up areas, due to factors such as multipath effects and signal blockages, the Beidou positioning system is prone to sub-meter positioning errors. When the distances between multiple vehicles are close or in the critical area of the same parking space boundary, the Beidou positioning system may misidentify vehicles actually distributed in adjacent parking spaces as being in an "overlapping parking" state within the same parking unit. Such misjudgments not only trigger incorrect penalties for overlapping illegal parking, but may also cause non-violating vehicles to be forcibly locked, damage the vehicle body, or prevent normal departure when linked with automated devices (such as locking devices and lifting rod systems), thereby causing serious consequences such as property losses and even traffic jams.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to provide a roadside parking space management method and system based on the Beidou positioning system. By introducing intelligent prediction of building density based on three-dimensional urban semantic perception before the Beidou static grid judgment logic, and adaptively expanding the fuzzy buffer zone of the parking space grid and superimposing the secondary verification mechanism of vehicle body orientation and inertial navigation trajectory in high-density scenarios, the Beidou positioning system can dynamically offset the sub-meter positioning drift error caused by the building complex, achieve high-fault-tolerance determination for critical parking space boundary scenarios, thereby significantly reducing the misjudgment rate of "double parking" and equipment risks such as mis-triggering of automatic vehicle locking, and improving the accuracy, safety and user experience of roadside parking management in complex urban environments to solve the problems in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: A roadside parking space management method based on the Beidou positioning system, including the following steps: Collect the current position coordinates of the vehicle through the Beidou positioning system, and combine with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located. According to the center coordinates of the parking space, generate an analysis area centered on the center of the parking space with a preset analysis radius (such as 10 meters or 15 meters). Call the accessed urban three-dimensional geographic information platform (such as urban BIM data, urban LIDAR mapping data or GIS building vector map), retrieve all building boundary objects within the analysis area, and map each building covering the analysis area to the coordinate system where the center of the parking space is located according to the spatial relationship. Graphically fuse the extracted building boundary information to generate a building information map. Extract the key features reflecting the existence of high-density building complexes in the analysis area from the building information map through feature engineering techniques. After deeply analyzing the extracted features, input the analyzed features into a convolutional neural network that has been trained with a large amount of data in advance, and predict the building complex density in the analysis area through the convolutional neural network. When it is identified that the analysis area is a building-intensive area, expand a fuzzy edge buffer zone around each grid, and adaptively adjust the width of the fuzzy edge buffer zone according to the density of the building complex. When the vehicle positioning point is within the buffer zone, it is not immediately judged as straddling the position, but the secondary verification mechanism is triggered, and the vehicle body orientation recognition and inertial navigation displacement tracking methods are used to judge whether the vehicle is parked reasonably.

[0007] Preferably, the process of identifying the standard parking space number where the vehicle is located through the Beidou positioning system can be divided into the following key steps: First, use the high-precision Beidou positioning terminal installed on the vehicle to collect the longitude and latitude coordinates of the vehicle's current position in real time, and upload this position information to the urban road parking management platform; Second, the platform calls the pre-constructed parking space position information database, which records the spatial attributes such as the numbers, central coordinates, and boundary polygons of all standard parking spaces; Subsequently, the Beidou positioning system based on the spatial matching algorithm (such as point-in-polygon judgment, nearest neighbor matching, or vector space search) geometrically compares the vehicle's current position with all parking space boundaries to identify the standard parking space number where the vehicle is located; If the vehicle positioning point falls in the overlapping area of multiple parking space boundaries, an optimized determination is made according to the principle of the nearest distance or the directionality of the historical trajectory, so as to achieve the accurate identification of the actual parked parking space of the vehicle.

[0008] Preferably, mapping each building in the analysis area to the coordinate system where the parking space center is located according to the spatial relationship can be carried out according to the following steps: First, determine the geographical coordinates of the center point of the target parking space (such as WGS-84 longitude and latitude or XY coordinates in the urban projection coordinate system), and set this point as the origin of the local coordinate system; Second, perform coordinate transformation on the coordinate data of the retrieved building boundary objects one by one, and calculate the relative displacement of each building vertex relative to the center point of the parking space, that is, convert the original absolute coordinates into local coordinate representation based on the parking space center through translation transformation; Subsequently, if the urban data uses different projections or coordinate systems (such as LIDAR data is local coordinates, BIM is building internal coordinates), it is necessary to unify the coordinate reference system first and then perform the mapping; Finally, a set of building boundary polygon data aligned with the parking space center is formed, which is convenient for subsequent spatial relative position relationship analysis and graphic fusion operations.

[0009] Preferably, the process of graphically fusing the extracted building boundary information to generate a building information map can be carried out according to the following steps: First, based on the building boundary data that has been mapped to the parking space center coordinate system, draw the boundary polygons of each building on a unified two-dimensional plane, and use a vector graphics format (such as GeoJSON, Shapefile, or SVG path) to express its spatial contour. Secondly, use a graphics processing library (such as OpenCV, matplotlib, QGIS, or a custom rendering engine) to overlay and render all the building boundary layers and fuse them into a complete regional building information map, where different buildings can be given different attribute layer information (such as height, material, use, etc.) to enhance the layer semantics. Subsequently, to improve the analysis accuracy, auxiliary image information such as occlusion areas, boundary continuity, the number of enclosed spaces, and building density heat distribution in the graph can be further calculated, and this information can be embedded in the building information map in the form of a pixel matrix, partitioned vector grid, or raster encoding. The final output building information map will have visibility and computability, and can be used both as an input image for neural network recognition and for subsequent spatial structure feature extraction and analysis.

[0010] Preferably, key features reflecting the existence of high-density building clusters in the analysis area are extracted from the building information map through feature engineering techniques. Among them, the extracted features include the overlap ratio of adjacent building boundary lines on the projection plane and the concentration of the direction vectors of the edges of polygon buildings pointing to the parking space center. In the analysis area, after in-depth analysis of the extracted features, a boundary overlap rate index and a polygon concentration rate index are respectively generated, and the density of the building clusters in the analysis area is quantified through the boundary overlap rate index and the polygon concentration rate index. The analyzed boundary overlap rate index and polygon concentration rate index are used as feature vectors for high-dimensional spatial structure expression and input into a convolutional neural network that has been trained in advance with a large amount of data. Through in-depth analysis of the input feature vectors by the convolutional neural network, a density index score value that comprehensively reflects the building distribution density in the analysis area is finally output, and the density of the building clusters in the analysis area is intelligently predicted based on the density index score value.

[0011] Preferably, the density index score value generated when predicting the density of the building clusters in the analysis area through the convolutional neural network is compared and analyzed with a pre-set density threshold to identify building-dense areas. The specific identification steps are as follows: If the density index score value is greater than the pre-set density threshold, the analysis area is identified as a building-dense area; if the density index score value is less than or equal to the pre-set density threshold, the analysis area is identified as a normal building area.

[0012] Preferably, when the analysis area is identified as a building-dense area, a fuzzy edge buffer zone is extended around each grid, and the width of the fuzzy edge buffer zone is adaptively adjusted according to the density of the building complex. The specific steps are as follows: According to the deviation degree between the density index score value output by the convolutional neural network and the preset density threshold, calculate the building density over-standard ratio of the current analysis area. Then, based on the building density over-standard ratio, calculate the basic expansion radius of the fuzzy edge buffer zone. The calculation formula is: , where: is the density index score value output by the convolutional neural network. The larger the value, the denser the building distribution; is the building density threshold, which is used to determine whether to enter the high-density area; is the building density over-standard ratio, which reflects the degree of exceeding the building density relative to the building density threshold; is the basic buffer zone expansion radius, which represents the default expansion width (such as 0.5 meters) in the critical density state; is the sensitivity amplification factor (such as 1.5 - 3.0), which is used to control the intensity of the expansion response; is the non-linear amplification power exponent (it is recommended to take 1.2 - 2.0), which enhances the growth trend of the buffer width response in the high-density area; Traverse all parking space grid units in the analysis area and extend the fuzzy buffer zone to their boundaries to construct a new adaptive fuzzy expansion grid area. The geometric expansion rule of the adaptive fuzzy expansion grid area is expressed as: , where: represents the morphological dilation operation (which can be analogous to the concept of structural dilation convolution); is a two-dimensional dilation structural element with a radius of is the w-th parking space grid area; is the fuzzy grid area after the expansion of the w-th parking space grid; In the actual construction process, to prevent unnecessary intersection of adjacent grid buffer zones, a buffer zone boundary compression function is introduced to compress and adjust the overlapping area. The adjustment formula is: , where: is the center distance between adjacent grids; and are the buffer compression control coefficients (for example , ), which are used to avoid excessive expansion in the high-density area resulting in buffer overlap; is the actual edge buffer radius of the w-th parking space grid dynamically compressed according to the interference relationship of adjacent grids.

[0013] Preferably, within the analysis area, the specific steps for generating the boundary overlap rate index after deeply analyzing the overlap ratio of adjacent building boundary lines in the building information map on the projection plane are as follows: Within the analysis area, for all pairs of adjacent buildings (meeting the minimum distance between two buildings , being the preset adjacency threshold) perform boundary overlap detection to identify all pairs of overlapping boundary line segments , for each pair of overlapping boundary line segments, obtain their overlapping length, and combine their spatial distance to calculate the overlapping weight factor of the boundary line segment pair. The calculation expression is: where: is the length of the overlapping part of the boundary line segment; is the Euclidean distance between the centroids of buildings and ; is a small constant used to suppress numerical explosion caused by extremely small spatial distances; is the included angle between the direction vectors of the boundary line segment pair ; is the direction consistency factor, encouraging higher weights for parallel overlapping boundary line segments, ; is the overlapping weight factor of the boundary line segment pair; Perform non-linear fusion processing on the overlapping weight factors in all pairs of adjacent buildings that meet the conditions to calculate the boundary overlap rate index of the analysis area. The calculation expression is: where: is the boundary overlap rate index; is used to traverse all pairs of buildings that meet the adjacency conditions within the analysis area ; represents counting all pairs of boundary line segments that actually have boundary overlap in the building pair , and this set of boundary line segment pairs is denoted as ; ; is the structure trend amplification coefficient (with a value range of 0.01 - 0.05) used to control the non-linear growth rate; is a high-order non-linear compression function that stabilizes the output of the boundary overlap rate index at for non-linear normalization.

[0014] Preferably, within the analysis area, the specific steps for generating the polygon concentration rate index after deeply analyzing the concentration degree of the direction vectors from the edges of polygon buildings to the center of parking spaces in the building information map are as follows: In the local coordinate system with the center of the parking space as the origin, edge vectors are extracted for all polygonal buildings within the analysis area. For each boundary segment, its unit direction vector is constructed, and a unit vector pointing from the midpoint of this boundary segment to the center of the parking space is constructed. The degree of direction alignment (i.e., direction consistency) between the two is calculated using the following formula: , where: is the unit direction vector of the nth boundary segment; is the unit vector pointing from the midpoint of the nth boundary segment to the center of the parking space; represents the dot product operation of vector and vector ; represents the absolute value, used to uniformly handle positive and negative directions; represents the consistency weight value of the nth boundary segment with the direction of the parking space center, used to quantify whether this boundary segment points to the center of the parking space; After obtaining the consistency weight values of all boundary segments with the direction of the parking space center, based on the sparsity of the direction alignment weights, a normalized entropy evaluation is performed, and a polygon concentration rate index is constructed. The constructed expression is: , where: N is the total number of all boundary segments within the analysis area; represents the energy-normalized distribution of each direction consistency weight value, ; is used to normalize the entropy value to the interval; is the finally output polygon concentration rate index, with a range of , and a higher value indicates that the building directions are more concentrated on the center of the parking space.

[0015] A roadside parking space management system based on the Beidou positioning system, including a vehicle position recognition module, an analysis area generation module, a building boundary retrieval and mapping module, a building information map construction module, a building density prediction module, a buffer zone adaptive expansion module, and a secondary verification and determination module; The vehicle position recognition module collects the current position coordinates of the vehicle through the Beidou positioning system, and combines with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located; The analysis area generation module generates an analysis area centered on the center of the parking space with a preset analysis radius according to the center coordinates of the parking space; The building boundary retrieval and mapping module calls the accessed urban three-dimensional geographic information platform to retrieve all building boundary objects within the analysis area, and maps each building covering the analysis area to the coordinate system where the center of the parking space is located according to the spatial relationship; The building information map construction module graphically fuses the extracted building boundary information to generate a building information map; The building density prediction module extracts key features reflecting the existence of high-density building clusters in the analysis area from the building information map through feature engineering techniques. After deeply analyzing the extracted features, the analyzed features are input into a convolutional neural network that has been trained in advance with a large amount of data, and the convolutional neural network is used to predict the density of building clusters in the analysis area; The buffer zone adaptive expansion module, when it is recognized that the analysis area is a building-dense area, expands a fuzzy edge buffer zone around each grid, and adaptively adjusts the width of the fuzzy edge buffer zone according to the density of the building clusters; The secondary verification determination module, when the vehicle positioning point is within the buffer zone, does not immediately determine it as out-of-position, but triggers a secondary verification mechanism, and uses the methods of vehicle body orientation recognition and inertial navigation displacement tracking to determine whether the vehicle is parked reasonably.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention: The present invention introduces intelligent prediction of building density based on three-dimensional urban semantic perception before the Beidou static grid judgment logic, and adaptively expands the fuzzy buffer zone of the parking space grid and superimposes the secondary verification mechanism of vehicle body orientation and inertial navigation trajectory in high-density scenarios, enabling the Beidou positioning system to dynamically offset the sub-meter positioning drift error caused by building clusters, achieve high-fault-tolerance determination for critical parking space boundary scenarios, thereby significantly reducing the misjudgment rate of "stacked parking" and equipment risks such as mis-triggering of automatic vehicle locking, and improving the accuracy, safety and user experience of roadside parking management in complex urban environments. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings.

[0018] Figure 1 It is a method flow chart of a roadside parking space management method based on the Beidou positioning system of the present invention.

[0019] Figure 2 It is a module schematic diagram of a roadside parking space management system based on the Beidou positioning system of the present invention. Detailed Embodiments

[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0021] The present invention provides a roadside parking space management method based on the Beidou positioning system as Figure 1 shown, including the following steps: Collect the current position coordinates of the vehicle through the Beidou positioning system, and combine with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located. According to the center coordinates of the parking space, generate an analysis area centered on the center of the parking space with a preset analysis radius (such as 10 meters or 15 meters); The process of realizing the identification of the standard parking space number where the vehicle is located through the Beidou positioning system can be divided into the following key steps: First, use the Beidou high-precision positioning terminal installed on the vehicle to collect the longitude and latitude coordinates of the current position of the vehicle in real time, and upload this position information to the urban road parking management platform; Second, the platform calls the pre-constructed parking space position information database, which records the spatial attributes such as the numbers, center coordinates, and boundary polygons of all standard parking spaces; Subsequently, the Beidou positioning system, based on a spatial matching algorithm (such as point-in-polygon judgment, nearest neighbor matching, or vector space search), geometrically compares the current position of the vehicle with all parking space boundaries to identify the standard parking space number where the vehicle is located; If the vehicle positioning point falls in the overlapping area of multiple parking space boundaries, an optimized determination is made according to the principle of the nearest distance or the directionality of the historical trajectory, so as to achieve the accurate identification of the actual parked parking space of the vehicle.

[0022] This process provides basic positioning support for subsequent parking behavior determination and intelligent management.

[0023] This analysis area serves as a spatial window for subsequent spatial environment semantic analysis, ensuring that potential influencing factors such as high-rise building occlusion and dense street structures can be completely captured when the vehicle's surrounding environment is complex.

[0024] Call the accessed urban three-dimensional geographic information platform (such as urban BIM data, urban LIDAR surveying and mapping data, or GIS building vector maps), retrieve all building boundary objects within the analysis area, and map each building covering the analysis area to the coordinate system where the center of the parking space is located according to the spatial relationship; The retrieval of building boundary objects within the analysis area can be carried out according to the following steps: First, based on the identified target parking space center coordinates and the set analysis radius, the Beidou positioning system constructs a spatial query area and converts the coordinates of this area into a unified geospatial reference system (such as WGS-84 or the unified urban GIS projection coordinate system); Second, the Beidou positioning system calls the connected urban three-dimensional geographic information platform, such as the urban BIM data platform, the LIDAR mapping database, or the GIS building vector map interface, to perform a spatial range query operation (such as GeoBoundingBox query or GeoWithin spatial filtering), and retrieves all building boundary objects within the analysis area.

[0025] Map each building within the analysis area to the coordinate system where the parking space center is located according to the spatial relationship, which can be carried out in the following steps: First, determine the geographic coordinates of the target parking space center point (such as WGS-84 longitude and latitude or XY coordinates in the urban projection coordinate system), and set this point as the origin of the local coordinate system; Second, perform coordinate transformation on the coordinate data of the retrieved building boundary objects one by one, calculate the relative displacement of each building vertex relative to the parking space center point, that is, convert the original absolute coordinates into local coordinate representation based on the parking space center through translation transformation; Subsequently, if the urban data uses different projections or coordinate systems (such as LIDAR data being local coordinates and BIM being building internal coordinates), it is also necessary to unify the coordinate reference system first and then perform the mapping; Finally, a set of building boundary polygon data aligned with the parking space center is formed, which is convenient for subsequent analysis of spatial relative position relationships and graphic fusion operations.

[0026] This process ensures the consistency of spatial semantic expression and is the basis for accurately analyzing building density and occlusion relationships.

[0027] Graphically fuse the extracted building boundary information to generate a building information map; The process of graphically fusing the extracted building boundary information to generate a building information map can be carried out in the following steps: First, based on the building boundary data mapped to the parking space center coordinate system, the boundary polygons of each building are drawn on a unified two-dimensional plane and their spatial contours are expressed in vector graphic formats (such as GeoJSON, Shapefile, or SVG paths). Secondly, using a graphics processing library (such as OpenCV, matplotlib, QGIS, or a custom rendering engine), all building boundary layers are superimposed and rendered to be integrated into a complete regional building information map, where different buildings can be given different attribute layer information (such as height, material, use, etc.) to enhance the layer semantics. Subsequently, to improve the analysis accuracy, auxiliary image information such as occlusion areas, boundary continuity, the number of enclosed spaces, and building density heat distribution in the graph can be further calculated and these information are embedded in the building information map in the form of pixel matrices, partitioned vector grids, or raster encodings. The finally output building information map will have visibility and computability, which can be used both as an input image for neural network recognition and for subsequent spatial structure feature extraction and analysis.

[0028] Key features reflecting the existence of high-density building clusters in the analysis area are extracted from the building information map through feature engineering techniques. After in-depth analysis of the extracted features, the analyzed features are input into a convolutional neural network that has been trained in advance with a large amount of data, and the convolutional neural network is used to predict the building cluster density in the analysis area. Key features reflecting the existence of high-density building clusters in the analysis area are extracted from the building information map through feature engineering techniques. Among them, the extracted features include the overlapping ratio of adjacent building boundary lines on the projection plane and the concentration degree of the direction vectors of the polygon building edges pointing to the parking space center. In the analysis area, after in-depth analysis of the extracted features, a boundary overlap rate index and a polygon concentration rate index are respectively generated, and the building cluster density situation in the analysis area is quantified through the boundary overlap rate index and the polygon concentration rate index.

[0029] In the analysis area, if the overlapping ratio of the boundary lines of adjacent buildings in the building information map on the projection plane is higher, it usually indicates a higher building cluster density in this area. This is because in a high-density building environment, buildings often line up closely along the street, their walls are close to each other or even share structural boundaries, resulting in partial coincidence or very small spacing fitting of their contour lines on the two-dimensional projection plane. When there is a high line segment overlapping rate among the boundaries of multiple buildings in space, it reflects the physical structure characteristics of small gaps, crowded arrangements, and lack of open space between buildings. This geometric overlapping relationship can be used as a direct quantification index of spatial compression degree, which is a relatively stable and highly sensitive feature that is easy to extract at the graphic level. Therefore, the higher the overlapping ratio of the boundary lines, the more it can be regarded as the existence of a high-density building spatial structure in this area, which has important value for judging the "high occlusion area" or "high misjudgment risk area" of the city.

[0030] In the analysis area, the specific steps for generating the boundary overlap rate index after deeply analyzing the overlap ratio of adjacent building boundary lines in the building information map on the projection plane are as follows: In the analysis area, for all pairs of adjacent buildings (meeting the minimum distance between two buildings , being the preset adjacency threshold) conduct boundary overlap detection to identify all pairs of overlapping boundary line segments , for each pair of overlapping boundary line segments, obtain their overlapping length, and combine their spatial distance to calculate the overlapping weight factor of the boundary line segment pair. The calculation expression is: , where: is the length of the overlapping part of the boundary line segment; is the Euclidean distance of the centroids of buildings and ; is a small constant used to suppress numerical explosion caused by extremely small spatial distances; is the included angle of the direction vectors of the boundary line segment pair ; is the direction consistency factor, encouraging higher weights for parallel overlapping boundary line segments, ; is the overlapping weight factor of the boundary line segment pair; For each pair of overlapping boundary line segments, the acquisition of their overlapping length can be achieved by combining projection coincidence detection and line segment clipping calculation. Specifically, first, determine whether the two boundary line segments are collinear or approximately collinear in the two-dimensional plane. The determination condition can be that the included angle of the direction vectors is less than a set threshold (such as 5°); if satisfied, project the two line segments onto a one-dimensional axis (such as the main axis direction) in the same direction, and calculate the overlapping interval length by comparing the starting and ending projection intervals of the two line segments on this axis; this length is the spatial projection distance of the actual overlapping part. This method is not only applicable to strictly collinear situations but also tolerates "approximate overlapping" structures with slight offsets, has high robustness, and is convenient for embedding in a batch calculation framework in graphics processing or spatial analysis, and is an efficient means for identifying building fitting relationships.

[0031] This step comprehensively models the length, direction consistency, and spatial proximity of the overlapping line segments to generate an overlapping weight factor for characterizing the "local spatial fitting degree" , providing a physical representation of the boundary structure for subsequent intensive metrics.

[0032] Perform non-linear fusion processing on the overlapping weight factors in all pairs of adjacent buildings that meet the conditions to calculate the boundary overlap rate index of the analysis area. The calculation expression is: , where: is the boundary overlap rate index; used to traverse all pairs of buildings in the analysis area that meet the adjacency condition ; represents in the building pair count all boundary segment pairs that actually overlap in boundaries , and this set of boundary segment pairs is denoted as ; is the structural trend amplification factor (the value ranges from 0.01 to 0.05) used to control the non-linear growth rate; is a high-order non-linear compression function, which makes the output of the boundary overlap rate index stable at , and is used for non-linear normalization; This step globally weights and superimposes the local overlap relationships and uses a non-linear suppression function for normalization, so that the boundary overlap rate index not only reflects the building fitting degree, but also can produce a response amplification effect on extremely dense structures.

[0033] It can be seen from the boundary overlap rate index that in the analysis area, the larger the performance value of the boundary overlap rate index generated after in-depth analysis of the overlap ratio of the adjacent building boundary lines in the building information map on the projection plane, the higher the density of the building group in this analysis area, and vice versa, the lower the density of the building group in this analysis area. The reason is that in high-density building areas, buildings are often arranged compactly, and the walls are close to each other or even show spatial relationships such as shared boundaries and parallel adjacent edges, resulting in a large number of boundary segment overlaps or near overlaps in the two-dimensional projection map; the boundary overlap rate index accurately captures the structural characteristics of this "spatial fitting degree" by integrating multiple factors such as overlap length, building spacing, and boundary direction consistency. Therefore, the larger its value, the more crowded the building distribution, the smaller the spacing, and the more consistent the direction arrangement, which is a direct structural response to the high density of the building group; on the contrary, in areas where buildings are sparsely distributed and the boundary lines are far from each other, the overlap rate is significantly reduced, and the boundary overlap rate index naturally decreases, thus reflecting the low-density spatial characteristics of this area.

[0034] Within the analysis area, if the edges or normal vectors of the polygon buildings in the building information map are concentrated and point towards the center of the parking space, it indicates that these buildings are arranged in a certain degree of surrounding, converging or array pattern in space. This phenomenon often appears in densely built urban blocks or enclosed community environments. Because in high-density building clusters, limited space resources prompt buildings to be more inclined to enclose, border or arrange inwardly around certain core areas during planning, resulting in the edge directions of building outlines showing a convergence towards the center. By calculating the concentration of these direction vectors (such as using the average cosine similarity or vector concentration coefficient), this aggregation characteristic can be quantified. The higher the value, the more compact the spatial distribution of the buildings, the more continuous the boundaries, and the narrower the gaps, thus reflecting a higher building density in this area. Therefore, the edge direction vector concentration is a spatial structure index with practical semantic significance and can effectively identify the characteristics of local high-density building clusters.

[0035] Within the analysis area, the specific steps to generate the polygon concentration rate index after a deep analysis of the direction vector concentration of the edges of polygon buildings in the building information map pointing towards the center of the parking space are as follows: In the local coordinate system with the center of the parking space as the origin, extract the edge vectors of all polygon buildings within the analysis area. For each boundary segment, construct its unit direction vector and the unit vector pointing from the midpoint of this boundary segment to the center of the parking space. Calculate the degree of direction alignment (i.e., direction consistency) between the two using the following formula: , where: is the unit direction vector of the nth boundary segment; is the unit vector pointing from the midpoint of the nth boundary segment to the center of the parking space; represents the dot product operation of vector and vector ; represents the absolute value, used to uniformly handle positive and negative directions; represents the consistency weight value of the nth boundary segment with the direction of the center of the parking space, used to quantify whether this boundary segment points towards the center of the parking space; In this step, the consistency weight values of all boundary segments with the direction of the center of the parking space range from

[0036] to , where: N is the total number of all boundary line segments in the analysis area; represents the energy normalization distribution of the consistency weight value in each direction, ; used to normalize the entropy value to interval; is the polygon concentration rate index of the final output, and the range is , and the higher the value, the more concentrated the building direction is around the parking space center; Model the spatial direction aggregation effect through entropy, and strengthen the response to highly aligned boundaries through square normalization to eliminate the disturbance of a small number of extreme values. The polygon concentration rate index close to 1 indicates high concentration, which is suitable for identifying "enclosed" building layouts or high-density enclosed blocks.

[0037] It can be seen from the polygon concentration rate index that in the analysis area, the larger the performance value of the polygon concentration rate index generated after deeply analyzing the concentration of the direction vectors of the polygon building edges pointing to the parking space center in the building information map, the higher the density of the building group in the analysis area, and vice versa, the lower the density of the building group in the analysis area. The reason is that: this polygon concentration rate index measures the spatial aggregation trend by calculating the consistency degree between the direction vectors of the building edge line segments and the direction of the parking space center. The larger the value, the more the building edges in the area tend to surround, approach or point to the parking space center, reflecting an obvious "space congestion" or "enclosure compression" layout feature; this feature is common in dense blocks, enclosed communities or street-side continuous building groups, and belongs to the spatial form of high-density development. Therefore, the larger the polygon concentration rate index, the stronger the structural concentration and direction aggregation of the building in the limited space, which indirectly reflects a higher building group density.

[0038] Input the analyzed boundary overlap rate index and polygon concentration rate index as the feature vectors of the high-dimensional space structure expression into the convolutional neural network that has been trained in advance with a large amount of data. Through the convolutional neural network, deeply analyze the input feature vectors, and finally output a dense index score value that comprehensively reflects the building distribution density in the analysis area, and intelligently predict the building group density in the analysis area based on the dense index score value.

[0039] A convolutional neural network that has been pre-trained with a large amount of data in advance refers to the situation where, before formally deploying the intelligent recognition task of building density, developers use the supervised learning method to pre-train a specially designed convolutional neural network model (CNN, Convolutional Neural Network) based on the existing large-scale urban three-dimensional space data, building distribution samples, and known density labels, so that it has the ability to automatically identify boundary features and building distribution patterns. In this process, the input training data consists of the boundary overlap rate index and the polygon concentration rate index extracted from a large number of real urban areas. The boundary overlap rate index and the polygon concentration rate index are fed into the network in the form of a two-dimensional matrix or an image as structural feature vectors, and the output is the building density score or classification label (such as sparse, medium, dense) corresponding to this area. During the training process, the neural network extracts local spatial patterns and trends in the features through the convolutional layer, such as the overlapping cluster area of building boundaries, the change direction of contour concentration, etc., and then continuously adjusts the weights through multiple hidden layers to minimize the difference between the output score and the manually labeled density, thereby forming a set of generalizable building density pattern recognition models.

[0040] Specifically, in the training stage, the convolutional neural network learns how to extract spatial structure features highly relevant to the building density from the boundary overlap rate index and the polygon concentration rate index by receiving a large amount of labeled data. The boundary overlap rate index reflects the degree of edge overlap between buildings, and it can reveal whether the buildings show a spatial trend of "seamless splicing" or "adjacent layout"; while the polygon concentration rate index measures the geometric concentration degree of building boundaries towards the center of the parking space, reflecting whether the buildings are arranged in a centralized manner around a certain functional core. The neural network slides the convolutional operation through multiple groups of convolutional kernels in the input feature matrix to extract edge distribution patterns and spatial density patterns at different scales, and then performs non-linear transformation through an activation function (such as ReLU) to enhance the model's ability to model the relationship between local changes and the overall structure; subsequently, the pooling layer further reduces the dimension of the pattern, retains the most representative structural information, and reduces the model complexity and overfitting risk. After multiple forward propagations and backward error propagations, the network can gradually adjust its internal parameters to make it have the ability to efficiently identify the distribution density of building groups. After the training is completed, the model is "frozen", that is, the network structure and parameters are fixed and can be directly used for rapid prediction in practical applications.

[0041] In the application stage, the trained CNN model takes the boundary overlap rate index and polygon concentration rate index extracted from the target analysis area as input feature vectors and inputs them into the model for in-depth analysis. These feature vectors can be constructed into a two-dimensional spatial grid map, a heat map, or a vector stack map to retain the spatial correlation and geometric structure features of the area. The model will automatically discriminate the input data based on the pre-trained parameters and output a dense index score value. This score value is the comprehensive evaluation result of the density of the building complex in this area. Usually, the value range is between 0 and 1. The higher the value, the denser the building layout, and the more likely it is to cause Beidou positioning errors and cross-position misjudgments.

[0042] Compare and analyze the dense index score value generated when predicting the density of the building complex in the analysis area through a convolutional neural network with a pre-set density threshold to identify the building-dense area. The specific identification steps are as follows: If the dense index score value is greater than the pre-set density threshold, then identify this analysis area as a building-dense area; if the dense index score value is less than or equal to the pre-set density threshold, then identify this analysis area as a normal building area.

[0043] When it is identified that this analysis area is a building-dense area, expand a fuzzy edge buffer zone around each grid, and adaptively adjust the width of the fuzzy edge buffer zone according to the density of the building complex. When the vehicle positioning point is within the buffer zone, it is not immediately judged as cross-position, but instead triggers a secondary verification mechanism, and uses the vehicle body orientation recognition and inertial navigation displacement tracking methods to judge whether the vehicle is parked reasonably; When it is identified that this analysis area is a building-dense area, expand a fuzzy edge buffer zone around each grid, and adaptively adjust the width of the fuzzy edge buffer zone according to the density of the building complex. The specific steps are as follows: According to the deviation degree between the dense index score value output by the convolutional neural network and the pre-set density threshold, calculate the building density over-standard ratio of the current analysis area. Then, based on this building density over-standard ratio, calculate the basic expansion radius of the fuzzy edge buffer zone. The calculation formula is: , where: is the dense index score value output by the convolutional neural network. The larger the value, the denser the building distribution; is the building density threshold, which is used to judge whether to enter the high-density area; is the building density over-standard ratio, which reflects the degree of excess of the building density relative to the building density threshold; is the basic buffer zone expansion radius, which represents the default expansion width (such as 0.5 meters) in the critical density state; is the sensitivity amplification factor (such as 1.5 - 3.0), which is used to control the intensity of the expansion response; is the non-linear amplification power exponent (it is recommended to take 1.2 - 2.0), enhancing the growth trend of the buffer width response in the high-density area; This step realizes the non-linear adaptive control mechanism of "the higher the density, the wider the buffer" by converting the difference between the building density score value and the set density threshold into the spatial expansion intensity of the fuzzy buffer zone, thereby improving the robustness of the Beidou positioning system in the high-risk misjudgment area.

[0044] Traverse all parking space grid cells in the analysis area and expand the fuzzy buffer zone on their boundaries to construct a new adaptive fuzzy expansion grid area. The geometric expansion rule of the adaptive fuzzy expansion grid area is expressed as: , where: represents the morphological dilation operation (which can be analogous to the concept of structural dilation convolution); is the two-dimensional dilation structural element with a radius of (usually a circle or an eight-connected neighborhood); is the w-th parking space grid area; is the fuzzy grid area after the expansion of the w-th parking space grid; In the actual construction process, to prevent unnecessary intersections of adjacent grid buffer zones, a buffer zone boundary compression function is introduced to compress and adjust the overlapping area. The adjustment expression is: , where: is the central distance between adjacent grids; and are the buffer compression control coefficients (for example , ), used to avoid excessive expansion in the high-density area resulting in buffer overlap; is the actual edge buffer radius of the w-th parking space grid dynamically compressed according to the interference relationship of adjacent grids; This step spatially dilates each parking space grid area in a structured manner and adaptively adjusts the expansion radius according to the density relationship between adjacent parking spaces, not only retaining the error tolerance ability for high-risk areas but also avoiding misidentification interference between grids in space, enhancing the stability and discrimination accuracy of the spatial resolution of the Beidou positioning system.

[0045] When the vehicle positioning point is within the fuzzy edge buffer zone, the Beidou positioning system does not directly determine it as "out of position" or "illegal parking", but triggers a secondary verification mechanism to improve the accuracy and fault tolerance of judgment. First, the Beidou positioning system calls the sensor data of the IMU (Inertial Measurement Unit) installed in the vehicle, extracts the current body orientation angle (such as the yaw angle or the angle with the geographic north), and compares this orientation with the preset standard parking direction of the parking space. If the angle between the two is within a reasonable range (such as less than 15°), it is considered that the vehicle parking posture meets the specifications. Secondly, the Beidou positioning system synchronously calls the inertial navigation trajectory tracking unit to retrospectively analyze the continuous movement trajectory of the vehicle from entering the current area to the stationary position, and determines whether it presents a reasonable behavior pattern of "along the lane - towards the side - docking"; if the trajectory shows that the vehicle drives in smoothly at a low speed and does not deviate from the main direction of the target parking space, it further supports that it is parked legally. Through the dual judgment of the body orientation and trajectory characteristics, the Beidou positioning system can effectively eliminate misjudgments caused by positioning drift, ensuring more refined and intelligent judgment of parking behaviors at critical positions in densely built-up areas.

[0046] By introducing intelligent prediction of building density based on three-dimensional urban semantic perception before the Beidou static grid judgment logic, and adaptively expanding the fuzzy buffer zone of the parking space grid and superimposing the secondary verification mechanism of body orientation and inertial navigation trajectory in high-density scenarios, the Beidou positioning system can dynamically offset the sub-meter positioning drift error caused by the building complex, achieve high-fault-tolerance judgment of critical parking space boundary scenarios, thereby significantly reducing the misjudgment rate of "stacked parking" and equipment risks such as mis-triggering of automatic car locking, and improving the accuracy, safety and user experience of roadside parking management in complex urban environments.

[0047] The present invention provides a Figure 2 roadside parking space management system based on the Beidou positioning system as shown in The vehicle position recognition module collects the current position coordinates of the vehicle through the Beidou positioning system, and combines with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located; The analysis area generation module generates an analysis area centered on the center of the parking space with a preset analysis radius according to the center coordinates of the parking space; The building boundary retrieval and mapping module calls the accessed urban three-dimensional geographic information platform to retrieve all building boundary objects in the analysis area, and maps each building covering the analysis area to the coordinate system where the center of the parking space is located according to the spatial relationship; The building information map construction module graphically integrates the extracted building boundary information to generate a building information map; The building density prediction module extracts key features reflecting the existence of high-density building clusters in the analysis area from the building information map through feature engineering techniques. After deeply analyzing the extracted features, the analyzed features are input into a convolutional neural network that has been trained in advance with a large amount of data, and the convolutional neural network is used to predict the density of building clusters in the analysis area; The buffer zone adaptive expansion module, when it is recognized that the analysis area is a building-dense area, expands a fuzzy edge buffer zone around each grid, and adaptively adjusts the width of the fuzzy edge buffer zone according to the density of the building clusters; The secondary verification determination module, when the vehicle positioning point is within the buffer zone, does not immediately determine it as out-of-position, but triggers a secondary verification mechanism, and uses the vehicle body orientation recognition and inertial navigation displacement tracking methods to determine whether the vehicle is parked reasonably.

[0048] A roadside parking space management method based on the Beidou positioning system provided by an embodiment of the present invention is implemented through the above-mentioned roadside parking space management system based on the Beidou positioning system. The specific methods and processes of a roadside parking space management system based on the Beidou positioning system are detailed in the embodiments of the above-mentioned roadside parking space management method based on the Beidou positioning system, and will not be elaborated here.

[0049] The above formulas are all dimensionless and take their numerical calculations. The formula is a formula obtained by collecting a large amount of data for software simulation to obtain the closest real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0050] Only some exemplary embodiments of the present invention have been described by way of illustration above. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

[0051] As described above, only the specific embodiments of the present application are provided, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A roadside parking space management method based on the Beidou positioning system, characterized in that, It includes the following steps: Collect the current position coordinates of the vehicle through the Beidou positioning system, and combine with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located. According to the center coordinates of the parking space, generate an analysis area centered on the parking space center with a preset analysis radius; Call the accessed urban three-dimensional geographic information platform to retrieve all building boundary objects within the analysis area. For each building covering the analysis area, map it to the coordinate system where the parking space center is located according to the spatial relationship; Graphically fuse the extracted building boundary information to generate a building information map. Extract the key features reflecting the existence of high-density building clusters from the building information map through feature engineering techniques. After deeply analyzing the extracted features, input the analyzed features into a convolutional neural network that has been trained with a large amount of data in advance, and predict the building cluster density within the analysis area through the convolutional neural network; When it is identified that the analysis area is a building-intensive area, expand a fuzzy edge buffer zone around each grid, and adaptively adjust the width of the fuzzy edge buffer zone according to the density of the building cluster. When the vehicle positioning point is within the buffer zone, it is not immediately judged as out of position, but a secondary verification mechanism is triggered, and the vehicle body orientation recognition and inertial navigation displacement tracking methods are used to judge whether the vehicle is parked reasonably.

2. The roadside parking space management method based on the Beidou positioning system according to claim 1, wherein, Identify the standard parking space number where the vehicle is located through the Beidou positioning system, using the following steps: Use the Beidou high-precision positioning terminal installed on the vehicle to collect the longitude and latitude coordinates of the vehicle's current position in real time, and upload this position information to the urban road parking management platform; Call the pre-constructed parking space position information database; Based on the spatial matching algorithm, geometrically compare the vehicle's current position with all parking space boundaries to identify the standard parking space number where the vehicle is located; if the vehicle positioning point falls within the overlapping area of multiple parking space boundaries, optimize the determination according to the principle of the nearest distance to accurately identify the actual parked parking space of the vehicle.

3. The roadside parking space management method based on the Beidou positioning system according to claim 1, wherein, Map each building within the analysis area to the coordinate system where the parking space center is located according to the spatial relationship, which is achieved through the following steps: Determine the geographic coordinates of the center point of the target parking space, and set this point as the origin of the local coordinate system; Perform coordinate transformation on the coordinate data of the retrieved building boundary objects one by one, and calculate the relative displacement of each building vertex relative to the parking space center point, that is, convert the original absolute coordinates into local coordinate representations based on the parking space center through translation transformation; First unify the coordinate reference system, and then perform mapping to finally form a set of building boundary polygon data aligned with the parking space center.

4. A roadside parking space management method based on the Beidou positioning system according to claim 1, characterized in that, The process of graphically fusing the extracted building boundary information to generate a building information map is achieved through the following steps: Based on the building boundary data mapped to the parking space center coordinate system, draw the boundary polygon of each building on a unified two-dimensional plane, and express its spatial contour in a vector graphics format; Use the graphics processing library to overlay and render all building boundary layers to fuse them into a complete regional building information map, and assign different attribute layer information to different buildings to enhance the layer semantics; Calculate the auxiliary image information in the graph and embed the auxiliary image information into the building information graph in the form of a pixel matrix, a partition vector grid or a raster code.

5. A roadside parking space management method based on the Beidou positioning system according to claim 1, characterized in that, Extract the key features reflecting the existence of high-density building clusters in the analysis area from the building information graph through feature engineering techniques. Among them, the extracted features include the overlapping ratio of adjacent building boundary lines on the projection plane and the concentration degree of the direction vectors of the polygon building edges pointing to the center of the parking space. In the analysis area, after deeply analyzing the extracted features, a boundary overlap rate index and a polygon concentration rate index are respectively generated, and the building cluster density of the analysis area is quantified through the boundary overlap rate index and the polygon concentration rate index. Take the analyzed boundary overlap rate index and polygon concentration rate index as the feature vectors for high-dimensional space structure expression and input them into a convolutional neural network that has been trained in advance with a large amount of data. Through the convolutional neural network, deeply analyze the input feature vectors, and finally output a density index score value that comprehensively reflects the building distribution density of the analysis area. Based on the density index score value, make an intelligent prediction of the building cluster density in the analysis area.

6. The roadside parking space management method based on the Beidou positioning system according to claim 5, characterized in that, Compare and analyze the density index score value generated when predicting the building cluster density in the analysis area through the convolutional neural network with a pre-set density threshold to identify the building-dense area. The specific identification steps are as follows: If the density index score value is greater than the pre-set density threshold, identify the analysis area as a building-dense area; if the density index score value is less than or equal to the pre-set density threshold, identify the analysis area as a normal building area.

7. A roadside parking space management method based on the Beidou positioning system according to claim 6, characterized in that, When it is identified that the analysis area is a building-dense area, expand a fuzzy edge buffer zone around each grid, and adaptively adjust the width of the fuzzy edge buffer zone according to the density of the building cluster. The specific steps are as follows: Calculate the exceeding ratio of building density in the current analysis area based on the deviation degree between the dense index score value output by the convolutional neural network and the preset density threshold, and calculate the basic expansion radius of the fuzzy edge buffer zone based on the exceeding ratio of building density. The calculation formula is as follows: , where: is the dense index score value output by the convolutional neural network; is the building density threshold; is the exceeding ratio of building density, reflecting the exceeding degree of building density relative to the building density threshold; is the expansion radius of the basic buffer zone; is the sensitivity amplification factor, used to control the intensity of the expansion response; is the non-linear amplification power exponent, enhancing the growth trend of the buffer width response in the high-density area; Traverse all parking space grid cells within the analysis area and perform fuzzy buffer zone expansion on their boundaries to construct a new adaptive fuzzy expansion grid area. The geometric expansion rule of the adaptive fuzzy expansion grid area is expressed as: , where: represents the morphological dilation operation; is the two-dimensional dilation structuring element with as the radius; is the w-th parking space grid area; is the fuzzy grid area after expansion of the w-th parking space grid; Introduce a buffer zone boundary compression function to compress and adjust the overlapping area. The adjustment expression is: , where: is the center distance between adjacent grids; and are buffer compression control coefficients to avoid excessive expansion of high-density areas resulting in buffer overlap; The actual edge buffer radius of the w-th parking space grid dynamically compressed according to the interference relationship of adjacent grids.

8. A roadside parking space management method based on the Beidou positioning system according to claim 5, characterized in that, In the analysis area, the specific steps for generating the boundary overlap rate index after deeply analyzing the overlapping ratio of adjacent building boundary lines on the projection plane in the building information graph are as follows: Within the analysis area, for all pairs of adjacent buildings perform boundary overlap detection to identify all pairs of overlapping boundary line segments , for each pair of overlapping boundary line segments, obtain their overlapping length, and combine their spatial spacing to calculate the overlapping weight factor of the boundary line segment pair. The calculation expression is: , where: is the length of the overlapping part of the boundary line segment; is the Euclidean distance between the centroids of buildings and ; is a small constant; is the included angle of the direction vectors of the boundary line segment pair ; is the direction consistency factor, ; is the overlapping weight factor of the boundary line segment pair; The overlapping weight factors in all adjacent building pairs that meet the conditions are subjected to non-linear fusion processing to calculate the boundary overlap rate index of the analysis area. The calculation expression is as follows: , where: is the boundary overlap rate index; is used to traverse all building pairs that meet the adjacency conditions within the analysis area ; represents the boundary segment pairs that actually overlap at the boundary in the building pair ; ; is the structural trend amplification factor used to control the non-linear growth rate.

9. A roadside parking space management method based on the Beidou positioning system according to claim 5, characterized in that In the analysis area, the specific steps for generating the polygon concentration rate index after deeply analyzing the concentration degree of the direction vectors of the polygon building edges pointing to the center of the parking space in the building information graph are as follows: In the local coordinate system with the center of the parking space as the origin, edge vectors of all polygonal buildings within the analysis area are extracted. For each boundary segment, its unit direction vector is constructed, and a unit vector pointing from the midpoint of this boundary segment to the center of the parking space is also constructed. The following formula is used to calculate the degree of direction alignment between the two: , where: is the unit direction vector of the nth boundary segment; is the unit vector pointing from the midpoint of the nth boundary segment to the center of the parking space; represents the dot product operation of vector and vector ; represents the consistency weight value of the nth boundary segment with the direction of the center of the parking space; After obtaining the consistency weight values of all boundary segments and the direction of the parking space center, normalize the entropy evaluation based on the sparsity of the direction alignment weights, and construct a polygon concentration rate index. The constructed expression is: , where: N is the total number of all boundary segments in the analysis area; represents the energy normalization distribution of each direction consistency weight value, ; is the finally output polygon concentration rate index.

10. A roadside parking space management system based on the Beidou positioning system, which is used to implement the roadside parking space management method based on the Beidou positioning system described in any one of the above claims 1-9, and is characterized in that, It includes a vehicle position recognition module, an analysis area generation module, a building boundary retrieval and mapping module, a building information graph construction module, a building density prediction module, a buffer zone adaptive expansion module, and a secondary verification and determination module; The vehicle position recognition module collects the current position coordinates of the vehicle through the Beidou positioning system and combines with the parking space position information database in the urban road parking management platform to identify the standard parking space number where the vehicle is currently located. The analysis area generation module generates an analysis area centered on the parking space with a preset analysis radius according to the center coordinates of the parking space. The building boundary retrieval and mapping module calls the accessed urban three-dimensional geographic information platform to retrieve all building boundary objects in the analysis area. For each building covering the analysis area, map it to the coordinate system where the parking space center is located according to the spatial relationship. Building Information Map Construction Module, which graphically integrates the extracted building boundary information to generate a building information map; Building Density Prediction Module, which extracts key features reflecting the existence of high-density building clusters in the analysis area from the building information map through feature engineering techniques. After deeply analyzing the extracted features, the analyzed features are input into a convolutional neural network that has been trained in advance with a large amount of data, and the building cluster density in the analysis area is predicted through the convolutional neural network; Buffer Zone Adaptive Expansion Module, when it is recognized that the analysis area is a building-intensive area, a fuzzy edge buffer zone is expanded around each grid, and the width of the fuzzy edge buffer zone is adaptively adjusted according to the density of the building clusters; Secondary Verification Judgment Module, when the vehicle positioning point is within the buffer zone, it is not immediately judged as out of position, but a secondary verification mechanism is triggered, and the vehicle's reasonable parking is judged by means of vehicle body orientation recognition and inertial navigation displacement tracking.

Images