A rail station surrounding overpass passage demand evaluation method based on multi-source data

Abstract

The application discloses a kind of based on multi-source data's track site periphery crossing passage demand evaluation method, it is related to traffic technical field, the method fuses mobile phone signaling, land use and POI, present situation crossing facility and road network and so on multi-source heterogeneous data, and through data cleaning, track reconstruction and deep learning algorithm identification rail transit travel path, extract off-station walking track features, in combination with DBSCAN clustering identification crossing demand heat area and match determine crossing gap;Multi-dimensional evaluation index system is constructed, and the demand urgency of crossing gap is calculated after standardization and weight assignment, and the demand evaluation result is visualized, improves crossing demand identification precision and evaluation scientificity, provides quantitative decision support for urban crossing passage planning, effectively reduces planning cost, improves traffic operation efficiency.

Description

Technical Field

[0001] This application relates to the field of transportation technology, specifically to a method for assessing the demand for pedestrian crossings around rail stations based on multi-source data. Background Technology

[0002] Currently, the following technical solutions are mainly used for identifying pedestrian crossing demand at rail transit stations: (1) Station passenger flow identification technology based on computer vision: By installing cameras in subway stations, computer vision technology is used to identify passenger flow density, pedestrian behavior and traffic bottlenecks in the station. It is mainly used for optimizing facilities in the station, but it cannot cover pedestrian crossing demand outside the station. (2) Travel mode identification technology based on mobile phone signaling data: Existing technology uses mobile phone signaling data to identify rail transit travel modes. It judges whether it is rail transit travel by location matching, trajectory similarity and speed characteristics, but mainly focuses on OD identification and passenger flow prediction, and lacks refined analysis of pedestrian crossing behavior outside the station. (3) Passenger flow corridor identification method based on OD clustering: Passenger flow corridors are identified by considering factors such as distance from O point, distance from D point and path overlap through OD point clustering, but it does not specifically model pedestrian crossing demand, nor does it combine multi-source data for comprehensive evaluation. (4) Traditional manual survey method: Pedestrian crossing demand data are obtained through questionnaire surveys, manual observation and other methods. It has problems such as small sample size, short coverage time and high cost, and it is difficult to reflect the real dynamic demand.

[0003] The existing technologies mentioned above have the following problems: (1) Incomplete data coverage: Existing technologies mostly rely on a single data source (such as only mobile phone signaling or only AFC data), which cannot fully reflect the complex characteristics of cross-street demand around rail stations, especially the lack of identification of implicit demand such as "detour" and "crossing the road". (2) Lack of accurate spatial positioning: Traditional methods have low accuracy in identifying the specific walking trajectory and waiting behavior of users on both sides of the road outside the station, and cannot accurately locate the hot spots of cross-street demand, resulting in a lack of scientific basis for the selection of cross-street passages. (3) Single evaluation dimension: Existing technologies mainly focus on the size of passenger flow, without comprehensively considering multiple dimensions such as road grade, red line width, intensity of attraction points on the opposite side, population and job density, etc., and cannot fully assess the urgency of cross-street demand. (4) Poor real-time performance: Manual survey methods have long cycles and slow updates, which cannot adapt to the dynamic changes in the city and are difficult to support refined and real-time traffic planning decisions. (5) Insufficient algorithm accuracy: Existing trajectory matching algorithms have drift errors in complex urban environments, and the road network matching accuracy is not high, which affects the accuracy of backtracking the user's real walking path. Summary of the Invention

[0004] The purpose of this application is to provide a method for assessing the demand for pedestrian crossings around rail stations based on multi-source data, which solves the problems of incomplete data coverage, lack of accurate spatial positioning, single assessment dimension, poor real-time performance and / or insufficient algorithm accuracy.

[0005] This application is achieved through the following technical solution:

[0006] A method for assessing the demand for pedestrian crossings around rail stations based on multi-source data includes:

[0007] Collect multi-source heterogeneous data; the multi-source heterogeneous data includes mobile phone signaling data, land use and POI data points, existing municipal pedestrian crossing data, and road network data.

[0008] The mobile phone signaling data is cleaned and the trajectory is reconstructed to obtain continuous spatiotemporal trajectory data of pedestrians. Based on the continuous spatiotemporal trajectory data, a deep learning algorithm is used to identify rail transit travel and backtrack the path to and from the station.

[0009] Based on the entry and exit paths, and combined with the real location information of the rail station entrances and exits, the walking trajectory features of users on both sides of the road outside the station are extracted;

[0010] Based on the walking trajectory characteristics of users on both sides of the road outside the station, the heat areas of pedestrian crossing demand are identified, and the heat areas of pedestrian crossing demand are matched and analyzed with the entrances and exits of the rail station to determine the location of the pedestrian crossing gap.

[0011] A multi-dimensional evaluation index was constructed using land use and POI data, existing municipal pedestrian crossing data, and road network data. The multi-dimensional evaluation index was then standardized and its weights were determined to obtain the standardized multi-dimensional evaluation index and its corresponding weights.

[0012] Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, the demand urgency index corresponding to the location of the pedestrian gap is obtained, and the demand urgency index is used to assess the demand for pedestrian passages around the rail station and visualize the assessment results.

[0013] In one possible implementation, the mobile phone signaling data is cleaned and its trajectory reconstructed to obtain continuous spatiotemporal trajectory data of the pedestrian, including:

[0014] The raw mobile signaling data is subjected to quality checks, and abnormal data is removed to obtain preprocessed mobile signaling data.

[0015] The Kalman filter algorithm is used to dynamically correct the preprocessed mobile phone signaling data to obtain continuous spatiotemporal trajectory data of pedestrians.

[0016] In one possible implementation, based on the continuous spatiotemporal trajectory data, a deep learning algorithm is used to identify rail transit travel and trace routes to obtain entry and exit routes, including:

[0017] The study area is divided into grids to obtain multiple grid cells for analysis; the study area refers to the area where the demand for pedestrian crossings around rail stations is assessed.

[0018] For any given grid cell, feature parameters are extracted from the continuous spatiotemporal trajectory data corresponding to that grid cell; the feature parameters include one or more of the following: travel duration, travel distance, average speed, number of trajectory points, number of stop points, whether it passes through the track station buffer zone, and density of surrounding roads;

[0019] The feature parameters are analyzed using a pre-set deep learning algorithm to identify rail transit travel trajectories and extract the origin and destination information for each rail transit trip to obtain the travel mode.

[0020] Based on the aforementioned travel mode, a hidden Markov model is used for path backtracking to obtain the entry and exit routes.

[0021] In one possible implementation, based on the travel mode, a hidden Markov model is used for path backtracking to obtain the entry and exit paths, including:

[0022] Based on the aforementioned travel mode, each road segment in the road network is considered as a hidden state, and each observation point in the continuous spatiotemporal trajectory data is considered as an observation sequence. Hidden Markov Model construction, state transition probability calculation, observation probability calculation, Viterbi algorithm solution, and road network matching are performed sequentially to achieve path backtracking and obtain the entry and exit paths.

[0023] In one possible implementation, the multi-source heterogeneous data also includes rail transit AFC data;

[0024] After obtaining the travel mode, the process also includes: using rail transit AFC data to perform spatiotemporal matching on the travel mode to obtain a spatiotemporal matching result; the spatiotemporal matching result includes matching or non-matching.

[0025] Based on the premise that the spatiotemporal matching result is a mismatch, the accuracy of travel identification and route tracing is evaluated by comparing the difference between the identification result and the AFC data, and the evaluation result is obtained.

[0026] Based on the evaluation results, adjust the parameters of the deep learning algorithm or the threshold for road network matching.

[0027] In one possible implementation, based on the entry / exit path and combined with the actual location information of the rail station entrances and exits, the user's walking trajectory features on both sides of the road outside the station are extracted, including:

[0028] Based on the aforementioned entry and exit routes, each entry and exit route is segmented according to the road boundaries, and the walking trajectory segments on both sides of the road are identified. For walking trajectory segments that cross the road, they are marked as potential street crossing behaviors.

[0029] For pedestrian trajectory segments with potential street crossing behavior, analyze the stopping points in the pedestrian trajectory segments to identify users' waiting behavior at the road edge;

[0030] For any walking trajectory segment with waiting behavior, the user's detour behavior is identified by calculating the geometric features of the walking trajectory segment, and the detour point is obtained;

[0031] By analyzing the crossing angle and crossing position of the trajectory, the user's cross-road behavior is identified, and the cross-road point is obtained.

[0032] The detour points and the road crossing points are used together as the characteristics of the user's walking trajectory on both sides of the road outside the station.

[0033] In one possible implementation, based on the user's walking trajectory characteristics on both sides of the road outside the station, heat areas of pedestrian crossing demand are identified, including:

[0034] The detour points and road crossing points in the pedestrian trajectory features are used as sample points for pedestrian crossing demand. The DBSCAN spatial clustering algorithm is used to automatically identify the heat areas of pedestrian crossing demand.

[0035] In one possible implementation, the location of the pedestrian crossing gap is determined by matching the heat zone of the pedestrian crossing demand with the entrances and exits of the rail station, including:

[0036] The precise location coordinates of each entrance and exit of the station are obtained from the rail transit design drawings or on-site surveys; the precise location coordinates include the entrance / exit number, location coordinates, orientation, and width information;

[0037] Based on the precise location coordinates, the distance between each entrance / exit and each hotspot area is calculated. If the distance is less than the set matching radius, the entrance / exit is determined to match the hotspot area.

[0038] For any hotspot area that matches an entrance or exit, analyze the location of pedestrian crossing gaps that are missing, lack facilities, have insufficient facilities, or are poorly located.

[0039] Based on the density of hotspot areas, traffic flow at entrances and exits, and road grade, the identified pedestrian gaps are prioritized and their locations are output.

[0040] In one possible implementation, a multi-dimensional evaluation index is constructed using land use and POI point data, existing municipal pedestrian crossing data, and road network data. The multi-dimensional evaluation index is then standardized and its weights are determined to obtain the standardized multi-dimensional evaluation index and its corresponding weights, including:

[0041] Based on the land use and POI data, existing municipal pedestrian crossing data, and road network data, multiple evaluation indicators are constructed in the behavioral characteristic dimension, traffic flow characteristic dimension, land use dimension, road characteristic dimension, and facility characteristic dimension to obtain multi-dimensional evaluation indicators.

[0042] The multidimensional evaluation indicators are dedimensionalized to obtain standardized multidimensional evaluation indicators.

[0043] The weights of the standardized multi-dimensional evaluation indicators are obtained by using the analytic hierarchy process (AHP), thus obtaining the standardized multi-dimensional evaluation indicators and their corresponding weights.

[0044] In one possible implementation, based on the standardized multi-dimensional evaluation indicators and their corresponding weights, a demand urgency index corresponding to the location of the pedestrian gap is obtained. Then, based on the demand urgency index, a demand assessment of pedestrian crossings around the rail station is performed, and the assessment results are visualized, including:

[0045] Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, a weighted sum is performed to obtain the demand urgency index corresponding to the location of the pedestrian gap;

[0046] The urgency index corresponding to the location of the pedestrian crossing gap is matched with the preset demand level classification range to determine the demand level of the pedestrian crossings around the rail station.

[0047] For any given demand level, the locations of the pedestrian crossing gaps are arranged in descending order of demand urgency index to generate a priority assessment list.

[0048] The required levels and priority assessment list will be visualized and output.

[0049] Compared with the prior art, this application has the following advantages and beneficial effects:

[0050] This application discloses a method for assessing the demand for pedestrian crossings around rail transit stations based on multi-source data. This method integrates heterogeneous data from multiple sources, including mobile phone signaling, land use and POI, existing pedestrian crossing facilities, and road networks. It identifies rail transit travel paths through data cleaning, trajectory reconstruction, and deep learning algorithms, extracts pedestrian trajectory features outside stations, and uses DBSCAN clustering to identify pedestrian demand heatmaps and match them to determine pedestrian gaps. A multi-dimensional evaluation index system is constructed, and the urgency index of pedestrian gap demand is calculated through standardization and weighting. The demand assessment results are also visualized, improving the accuracy and scientific rigor of pedestrian demand identification and assessment. This provides quantitative decision support for urban pedestrian crossing planning, effectively reducing planning costs and improving traffic efficiency. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the exemplary embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0052] Figure 1 A flowchart illustrating a method for assessing the demand for pedestrian crossings around rail stations based on multi-source data, provided in this application embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this application are only for explaining this application and are not intended to limit this application.

[0054] like Figure 1 As shown in the figure, this application provides a method for assessing the demand for pedestrian crossings around rail stations based on multi-source data, including:

[0055] S101. Collect multi-source heterogeneous data; the multi-source heterogeneous data includes mobile phone signaling data, land use and POI data points, existing municipal pedestrian crossing data, and road network data.

[0056] S102. Perform data cleaning and trajectory reconstruction on the mobile phone signaling data to obtain continuous spatiotemporal trajectory data of pedestrians. Based on the continuous spatiotemporal trajectory data, use deep learning algorithms to identify rail transit travel and backtrack the path to and from the station.

[0057] S103. Based on the entry and exit paths, and combined with the real location information of the rail station entrances and exits, extract the walking trajectory features of the user on both sides of the road outside the station.

[0058] S104. Based on the walking trajectory characteristics of the users on both sides of the road outside the station, identify the heat areas of pedestrian crossing demand, and match and analyze the heat areas of pedestrian crossing demand with the entrances and exits of the rail station to determine the location of the pedestrian crossing gap.

[0059] S105. Using land use and POI data, existing municipal pedestrian crossing data and road network data, construct multi-dimensional evaluation indicators, and standardize and determine the weights of the multi-dimensional evaluation indicators to obtain standardized multi-dimensional evaluation indicators and their corresponding weights.

[0060] S106. Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, obtain the demand urgency index corresponding to the location of the pedestrian crossing gap, and conduct a demand assessment of the pedestrian crossing around the rail station and visualize the assessment results based on the demand urgency index.

[0061] In one possible implementation, multi-source heterogeneous data can be:

[0062] Mobile signaling data: Anonymized mobile signaling data obtained from mobile operators, including user identifiers, timestamps, base station location information, signal strength, etc. This type of data is characterized by wide coverage and strong temporal continuity, and can reflect the spatiotemporal activity patterns of urban residents. Data collection time is usually selected for a consecutive 7 days, including weekdays and weekends, to comprehensively reflect travel characteristics of different date types.

[0063] Rail transit AFC data: Data obtained from the automatic fare collection system of rail transit operating companies, including card number (anonymized), entry and exit time, station number, gate number, transaction amount, etc. This type of data can accurately reflect the origin and destination points and time distribution of rail transit passengers, and is an important basis for identifying rail transit trips.

[0064] Land use and POI point data, including land use categories: land use data and interest types, building area, population and job density, POI type and location coordinates, obtained from geographic information systems or commercial map service providers. This type of data is used to analyze the land use characteristics and attraction point distribution around the site, providing socio-economic background information for pedestrian crossing demand assessment.

[0065] Current status data on municipal pedestrian crossings: This data, obtained from municipal management departments, includes the location coordinates, type, width, and usage status of facilities such as overpasses, underpasses, zebra crossings, and traffic lights. This type of data is used to identify the distribution and capacity of existing pedestrian crossings, providing a basis for gap analysis.

[0066] Road network data: Road network data obtained from transportation planning departments or map service providers, including information such as road grade, right-of-way width, number of lanes, speed limits, intersection locations, and sidewalk widths. This type of data is used to analyze the physical characteristics and traffic conditions of roads, providing spatial context for assessing pedestrian crossing demand.

[0067] In one possible implementation, the mobile phone signaling data is cleaned and its trajectory reconstructed to obtain continuous spatiotemporal trajectory data of the pedestrian, including:

[0068] S102.1 Perform quality checks on the original mobile phone signaling data and remove abnormal data to obtain preprocessed mobile phone signaling data;

[0069] S102.2. The Kalman filter algorithm is used to dynamically correct the preprocessed mobile phone signaling data to obtain continuous spatiotemporal trajectory data of pedestrians.

[0070] For example, since mobile phone signaling data comes from base station handover records, there are problems such as limited location accuracy and uneven time intervals, requiring specialized data cleaning and trajectory reconstruction processing.

[0071] First, the raw signaling data undergoes a quality check, removing obviously abnormal data records, such as those with excessively large location jumps, incorrect timestamps, or abnormal signal strength. Then, a Kalman filter algorithm is used to dynamically correct the signaling data, effectively reducing location drift caused by base station handover. By establishing a pedestrian motion state model and combining observation data with prior knowledge, the Kalman filter estimates the pedestrian's optimal position in real time, significantly improving the continuity and accuracy of the trajectory.

[0072] In one possible implementation, based on the continuous spatiotemporal trajectory data, a deep learning algorithm is used to identify rail transit travel and trace routes to obtain entry and exit routes, including:

[0073] S102.3. Divide the study area into grids to obtain multiple grid cells for analysis; wherein, the study area refers to the area where the demand assessment of pedestrian crossings around rail stations is carried out;

[0074] S102.4 For any given grid cell, extract feature parameters from the continuous spatiotemporal trajectory data corresponding to that grid cell; the feature parameters include one or more of the following: travel duration, travel distance, average speed, number of trajectory points, number of stop points, whether it passes through the track station buffer zone, and density of surrounding roads;

[0075] S102.5. The feature parameters are analyzed using a pre-set deep learning algorithm to identify the rail transit travel trajectory and extract the origin and destination information of each rail transit trip to obtain the travel mode.

[0076] S102.6. Based on the aforementioned travel mode, a hidden Markov model is used to perform path backtracking to obtain the entry and exit paths.

[0077] For example, after trajectory reconstruction is completed, the study area is divided into regular traffic analysis zones (TAZs), typically using 100m x 100m grid cells. All multi-source data are uniformly mapped to this standardized spatial framework, providing a consistent spatial benchmark for subsequent fusion analysis. This standardization process not only improves data comparability but also facilitates spatial statistical and visualization analysis.

[0078] Based on the continuous spatiotemporal trajectory data generated in step S1, it is first necessary to identify which trajectories belong to rail transit travel. This application uses machine learning algorithms for travel mode identification, specifically including the following steps:

[0079] Feature extraction: Multiple feature parameters are extracted from each trajectory, including travel duration, travel distance, average speed, number of trajectory points, number of stops, whether it passes through the rail station buffer zone, and surrounding road density. These features can effectively distinguish rail transit travel from other modes of travel (such as walking, cycling, public transportation, and cars).

[0080] Model Training: Machine learning algorithms such as Random Forest or Support Vector Machine are used to train the travel mode recognition model using labeled sample data. The Random Forest algorithm, by constructing multiple decision trees and voting, has strong generalization ability and anti-overfitting properties; the Support Vector Machine algorithm, by finding the optimal classification hyperplane, is suitable for classification problems in high-dimensional feature spaces.

[0081] Trip identification: The trained model is applied to unlabeled trajectory data to automatically identify rail transit trip trajectories and extract the origin and destination information for each rail transit trip. The identification results include key information such as the origin location, destination location, trip time, and trip duration.

[0082] In one possible implementation, based on the travel mode, a hidden Markov model is used for path backtracking to obtain the entry and exit paths, including:

[0083] Based on the aforementioned travel mode, each road segment in the road network is considered as a hidden state, and each observation point in the continuous spatiotemporal trajectory data is considered as an observation sequence. Hidden Markov Model construction, state transition probability calculation, observation probability calculation, Viterbi algorithm solution, and road network matching are performed sequentially to achieve path backtracking and obtain the entry and exit paths.

[0084] For example, after identifying rail transit travel, it is necessary to further trace the user's entry and exit routes around the rail station in order to analyze their street crossing behavior.

[0085] This application uses a Hidden Markov Model (HMM) for road network matching and path backtracking. The specific process is as follows:

[0086] Hidden Markov Model Construction: Each road segment in the road network is considered a hidden state, and each observation point in the signaling trajectory is considered an observation sequence. The model includes a state transition probability matrix and an observation probability matrix, which respectively describe the transition rules between road segments and the matching relationship between observation points and road segments.

[0087] State transition probability calculation: Based on the road network topology and pedestrian walking habits, the transition probability between adjacent road segments is calculated. Typically, pedestrians tend to choose the shortest path or a familiar path; therefore, the transition probability is related to the topological distance between road segments and the frequency of historical path selection. Let the state of each road segment in the road network be... , The state transition function is then defined as:

[0088] ;

[0089] in, This represents the hidden state corresponding to time t; Indicates from road segment Transfer to The probability of;

[0090] Furthermore, it can be determined based on topological distance and historical path frequency:

[0091] ;

[0092] in, The topological distance between road segments; Select frequency for historical paths; Here, e is the attenuation distance coefficient, and e is the natural constant.

[0093] Observation probability calculation: The matching probability between each observation point and each road segment is calculated, primarily considering the geometric distance from the observation point to the road segment. The closer the road segment, the higher the matching probability. Factors such as road grade and sidewalk conditions are also considered to improve matching accuracy. With candidate road sections Matching probability:

[0094] ;

[0095] Represented using a Gaussian distribution:

[0096] ;

[0097] in: This represents the vertical distance from the observation point to the candidate road segment; The standard deviation of the positioning error. Pi is the mathematical constant of a circle.

[0098] Viterbi Algorithm Solution: The Viterbi algorithm is used to find the optimal path, i.e., to find the hidden state sequence most likely to generate the observation sequence. This algorithm uses dynamic programming to efficiently find the global optimum, avoiding the computational complexity of exhaustive search. The optimal path satisfies the following conditions:

[0099] ;

[0100] The recursive process is as follows:

[0101] ;

[0102] in: For a moment to state The maximum probability; For observation sequence; This is the hidden state sequence. The optimal path sequence is ultimately obtained through backtracking.

[0103] In one possible implementation, the multi-source heterogeneous data also includes rail transit AFC data;

[0104] After obtaining the travel mode, the process also includes: using rail transit AFC data to perform spatiotemporal matching on the travel mode to obtain a spatiotemporal matching result; the spatiotemporal matching result includes matching or non-matching.

[0105] Based on the premise that the spatiotemporal matching result is a mismatch, the accuracy of travel identification and route tracing is evaluated by comparing the difference between the identification result and the AFC data, and the evaluation result is obtained.

[0106] Based on the evaluation results, adjust the parameters of the deep learning algorithm or the threshold for road network matching.

[0107] For example, in order to improve the accuracy of travel identification and route tracing, this application introduces AFC data for verification and correction:

[0108] Spatiotemporal matching verification: Identified rail transit trips are spatiotemporally matched with AFC data to check if the trip time matches the card swipe time and if the trip origin and destination are consistent with the card swipe station. Trips with successful matches have high confidence; trips with failed matches require further inspection and correction.

[0109] Accuracy Assessment and Feedback: The identified rail transit travel results are spatiotemporally matched and verified with AFC data. A successful match is determined when the difference between the identified travel time and the AFC recorded time is less than a preset time threshold Δt, and the distance between the identified entry / exit location and the corresponding AFC station is less than a preset spatial threshold Δd. The travel identification accuracy is calculated based on the number of successfully matched samples and the total number of samples. Simultaneously, using the entry / exit location corresponding to the AFC record as a benchmark, the deviation between the path backtracking result and the actual entry / exit location is calculated, and the path matching accuracy is also calculated. Based on the comprehensive accuracy evaluation results, the machine learning model parameters, as well as the matching radius R, observation probability standard deviation σ, and matching probability threshold θ in the Hidden Markov Model, are iteratively adjusted. Optimization stops when the comprehensive accuracy reaches the preset requirements, thus forming a closed-loop optimization mechanism based on AFC data feedback to continuously improve the accuracy of travel identification and path backtracking.

[0110] In one possible implementation, based on the entry / exit path and combined with the actual location information of the rail station entrances and exits, the user's walking trajectory features on both sides of the road outside the station are extracted, including:

[0111] S103.1 Based on the aforementioned entry and exit routes, each entry and exit route is segmented according to the road boundaries, and the walking trajectory segments on both sides of the road are identified. For walking trajectory segments that cross the road, they are marked as potential street crossing behaviors.

[0112] S103.2 For pedestrian trajectory segments with potential street crossing behavior, analyze the stopping points in the pedestrian trajectory segments and identify users' waiting behavior at the road edge;

[0113] S103.3 For any walking trajectory segment with waiting behavior, the user's detour behavior is identified by calculating the geometric features of the walking trajectory segment, and the detour point is obtained;

[0114] S103.4 By analyzing the crossing angle and crossing position of the trajectory, identify the user's cross-road behavior and obtain the cross-road point;

[0115] S103.5. The detour points and the road crossing points are used together as the user's walking trajectory features on both sides of the road outside the station.

[0116] For example, based on the backtracked entry and exit paths, combined with the actual location information of the rail station entrances and exits, the user's walking trajectory features on both sides of the road outside the station are extracted:

[0117] Trajectory segmentation: Each entry / exit path is segmented according to road boundaries, and pedestrian trajectory segments on both sides of the road are identified. Trajectory segments that cross roads are marked as potential street crossing behaviors.

[0118] Waiting behavior analysis: Analyze stopping points in the trajectory to identify user waiting behavior at road edges. Points with longer waiting times may indicate that users are looking for opportunities to cross the street or waiting for traffic lights.

[0119] Detour behavior recognition: By calculating the geometric features of the trajectory, detour behavior is identified. Detour behavior is usually manifested as the trajectory deviating from a straight path, forming a large curvature or angle change. This type of behavior often indicates that the user chooses to detour to avoid crossing the road, reflecting a hidden need for crossing the street.

[0120] Pedestrian crossing behavior recognition: By analyzing the crossing angle and location of the trajectory, this feature identifies the user's jaywalking behavior. Jaywalking behavior typically involves the trajectory directly crossing the road at an angle close to 90 degrees, and not at the location of existing pedestrian crossing facilities.

[0121] In one possible implementation, based on the user's walking trajectory characteristics on both sides of the road outside the station, heat areas of pedestrian crossing demand are identified, including:

[0122] S104.1. Using the detour points and road crossing points in the pedestrian trajectory features as sample points for street crossing demand, the DBSCAN spatial clustering algorithm is used to automatically identify the heat areas of street crossing demand.

[0123] For example, the identified detour points and road crossing points are used as sample points for pedestrian crossing demand, and the DBSCAN spatial clustering algorithm is used to automatically identify heat maps of pedestrian crossing demand:

[0124] DBSCAN Algorithm Principle: DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a density-based spatial clustering algorithm capable of discovering clusters of arbitrary shapes and effectively handling noisy points. The core idea of ​​the algorithm is: if a point's neighborhood contains a sufficient number of points, then that point belongs to a cluster; otherwise, the point is considered noise.

[0125] Parameter settings: Set two key parameters: neighborhood radius ε and minimum number of points MinPts. Neighborhood radius ε determines the granularity of clustering, and is usually set to 50-100 meters based on the spatial clustering characteristics of street crossing requirements; minimum number of points MinPts determines the minimum size of the cluster, and is usually set to 10-20 points.

[0126] Clustering process: The algorithm starts with any unvisited point and calculates the number of points in its ε-neighborhood. If the number of points is greater than or equal to MinPts, a new cluster is created, and the point and all its neighbors are added to the cluster. Then, the above process is repeated for each point in the cluster until it can no longer be expanded. Points with a number of points less than MinPts are marked as noise points.

[0127] Hotspot area generation: Multiple pedestrian crossing demand hotspot areas are obtained through DBSCAN clustering. Each hotspot area contains a spatially clustered group of pedestrian crossing demand points, reflecting a strong demand for crossings in that area. The size and density of hotspot areas can serve as preliminary indicators of demand intensity.

[0128] In one possible implementation, the location of the pedestrian crossing gap is determined by matching the heat zone of the pedestrian crossing demand with the entrances and exits of the rail station, including:

[0129] S104.2 Obtain the precise location coordinates of each entrance and exit of the station from the rail transit design drawings or on-site survey; the precise location coordinates include entrance / exit number, location coordinates, orientation and width information;

[0130] S104.3. Based on the precise location coordinates, calculate the distance between each entrance / exit and each hotspot area. If the distance is less than the set matching radius, then determine that the entrance / exit matches the hotspot area.

[0131] S104.4 For any hotspot area that matches an entrance or exit, analyze the location of pedestrian crossing gaps that are missing, lack facilities, have insufficient facilities, or are inappropriately located.

[0132] S104.5. Based on the density of hotspot areas, the traffic flow at entrances and exits, and the road grade, prioritize the identified pedestrian gaps and output the sorted pedestrian gap locations.

[0133] For example, after identifying areas with high pedestrian crossing demand, it is necessary to perform a matching analysis with the entrances and exits of rail stations to determine the specific locations of pedestrian crossing gaps:

[0134] Location of entrances and exits: Obtain the precise location coordinates of each entrance and exit of the station from rail transit design drawings or on-site surveys, including entrance / exit number, location coordinates, orientation, width and other information.

[0135] Matching algorithm: Calculate the distance between each entrance / exit and each hotspot area. If the distance is less than the set matching radius (usually 150 meters), the entrance / exit is considered to match the hotspot area. The matching degree can be calculated by weighting the distance; the closer the distance, the higher the matching degree.

[0136] Through matching analysis, the types of pedestrian crossing gaps are identified. The process is then layer by layer, following the order of "whether there is an entrance / exit → whether there are pedestrian crossing facilities → whether the facilities meet the standards → whether the facility location is suitable." Only one type of gap will be identified for the same area. The overall process is as follows:

[0137] Step 1: Determine if there are any track entrances or exits within 150m of the hotspot → Identify gaps or missing entrances / exits.

[0138] Step 2: Assuming an entrance / exit exists, determine if there are compliant pedestrian crossing facilities within 80 meters of the hotspot → Identify facility deficiency gaps.

[0139] Step 3: Assuming the existence of pedestrian crossing facilities, determine whether the peak-hour capacity of the facilities meets the pedestrian crossing demand → Identify facility insufficiency gaps.

[0140] Step 4: Assuming facilities meet the requirements, determine whether the spatial location of the facilities matches the core hotspot area → Identify gaps caused by inappropriate location.

[0141] After accurately identifying the four types of pedestrian crossing gaps, a weighted scoring model is constructed based on three core influencing factors: density of pedestrian crossing hotspots, peak traffic flow at rail entrances and exits, and road grade. This model is combined with the inherent safety risk differences of different gap types to prioritize pedestrian crossing gaps, distinguish the urgency of each gap's renovation, and finally output the coordinates and detailed information of the sorted pedestrian crossing gap locations according to their urgency. This ranking process involved expert-assigned weights for each indicator. Road grade, which relates to vehicle speed and pedestrian safety risks, carries a weight of 0.45 and is divided into four scoring tiers: expressway, arterial road, secondary arterial road, and local road. Peak traffic flow at entrances and exits reflects the overall pedestrian volume of rail connections, with a weight of 0.35, and is divided into four scoring levels based on hourly passenger flow. Density in hotspot areas reflects the intensity of local instantaneous pedestrian crossing demand, with a weight of 0.20, and is divided into scoring intervals based on hourly pedestrian density. Additionally, a base correction score is added based on the risk level of the four types of gaps: 5 points for missing entrances / exits, 4 points for missing facilities, 2 points for insufficient facilities, and 1 point for inappropriate location. Finally, a unified formula is used to calculate the comprehensive gap score: Comprehensive Gap Score. The score is calculated as follows: Road grade score × 0.45 + Peak traffic flow score at entrances / exits × 0.35 + Hotspot area density score × 0.20 + Gap type base score. All gaps are sorted in descending order of their comprehensive scores. If scores are the same, they are sorted by road grade, peak traffic flow at entrances / exits, and hotspot area density in that order. After sorting, a standardized gap list is output, including the sorting number, gap coordinates, gap type, corresponding entrance / exit information, comprehensive score, and rectification suggestions. This process can directly reuse all the measured and spatial data from the gap identification stage, forming a complete technical link from gap identification, type determination, priority sorting to result output, providing intuitive data support and decision-making basis for subsequent planning and renovation of pedestrian crossing facilities.

[0142] In one possible implementation, a multi-dimensional evaluation index is constructed using land use and POI point data, existing municipal pedestrian crossing data, and road network data. The multi-dimensional evaluation index is then standardized and its weights are determined to obtain the standardized multi-dimensional evaluation index and its corresponding weights, including:

[0143] S105.1 Based on the land use and POI data, existing municipal pedestrian crossing data and road network data, construct multiple evaluation indicators in the behavioral characteristic dimension, traffic flow characteristic dimension, land use dimension, road characteristic dimension and facility characteristic dimension respectively to obtain multi-dimensional evaluation indicators.

[0144] S105.2. Perform dimensionless processing on the multi-dimensional evaluation indicators to obtain standardized multi-dimensional evaluation indicators;

[0145] S105.3. Use the analytic hierarchy process (AHP) to obtain the index weights corresponding to the standardized multi-dimensional evaluation indicators, thus obtaining the standardized multi-dimensional evaluation indicators and their corresponding weights.

[0146] For example, in order to comprehensively and scientifically assess the intensity of pedestrian crossing demand, this patent constructs a comprehensive evaluation index system comprising five dimensions:

[0147] (1) Behavioral characteristics dimension;

[0148] This dimension relies on the original pedestrian crossing trajectory data across the entire area, directly reflecting the actual pedestrian crossing behavior and spontaneous crossing demands, and is the most core and direct indicator for demand assessment.

[0149] ①Crossing trajectory density: Based on the trajectory data after grid division, the number of pedestrian crossing trajectories that are illegal and compliant within a unit grid area is counted, and the spatial clustering degree of the trajectories is calculated;

[0150] ② Detour ratio: The optimal direct crossing route mileage and the actual detour route mileage of pedestrians are calculated separately, and the ratio of the number of detour routes to the total number of crossing routes is calculated to quantify the implicit crossing demand.

[0151] ③ Frequency of crossing the road: Using morning and evening rush hours as the statistical period, the number of times pedestrians cross the road without facilities within a unit of time is counted for each road segment, which directly reflects the urgency of the crossing gap.

[0152] (2) Traffic flow characteristics (extracted from passenger flow monitoring data at the entrances and exits of rail stations);

[0153] This dimension relies on time-sharing passenger flow data collected by automatic passenger flow monitoring equipment at stations to characterize the spatiotemporal distribution characteristics of rail transit passenger flow and pedestrian crossing demand.

[0154] ① Peak hour traffic flow at the station: Extract the total passenger flow entering and exiting the station at all entrances and exits during the morning and evening peak hours to reflect the overall pedestrian flow base in the area;

[0155] ② Inbound / Outbound Flow Ratio: Statistically analyze the inbound and outbound passenger flow during peak hours, calculate the ratio between the two, and determine the one-way / two-way distribution characteristics of pedestrian flow across the street.

[0156] (3) Land use dimension (extracted from urban GIS land use vector data and population and job census data);

[0157] This dimension relies on land use and population employment spatial data in the 500m buffer zone around the station to analyze the fundamental background of the demand for crossing the street.

[0158] ①Population density: Extract the distribution data of the permanent residents within the buffer zone, calculate the number of permanent residents per unit area, and reflect the normalized street crossing demand brought about by the residents;

[0159] ②Job density: Extract spatial distribution data of job positions in the area, count the number of job positions per unit area, and reflect the demand for crossing the street during commuting hours;

[0160] ③ Land use mix: Identify the proportion of commercial, residential, office, and transportation land within the buffer zone, calculate the land use mix balance, and determine the degree of diversity in pedestrian crossing demand.

[0161] (4) Road feature dimension (extracted from as-built drawings and on-site road survey data);

[0162] This dimension relies on actual road measurement data to characterize the constraints and risks of objective road traffic conditions on pedestrian crossings, while also reusing the core road indicators prioritized in the previous section.

[0163] ① Road Classification: Based on urban road design specifications, roads are classified into four levels: expressways, arterial roads, secondary arterial roads, and local roads, matching the previous sorting and classification standards;

[0164] ② Red line width: Directly read the road red line width parameter from the road survey drawing to measure the time and distance cost of a single pedestrian crossing;

[0165] ③Number of lanes: On-site survey and statistics of the number of two-way lanes for motor vehicles on the road section to quantify the traffic pressure on motor vehicles and the safety risks for pedestrians crossing the street.

[0166] (5) Facility characteristics dimension (based on the current pedestrian crossing facility survey and spatial vector data extraction);

[0167] This dimension relies on on-site surveys of pedestrian crossing space and operational data to measure the gap between the current supply of facilities and the demand for pedestrian crossings.

[0168] ① Density of attraction points on the opposite side: Count the number of pedestrian attraction facilities such as commercial, residential and office buildings on the opposite side of the road, calculate the density of attraction points per unit area, and determine the necessity for pedestrians to cross the road;

[0169] ② Current number of pedestrian crossing facilities: Statistics on the actual number of regular pedestrian crossing facilities such as zebra crossings, overpasses, and underpasses within an 80m matching radius of hot spots;

[0170] Facility service level: Based on the facility's capacity, peak queue length, and signal timing data, the quality of the current pedestrian crossing facilities' operation and service is evaluated in a graded manner.

[0171] Indicator Standardization: Due to the different dimensions and value ranges of the various indicators, standardization is required to convert all indicators to a unified [0,1] interval. The range standardization method is used, which involves subtracting the minimum value from the indicator value and then dividing by the difference between the maximum and minimum values.

[0172] Weight Determination: The Analytic Hierarchy Process (AHP) was used to determine the weights of each dimension and indicator. First, a judgment matrix was constructed, and experts were invited to conduct pairwise comparisons of each dimension and indicator to determine their relative importance. Then, the weight vector was calculated, and a consistency check was performed to ensure the rationality of the judgments. Based on expert opinions and practical experience, the recommended dimension weights are: behavioral characteristics 30%, traffic flow characteristics 25%, land use 20%, road characteristics 15%, and infrastructure characteristics 10%.

[0173] In one possible implementation, based on the standardized multi-dimensional evaluation indicators and their corresponding weights, a demand urgency index corresponding to the location of the pedestrian gap is obtained. Then, based on the demand urgency index, a demand assessment of pedestrian crossings around the rail station is performed, and the assessment results are visualized, including:

[0174] S106.1. Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, perform weighted summation to obtain the demand urgency index corresponding to the location of the pedestrian gap;

[0175] S106.2 Match the demand urgency index corresponding to the location of the pedestrian gap with the preset demand level classification range to determine the demand level of the pedestrian crossings around the rail station;

[0176] S106.3 For any demand level, arrange the locations of the pedestrian crossing gaps in descending order of demand urgency index to generate a priority assessment list.

[0177] S106.4 Visualize and output the required levels and priority assessment list.

[0178] For example, based on standardized indicator data and the hierarchical weights determined by the AHP (Analytic Hierarchy Process), a two-layer weighted comprehensive evaluation formula is constructed to complete the quantitative calculation of the urgency index of pedestrian crossing gap demand:

[0179] This assessment employs a two-layer weighted summation logic: first, the score for each dimension is calculated, and then the dimension weights are added together to obtain the final urgency index.

[0180] First layer: Single-dimensional score calculation formula (weighted summation of all sub-indicators within the dimension)

[0181]

[0182] in, This is the comprehensive score for the k-th evaluation dimension; This represents the indicator layer weight corresponding to the m-th sub-indicator at this latitude; is the index value after range standardization for the corresponding sub-indicator; n is the total number of evaluation indicators within the corresponding dimension.

[0183] Second layer: Formula for calculating the urgency index of pedestrian crossing gap demand (weighted sum of scores from five dimensions)

[0184]

[0185] In the formula: The demand urgency index corresponding to the pedestrian crossing gap; The weights of the criteria layer corresponding to the five assessment dimensions are as follows: behavioral characteristics 30%, traffic flow characteristics 25%, land use 20%, road characteristics 15%, and infrastructure characteristics 10%. This represents the single-layer weighted score for the corresponding dimension.

[0186] Index Explanation: The demand urgency index ranges from [0,1]. The higher the value, the more urgent the need for pedestrian crossings, and the more priority should be given to constructing pedestrian crossings. The index value comprehensively reflects information from five dimensions: behavioral characteristics, traffic flow characteristics, land use, road characteristics, and facility characteristics, providing a scientific and quantitative basis for pedestrian crossing planning.

[0187] Demand Level Classification: Based on the urgency index, the demand for crossing the street is divided into four levels:

[0188] High demand (index ≥ 0.75): The demand is very urgent, and pedestrian crossings should be prioritized for construction.

[0189] Medium to high demand (0.60 ≤ index < 0.75): Demand is relatively urgent, and the construction of pedestrian crossings should be given priority.

[0190] Medium demand (0.45 ≤ index < 0.60): Demand is moderate; construction of pedestrian underpasses could be considered.

[0191] Low demand (index < 0.45): Demand is not urgent, and the construction of pedestrian crossings is not considered at this time.

[0192] Prioritization: Within each demand level, pedestrian crossings are ranked from highest to lowest according to their urgency index, generating a priority list for their construction. Planning departments can then proceed with the construction of these pedestrian crossings according to the priority list, based on budget constraints and construction timelines.

[0193] Visualization output: The demand level and priority ranking results are visualized in the form of heat maps, hierarchical symbol maps, etc., which intuitively show the spatial distribution and priority ranking of pedestrian crossing demand, and provide intuitive reference for planning decisions.

[0194] Based on the above technical solution, the technical effects of the embodiments of this application are as follows:

[0195] Advantages of data fusion: This application integrates heterogeneous data from multiple sources, which, compared with single data source methods, can more comprehensively and accurately reflect the real pedestrian crossing demand around rail stations, especially significantly improving the accuracy of identifying implicit demand.

[0196] Improved algorithm accuracy: Kalman filtering is used to correct mobile phone signaling data drift, and hidden Markov models are combined for road network matching. Compared with traditional trajectory matching algorithms, the path backtracking accuracy is improved by more than 30%, effectively reducing position errors.

[0197] Precise demand identification: Identification rules are specifically established for "detour" and "jaywalking" behaviors. The DBSCAN spatial clustering algorithm automatically identifies demand hotspots, which has a wider coverage, higher update frequency, and lower cost compared to manual survey methods.

[0198] Scientific evaluation system: Construct a multi-dimensional comprehensive evaluation model that organically combines physical space, socio-economic factors, facility characteristics, and behavioral characteristics. Compared with single-dimensional evaluation methods, the decision-making basis is more scientific and comprehensive.

[0199] Intelligent planning and decision-making: By calculating the urgency index of demand, quantitative decision support is provided for the planning of pedestrian crossings. Compared with experience-based decision-making methods, the planning schemes are more accurate and reasonable, which can effectively improve the efficiency of urban traffic operation and the travel experience of citizens.

[0200] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for assessing the demand for pedestrian crossings around rail stations based on multi-source data, characterized in that, include: Collect multi-source heterogeneous data; the multi-source heterogeneous data includes mobile phone signaling data, land use and POI data points, existing municipal pedestrian crossing data, and road network data. The mobile phone signaling data is cleaned and the trajectory is reconstructed to obtain continuous spatiotemporal trajectory data of pedestrians. Based on the continuous spatiotemporal trajectory data, a deep learning algorithm is used to identify rail transit travel and backtrack the path to and from the station. Based on the entry and exit paths, and combined with the real location information of the rail station entrances and exits, the walking trajectory features of users on both sides of the road outside the station are extracted; Based on the walking trajectory characteristics of users on both sides of the road outside the station, the heat areas of pedestrian crossing demand are identified, and the heat areas of pedestrian crossing demand are matched and analyzed with the entrances and exits of the rail station to determine the location of the pedestrian crossing gap. A multi-dimensional evaluation index was constructed using land use and POI data, existing municipal pedestrian crossing data, and road network data. The multi-dimensional evaluation index was then standardized and its weights were determined to obtain the standardized multi-dimensional evaluation index and its corresponding weights. Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, the demand urgency index corresponding to the location of the pedestrian gap is obtained, and the demand urgency index is used to assess the demand for pedestrian passages around the rail station and visualize the assessment results.

2. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data as described in claim 1, characterized in that, The mobile phone signaling data is cleaned and its trajectory reconstructed to obtain continuous spatiotemporal trajectory data of pedestrians, including: The raw mobile signaling data is subjected to quality checks, and abnormal data is removed to obtain preprocessed mobile signaling data. The Kalman filter algorithm is used to dynamically correct the preprocessed mobile phone signaling data to obtain continuous spatiotemporal trajectory data of pedestrians.

3. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data as described in claim 1, characterized in that, Based on the continuous spatiotemporal trajectory data, a deep learning algorithm is used to identify rail transit travel and trace routes to obtain entry and exit routes, including: The study area is divided into grids to obtain multiple grid cells for analysis; the study area refers to the area where the demand for pedestrian crossings around rail stations is assessed. For any given grid cell, feature parameters are extracted from the continuous spatiotemporal trajectory data corresponding to that grid cell; the feature parameters include one or more of the following: travel duration, travel distance, average speed, number of trajectory points, number of stop points, whether it passes through the track station buffer zone, and density of surrounding roads; The feature parameters are analyzed using a pre-set deep learning algorithm to identify rail transit travel trajectories and extract the origin and destination information for each rail transit trip to obtain the travel mode. Based on the aforementioned travel mode, a hidden Markov model is used for path backtracking to obtain the entry and exit routes.

4. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 3, characterized in that, Based on the aforementioned travel mode, a hidden Markov model is used for path backtracking to obtain the entry and exit paths, including: Based on the aforementioned travel mode, each road segment in the road network is considered as a hidden state, and each observation point in the continuous spatiotemporal trajectory data is considered as an observation sequence. Hidden Markov Model construction, state transition probability calculation, observation probability calculation, Viterbi algorithm solution, and road network matching are performed sequentially to achieve path backtracking and obtain the entry and exit paths.

5. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 4, characterized in that, The multi-source heterogeneous data also includes rail transit AFC data; After obtaining the travel mode, the process also includes: using rail transit AFC data to perform spatiotemporal matching on the travel mode to obtain a spatiotemporal matching result; the spatiotemporal matching result includes matching or non-matching. Based on the premise that the spatiotemporal matching result is a mismatch, the accuracy of travel identification and route tracing is evaluated by comparing the difference between the identification result and the AFC data, and the evaluation result is obtained. Based on the evaluation results, adjust the parameters of the deep learning algorithm or the threshold for road network matching.

6. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 1, characterized in that, Based on the aforementioned entry and exit routes, and combined with the actual location information of the rail station entrances and exits, the user's walking trajectory features on both sides of the road outside the station are extracted, including: Based on the aforementioned entry and exit routes, each entry and exit route is segmented according to the road boundaries, and the walking trajectory segments on both sides of the road are identified. For walking trajectory segments that cross the road, they are marked as potential street crossing behaviors. For pedestrian trajectory segments with potential street crossing behavior, analyze the stopping points in the pedestrian trajectory segments to identify users' waiting behavior at the road edge; For any walking trajectory segment with waiting behavior, the user's detour behavior is identified by calculating the geometric features of the walking trajectory segment, and the detour point is obtained; By analyzing the crossing angle and crossing position of the trajectory, the user's cross-road behavior is identified, and the cross-road point is obtained. The detour points and the road crossing points are used together as the characteristics of the user's walking trajectory on both sides of the road outside the station.

7. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 6, characterized in that, Based on the walking trajectory characteristics of the users on both sides of the road outside the station, heat areas of pedestrian crossing demand are identified, including: The detour points and road crossing points in the pedestrian trajectory features are used as sample points for pedestrian crossing demand. The DBSCAN spatial clustering algorithm is used to automatically identify the heat areas of pedestrian crossing demand.

8. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 1, characterized in that, By matching the heat zone of the pedestrian crossing demand with the entrances and exits of rail stations, the location of pedestrian crossing gaps is determined, including: The precise location coordinates of each entrance and exit of the station are obtained from the rail transit design drawings or on-site surveys; the precise location coordinates include the entrance / exit number, location coordinates, orientation, and width information; Based on the precise location coordinates, the distance between each entrance / exit and each hotspot area is calculated. If the distance is less than the set matching radius, the entrance / exit is determined to match the hotspot area. For any hotspot area that matches an entrance or exit, analyze the location of pedestrian crossing gaps that are missing, lack facilities, have insufficient facilities, or are poorly located. Based on the density of hotspot areas, traffic flow at entrances and exits, and road grade, the identified pedestrian gaps are prioritized and their locations are output.

9. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 1, characterized in that, A multi-dimensional evaluation index was constructed using land use and POI point data, existing municipal pedestrian crossing data, and road network data. The multi-dimensional evaluation index was then standardized and its weights were determined, resulting in standardized multi-dimensional evaluation indicators and their corresponding weights, including: Based on the land use and POI data, existing municipal pedestrian crossing data, and road network data, multiple evaluation indicators are constructed in the behavioral characteristic dimension, traffic flow characteristic dimension, land use dimension, road characteristic dimension, and facility characteristic dimension to obtain multi-dimensional evaluation indicators. The multidimensional evaluation indicators are dedimensionalized to obtain standardized multidimensional evaluation indicators. The weights of the standardized multi-dimensional evaluation indicators are obtained by using the analytic hierarchy process (AHP), thus obtaining the standardized multi-dimensional evaluation indicators and their corresponding weights.

10. The method for assessing the demand for pedestrian crossings around rail stations based on multi-source data according to claim 1, characterized in that, Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, the demand urgency index corresponding to the location of the pedestrian crossing gap is obtained. Then, based on the demand urgency index, a demand assessment of pedestrian crossings around rail stations is performed, and the assessment results are visualized, including: Based on the standardized multi-dimensional evaluation indicators and their corresponding weights, a weighted sum is performed to obtain the demand urgency index corresponding to the location of the pedestrian gap; The urgency index corresponding to the location of the pedestrian crossing gap is matched with the preset demand level classification range to determine the demand level of the pedestrian crossings around the rail station. For any given demand level, the locations of the pedestrian crossing gaps are arranged in descending order of demand urgency index to generate a priority assessment list. The required levels and priority assessment list will be visualized and output.

Images