A road section hour vehicle flow identification and correction method based on map traffic big data

Abstract

The application discloses a kind of based on map traffic big data's road section hour vehicle flow identification and correction method, belong to intelligent transportation technical field.Method includes establishing target area road section basic table, complete map road and local road section matching and calculate matching credibility;To the quality control and weighted fusion of multi-platform road speed, output effective speed and calculate speed credibility;Combined with road parameter, calculate initial vehicle flow;Access measured section data to build same window correction sample, generate hierarchical correction coefficient, and insufficient hierarchical sample is executed shrink smoothing;After hierarchical correction, low credibility back-off and special traffic state residual secondary modification, obtain final vehicle flow.The application gives full play to the advantage that map traffic big data covers widely, and construction and operation cost of greatly reducing whole road network flow acquisition is obtained.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically to a method for identifying and correcting hourly traffic flow on road segments based on map traffic big data. Background Technology

[0002] Road traffic volume is a key fundamental parameter in traffic operation assessment, road activity level calculation, road dust emission calculation, motor vehicle emission inventory compilation, traffic organization optimization, and refined urban management. Traditional methods of obtaining traffic volume mainly rely on fixed-section detection methods such as loop detectors, geomagnetic sensors, checkpoints, video recognition, radar detection, drone observation, or manual surveys. While these methods can obtain relatively accurate traffic volume data at installed or surveyed sections, the costs of equipment construction, maintenance, and calibration are high, and the spatial coverage is limited, making it difficult to achieve continuous identification across the entire road network, by direction, and at the hourly level.

[0003] In recent years, map platforms have been able to provide road traffic situation data with wide coverage and high update frequency, such as average road speed, congestion level, traffic index, road spatial lines, and direction information. This type of data is suitable for describing road operating conditions; however, it is affected by factors such as congestion and vehicle length, and therefore cannot represent the actual cross-sectional traffic volume. The same speed or congestion level may correspond to different traffic volumes under conditions such as smooth traffic at night, high traffic volume during the day, severe congestion with queues, congestion dissipation, and truck traffic in industrial areas. Therefore, map traffic data cannot be directly used as actual traffic volume.

[0004] The prior art closest to this invention mainly includes the following three categories: (1) Fixed detectors or checkpoints for obtaining traffic flow. This type of method directly counts the number of vehicles passing through by coils, geomagnetic sensors, video, radar or checkpoint equipment. It has high accuracy, but only covers a limited detection section and cannot be naturally extended to road sections without equipment. In addition, the construction, maintenance and fault verification costs of the equipment are high.

[0005] (2) Traffic flow estimation methods based on floating cars, probe vehicles, or GPS trajectories. These methods estimate road flow using probe vehicle speed, trajectory, number of probe vehicle passes, vehicle penetration rate, or sample expansion coefficient, which can improve coverage. However, they usually require original trajectory, probe vehicle quantity, or penetration rate information. For scenarios where only aggregated speed and congestion level data from map platforms are available, these methods are difficult to implement directly.

[0006] (3) Estimation methods based on speed-flow theory models. This type of method uses speed, density, and flow relationships from Greenshields, Greenberg, Underwood, and triangulation diagrams to estimate traffic flow based on real-time road speed and road grade parameters. This method is simple to implement, but it is prone to systematic biases under conditions such as high speed and low flow at night, severe congestion and low speed and low throughput, and high proportion of trucks in industrial areas or construction zones.

[0007] Therefore, a method is needed to correct and identify aggregated traffic data from map platforms into road segments, directions, and hourly traffic flows when original vehicle trajectories, probe vehicle passage counts, or vehicle penetration rates are unavailable. This method leverages the advantages of wide coverage and high frequency of map traffic big data, while also improving the accuracy, stability, and reproducibility of traffic flow identification results through measured cross-sectional traffic flow and vehicle type classification data. Summary of the Invention

[0008] The purpose of this invention is to provide a method for identifying and correcting hourly traffic flow on road segments based on map traffic big data, so as to solve the technical problems of existing technologies that are prone to systematic deviations and data quality fluctuations affecting the accuracy of identification when using aggregated road condition data to identify traffic flow under atypical traffic flow conditions.

[0009] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a method for identifying and correcting hourly traffic flow on road segments based on map traffic big data, comprising: Acquire aggregated road condition data from at least one map traffic data platform within the target area, and establish a basic table of road segments in the target area; The matching score between the map road object spatial line and the local road spatial line is calculated based on the spatial overlap rate, directional consistency and road name similarity. The matching score is then used to determine the matching road segment based on the matching score, the preset matching threshold and the candidate difference threshold, and the matching confidence is calculated. Quality control is performed on the average speed of roads matched to the same road segment, the same direction, and the same collection time. The speed data from multiple platforms are weighted and fused to obtain the fused speed. The speed reliability is calculated, and the effective speed constrained by the upper and lower boundaries is output. The initial traffic flow for each road segment, direction, and time period is calculated based on the effective speed, road grade, number of lanes, free-flow speed, congestion density, and multi-lane distribution correction factor. The traffic flow data of the measured correction section is accessed, and the traffic flow data is spatiotemporally matched with the map traffic condition data of the same road segment, the same direction, and the same time window to form a window correction sample. The correction ratio and sample weight of each correction sample are calculated. Based on the correction ratio of the correction samples, stratified correction coefficients are generated according to road class, hour, congestion level, date type and functional area type. When the number of effective samples in the target stratum is insufficient, the correction coefficients of the parent layer or adjacent layers are used for shrinkage and smoothing to obtain the stratified correction coefficients. The initial traffic flow is corrected using the stratified correction coefficient to obtain the first-corrected traffic flow. When the matching confidence, speed confidence, or stratified sample coverage is lower than the preset threshold, a reference traffic flow is introduced for low-confidence backoff to obtain the backoffed traffic flow. For at least one of the special traffic conditions, namely, smooth traffic at night, severe congestion, and high proportion of trucks, a second residual correction is performed to obtain the final corrected traffic flow.

[0010] Furthermore, the road segment basic table includes the road segment name, road space line, road grade, direction, number of lanes in each direction, road segment length, speed limit, free flow speed, congestion density, functional zone type, and administrative district, street, and grid identifiers.

[0011] Furthermore, the matching score between the map road object and the candidate road segment in the basic road segment table is calculated based on the spatial overlap rate, directional consistency, and road name similarity between the spatial lines of the map road object and the local road spatial lines. Matching road segments are determined based on the matching score, a preset matching threshold, and a candidate difference threshold. The matching confidence level is then calculated using the following formula: In the formula, Indicates the matching score; Indicates spatial overlap rate; Indicates the score for consistency of direction; Indicates the similarity of road names; , , These are the weighting coefficients, and + + =1; In the formula, Indicates the reliability of the match; Indicates the highest match score; This indicates the second-highest matching score; This indicates a preset matching threshold; This represents the candidate difference threshold.

[0012] Furthermore, the average speed of roads matched to the same road segment, the same direction, and the same collection time is subjected to quality control. The speed data from multiple platforms is weighted and fused to obtain the fused speed. The speed reliability is calculated, and the effective speed constrained by upper and lower boundaries is output. The calculation formula is as follows: In the formula, Indicates the fusion speed; This represents the average road speed on the p-th map platform; This represents the fusion weight of the p-th platform, which is determined by the platform's historical reliability, multi-platform consistency, and data integrity. In the formula, Indicates speed reliability; C range Indicates the reliability of the speed range; C time Indicates the reliability of temporal continuity; C cons Indicates the consistency and reliability across multiple platforms; C plat Indicates the platform's historical reliability; a, b, c, and d are weighted indices. In the formula, Indicates effective speed; Indicates the minimum effective speed threshold; This represents the free-flow velocity corresponding to road class r.

[0013] Furthermore, the initial traffic flow for a road segment, direction, and time period is calculated based on effective speed, road grade, number of lanes, free-flow speed, congestion density, and multi-lane distribution correction coefficient. The calculation formula is as follows: In the formula, i represents the road segment; d represents the direction; t represents the time window; and r represents the road class. Indicates the initial density; Indicates the initial traffic flow; This represents the congestion density of road class r; The free-flow velocity represents the road class r; Indicates the shape parameters of the velocity-density curve; Indicates the number of lanes in each direction; This represents the correction factor for multi-lane distribution; This represents the upper limit of the single-lane saturation flow rate for road class r; when When the flow rate exceeds the preset high flow rate threshold and is occurring at night or in a low-demand state, a free-flow low-flow baseline function is used to replace the initial value of the basic graph to avoid high velocity being misjudged as high flow rate; when When the flow rate is below the congestion speed threshold, queuing release capacity or saturation flow rate upper limit constraints are used to avoid excessive density due to low speed, which would lead to unreasonable flow rates.

[0014] Furthermore, the traffic flow data from the measured correction section is spatiotemporally matched with map traffic data within the same road segment, direction, and time window to form a window-based correction sample. The correction ratio and sample weight of each correction sample are calculated using the following formula: In the formula, This represents the correction ratio of the j-th corrected sample; This represents the measured traffic flow of the j-th sample; This represents the initial traffic flow for the j-th sample; This represents the minimum flow threshold to prevent division by zero or near-zero amplification. In the formula, C represents the weight of the j-th corrected sample; m,j Indicates the reliability of the match; C v,j Indicates speed reliability; C o,j Indicates the reliability of the measured count; C s,j Indicates the credibility of spatially similar road segments; C t,j Indicates the reliability of time synchronization; γ1 to γ5 are weighting indices.

[0015] Furthermore, based on the correction ratio of the correction samples, stratified correction coefficients are generated according to road level, hour, congestion level, date type, and functional area type. When the number of effective samples in the target stratum is insufficient, the correction coefficients of the parent layer or adjacent layers are used for shrinkage and smoothing to obtain the shrunken stratified correction coefficients, calculated as follows: In the formula, The correction coefficient for the stratified combination g is given by g = (r, h, c, day, f), where r represents the road class, h represents the hour, c represents the congestion level, day represents the date type, and f represents the functional zone type. RobustWeightedEstimator represents the weighted median, truncated mean, or robust regression estimator. In the formula, This represents the stratification correction factor after shrinkage; Indicates the correction factor for the parent layer or adjacent layer; This represents the number of valid samples in the target stratification. This represents the shrinkage strength parameter; The value is dynamically determined based on the dispersion of the parent layer samples; the more dispersed the parent layer samples, the better. The smaller the value, the more stable the parent layer samples. The larger the value, the better.

[0016] Furthermore, a first sample threshold and a second sample threshold are set. When the number of effective samples in the target stratum is less than the first sample threshold, the parent layer correction coefficient is directly used. When the number of effective samples in the target stratum is between the first sample threshold and the second sample threshold, the shrinkage formula is used. When the number of effective samples in the target stratum is greater than the second sample threshold, the target stratum robust correction coefficient is directly used.

[0017] Furthermore, the initial traffic flow is corrected once using the shrunk stratification correction coefficient to obtain the first-corrected traffic flow; when the matching confidence, speed confidence, or stratified sample coverage is lower than a preset threshold, a reference traffic flow is introduced for low-confidence backoff to obtain the backoffed traffic flow, calculated as follows: In the formula, This indicates the traffic flow rate after one correction. This represents the contraction correction coefficient for the layer combination to which road segment i, direction d, and time window t belong; In the formula, ω represents the credibility fusion weight; Indicates the coverage of stratified samples; This indicates the traffic flow after a low-reliability rollback; This indicates the reference traffic volume.

[0018] Furthermore, the residual secondary correction is performed on at least one of the special traffic conditions, namely, smooth traffic at night, severe congestion, and high proportion of trucks, to obtain the final corrected traffic flow, calculated as follows: In the formula, This indicates the traffic flow after adjustments for nighttime traffic conditions. Indicates nighttime reference flow rate; Indicates the upper limit coefficient for nighttime; This indicates the traffic flow after adjustments for severe congestion. This indicates the queuing release capacity of road class r; This indicates the traffic flow after correction for a high proportion of trucks. This indicates the proportion of trucks on the target road segment during the target time period; Indicates the proportion of the benchmark truck; Indicates the corrected strength of the truck structure; , , These indicators respectively represent the status of smooth traffic at night, severe congestion, and high proportion of trucks. This represents the state residual correction function.

[0019] Based on the above technical solution, the embodiments of the present invention can produce at least the following technical effects: This invention provides a method for identifying and correcting hourly traffic flow on road segments based on map traffic big data. It can identify traffic flow across the entire road network, by direction, and at the hourly level, relying solely on road condition data aggregated on the map platform. It does not require obtaining original vehicle trajectories, probe vehicle penetration rates, or the number of times a single vehicle passes through. It fully leverages the advantages of map traffic big data, which has wide coverage and high update frequency, and significantly reduces the construction and maintenance costs of acquiring traffic flow across the entire road network. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0021] The objective of this invention is achieved through the following technical solution: A method for identifying and correcting hourly traffic flow on road segments based on map traffic big data includes the following steps: S1, Data Acquisition Acquire road condition data from at least one map traffic data platform within the study area. This road condition data includes map road object spatial lines, average road speed, congestion level, collection time, travel direction, and road name. Map platforms may include Amap, Baidu Maps, Tencent Maps, or other navigation / vehicle-to-everything (V2X) platforms. The road condition data is aggregated and does not need to include individual vehicle trajectories, individual vehicle routes, probe vehicle passage counts, or vehicle penetration rates.

[0022] S2, Construction of Road Segment Basic Table Establish a basic table of road segments for the study area. This table should include at least the road_id, road spatial lines, road grade, direction, number of lanes per direction, segment length, speed limit, free-flow speed, congestion density, functional zone type, and administrative district / street / grid identifier. Functional zone types may include commercial areas, residential areas, industrial areas, construction zones, logistics park zones, dock zones, and urban-rural fringe areas.

[0023] S3, Matching map road objects with local road segments Based on the spatial overlap rate, directional consistency, and road name similarity between the spatial lines of map road objects and local road spatial lines, a matching score is calculated between the map road objects and candidate road segments in the basic road segment table. Candidate road segments with the highest matching score exceeding a preset matching threshold, and where the difference between the highest and second-highest matching scores exceeds a preset difference threshold, are identified as matched road segments.

[0024] In the formula, Indicates the matching score; Indicates spatial overlap rate; Indicates the score for consistency of direction; Indicates the similarity of road names; , , These are the weighting coefficients, and + + =1; In the formula, Indicates the reliability of the match; Indicates the highest match score; This indicates the second-highest matching score when there is no second candidate road segment. 0 can be taken; This indicates a preset matching threshold; This represents the candidate difference threshold; by using the highest score and candidate differences as constraints, the risk of map road object mismatch can be reduced.

[0025] S4, Speed ​​Quality Control and Fusion Quality control is performed on the average speed of roads matched to the same road_id, direction, and collection time. Quality control includes speed missing detection, detection of zero speed anomalies, detection of speeds exceeding reasonable speed limits, detection of speed jumps between adjacent time windows, and detection of speed differences across multiple platforms. When the speed differences between multiple platforms are less than a preset threshold, weighted fusion is performed; when the differences exceed the preset threshold, the time period is marked as a low-confidence period, and the median speed or the platform speed with the highest historical reliability is preferentially used.

[0026] In the formula, Indicates the fusion speed; This represents the average road speed on the p-th map platform; This represents the fusion weight of the p-th platform, which is determined by the platform's historical reliability, multi-platform consistency, and data integrity.

[0027] In the formula, Indicates speed reliability; C range Indicates the reliability of the speed range; C time Indicates the reliability of temporal continuity; C cons Indicates the consistency and reliability across multiple platforms; C plat This indicates the platform's historical reliability; a, b, c, and d are weighted indices.

[0028] In the formula, Indicates effective speed; Indicates the minimum effective speed threshold; This represents the free-flow velocity corresponding to road class r. Boundary constraints are used to prevent abnormal velocities from being directly included in the initial flow calculation.

[0029] S5, Initial Traffic Flow Estimation The initial traffic flow for a road segment, direction, and time period is calculated based on the effective speed, road grade, number of lanes, free-flow speed, congestion density, and multi-lane distribution correction factor obtained through quality control. This initial traffic flow is only used as a correctable initial value for subsequent corrections and is not directly used as the final traffic flow output.

[0030] In the formula, i represents the road segment; d represents the direction; t represents the time window; and r represents the road class. Indicates the initial density; Indicates the initial traffic flow; This represents the congestion density of road class r; The free-flow velocity represents the road class r; Indicates the shape parameters of the velocity-density curve; Indicates the number of lanes in each direction; This represents the correction factor for multi-lane distribution; This represents the upper limit of the single-lane saturation flow rate for road class r.

[0031] when When the flow rate exceeds the preset high flow rate threshold and is occurring at night or in a low-demand state, a free-flow low-flow baseline function is used to replace the initial value of the basic graph to avoid high velocity being misjudged as high flow rate; when When the flow rate is below the congestion speed threshold, queuing release capacity or saturation flow rate upper limit constraints are used to avoid excessive density due to low speed, which would lead to unreasonable flow calculations.

[0032] S6, Measured Correction Section Access Acquire traffic flow data and vehicle type classification data from the measured correction sections, and spatially match the measured correction sections with the road_id and direction in the road segment base table. Measured data can be obtained from traffic detection sections, traffic police checkpoints, manual video counting, automatic video recognition, roadside radar, drone video, or temporary traffic surveys.

[0033] In the formula, This represents the converted hourly measured traffic flow. This represents the number of vehicles observed within τ minutes; τ represents the observation duration in minutes.

[0034] S7. Construction of peer correction samples The measured traffic flow data and vehicle type classification data of the calibration section are matched with the map traffic condition data within the same road_id, direction, and time window to form a window-based calibration sample. The calibration sample includes road_id, direction, time window, initial traffic flow, measured traffic flow, matching confidence, speed confidence, measured count confidence, spatial same road segment confidence, and time synchronization confidence.

[0035] In the formula, This represents the correction ratio of the j-th corrected sample; This represents the measured traffic flow of the j-th sample; This represents the initial traffic flow for the j-th sample; This represents the minimum flow threshold to prevent division by zero or near-zero amplification. In the formula, C represents the weight of the j-th corrected sample; m,j Indicates the reliability of the match; C v,j Indicates speed reliability; C o,j Indicates the reliability of the measured count; C s,j Indicates the credibility of spatially similar road segments; C t,j Indicates the reliability of time synchronization; γ1 to γ5 are weighting indices.

[0036] S8. Generation of Stratified Correction Coefficients Based on the ratio between measured traffic flow and initial traffic flow in the calibration sample, stratified calibration coefficients are generated according to road class, hour, congestion level, date type, and functional zone type. The stratified calibration coefficients are estimated using weighted median, truncated mean, or robust regression to reduce the impact of outlier calibration samples on the results.

[0037] In the formula, The correction coefficient for the stratified combination g is represented by g=(r,h,c,day,f), where r represents the road class, h represents the hour, c represents the congestion level, day represents the date type, and f represents the functional zone type; RobustWeightedEstimator represents the weighted median, truncated mean, or robust regression estimator.

[0038] S9. Stratified shrinkage when samples are insufficient When a certain stratified combination of samples is insufficient, a shrinkage smoothing is performed using the correction coefficient of the parent stratum or adjacent strata. The parent stratum can be a coarser-grained stratum derived from the same road class, adjacent hours, adjacent congestion levels, the same date type, or the same functional zone type.

[0039] In the formula, This represents the stratification correction factor after shrinkage; Indicates the correction factor for the parent layer or adjacent layer; This represents the number of valid samples in the target stratification. This represents the shrinkage strength parameter; The value is dynamically determined based on the dispersion of the parent layer samples; the more dispersed the parent layer samples, the better. The smaller the value, the more stable the parent layer samples. The larger the value, the better.

[0040] Preferably, a first sample threshold and a second sample threshold are set. When the number of effective samples in the target stratum is less than the first sample threshold, the parent layer correction coefficient is directly used. When the number of effective samples in the target stratum is between the first sample threshold and the second sample threshold, a shrinkage formula is used. When the number of effective samples in the target stratum is greater than the second sample threshold, the target stratum robust correction coefficient is directly used.

[0041] S10, Predictive Traffic Flow Calculation and Low-Reliability Backoff The initial traffic flow for each road_id, direction, and hour is corrected using a hierarchical correction coefficient.

[0042] In the formula, This indicates the traffic flow rate after one correction. This represents the contraction correction coefficient for the layer combination to which road segment i, direction d, and time window t belong.

[0043] When the matching confidence, speed confidence, or sample coverage is lower than a preset threshold, a reference traffic flow is introduced for low-confidence rollback. The reference traffic flow can be determined by historical corrected traffic flow of the same road level, hour, date type, and functional zone type, corrected traffic flow of adjacent road segments, or parent layer traffic flow.

[0044] In the formula, ω represents the credibility fusion weight; Indicates the coverage of stratified samples; This indicates the traffic flow after a low-reliability rollback; This represents the reference traffic flow. Through this mechanism, low-confidence map inputs will not directly dominate the final traffic flow output.

[0045] S11, Secondary Correction of Residuals under Special Traffic Conditions The residuals are corrected twice for at least one of the following traffic conditions: smooth traffic at night, severe congestion, and high proportion of trucks, to obtain the final corrected traffic flow.

[0046] In the formula, in the formula, This indicates the traffic flow after adjustments for nighttime traffic conditions. Indicates nighttime reference flow rate; Indicates the upper limit coefficient for nighttime; This indicates the traffic flow after adjustments for severe congestion. This indicates the queuing release capacity of road class r; This indicates the traffic flow after correction for a high proportion of trucks. This indicates the proportion of trucks on the target road segment during the target time period; Indicates the proportion of the benchmark truck; Indicates the corrected strength of the truck structure; , , These indicators respectively represent the status of smooth traffic at night, severe congestion, and high proportion of trucks. This represents the state residual correction function.

[0047] Among them, the smooth flow at night can be triggered by the hourly range, the proportion of speeds exceeding the free flow speed threshold, the congestion level being smooth and historically low demand characteristics; the severe congestion can be triggered by the speeds falling below the congestion speed threshold, the congestion level reaching severe congestion, and the continuous low speed values ​​in adjacent time periods; the high proportion of trucks can be triggered by the functional area type, the vehicle type recognition results from checkpoints or videos, the proportion of vehicle types at the measured cross-section, or the proportion of relocated vehicle types.

[0048] S12, vehicle type ratio, average vehicle weight, and vehicle mileage output The vehicle type ratio is generated based on vehicle type classification data, and the average vehicle weight per road segment, direction, and hour is calculated based on the vehicle type ratio and the average weight of each vehicle type. Vehicle type classification can include passenger cars, buses, light trucks, medium trucks, heavy trucks, construction vehicles, or dump trucks.

[0049] In the formula, This represents the proportion of car model k; This represents the number of vehicles of model k. Indicates average vehicle weight; VKT represents the average weight of vehicle model k; VKT represents the vehicle's mileage. This represents the length of road segment i.

[0050] When there is no direct vehicle type observation data for the target road segment, vehicle type proportion migration weights can be constructed according to road grade, functional area type, hour, date type, and spatial distance from adjacent correction sections to migrate the measured vehicle type proportions to the target road segment. When the target road segment is located in an industrial area, logistics park, dock area, or construction area, the prior weight of truck vehicle proportions should be increased. When there is checkpoint or video vehicle type recognition data, the actual vehicle type recognition results should be used first.

[0051] S13. Model Validation and Result Reliability Grading The calibration samples were divided into training set, optimization set and validation set, and time extrapolation validation, spatial extrapolation validation and road grade stratification validation were performed respectively. The mean absolute error, root mean square error, mean absolute percentage error and correlation coefficient were calculated.

[0052] Based on R, MAPE, matching confidence, speed confidence, hierarchical sample coverage, and special state identification confidence, the road segment-direction-hour traffic flow is classified into high confidence, medium confidence, and low confidence levels.

[0053] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying and correcting hourly traffic flow on road segments based on map traffic big data, characterized in that, include: Acquire aggregated road condition data from at least one map traffic data platform within the target area, and establish a basic table of road segments in the target area; The matching score between the map road object spatial line and the local road spatial line is calculated based on the spatial overlap rate, directional consistency and road name similarity. The matching score is then used to determine the matching road segment based on the matching score, the preset matching threshold and the candidate difference threshold, and the matching confidence is calculated. Quality control is performed on the average speed of roads matched to the same road segment, the same direction, and the same collection time. The speed data from multiple platforms are weighted and fused to obtain the fused speed. The speed reliability is calculated, and the effective speed constrained by the upper and lower boundaries is output. The initial traffic flow for each road segment, direction, and time period is calculated based on the effective speed, road grade, number of lanes, free-flow speed, congestion density, and multi-lane distribution correction factor. The traffic flow data of the measured correction section is accessed, and the traffic flow data is spatiotemporally matched with the map traffic condition data of the same road segment, the same direction, and the same time window to form a window correction sample. The correction ratio and sample weight of each correction sample are calculated. Based on the correction ratio of the correction samples, stratified correction coefficients are generated according to road class, hour, congestion level, date type and functional area type. When the number of effective samples in the target stratum is insufficient, the correction coefficients of the parent layer or adjacent layers are used for shrinkage and smoothing to obtain the stratified correction coefficients. The initial traffic flow is corrected using the stratified correction coefficient to obtain the first-corrected traffic flow. When the matching confidence, speed confidence, or stratified sample coverage is lower than the preset threshold, a reference traffic flow is introduced for low-confidence backoff to obtain the backoffed traffic flow. For at least one of the special traffic conditions, namely, smooth traffic at night, severe congestion, and high proportion of trucks, a second residual correction is performed to obtain the final corrected traffic flow.

2. The method according to claim 1, characterized in that, The basic road segment table includes the road segment name, road space line, road grade, direction, number of lanes in each direction, road segment length, speed limit, free flow speed, congestion density, functional zone type, and administrative district, street, and grid identifiers.

3. The method according to claim 1, characterized in that, The matching score between the map road object and the candidate road segment in the basic road segment table is calculated based on the spatial overlap rate, directional consistency, and road name similarity between the spatial lines of the map road object and the local road spatial lines. Matching road segments are determined based on the matching score, a preset matching threshold, and a candidate difference threshold. The matching confidence level is then calculated using the following formula: ; In the formula, Indicates the matching score; Indicates spatial overlap rate; Indicates the score for consistency of direction; Indicates the similarity of road names; , , These are the weighting coefficients, and + + =1; ; In the formula, Indicates the reliability of the match; Indicates the highest match score; This indicates the second-highest matching score; This indicates the preset matching threshold; This represents the candidate difference threshold.

4. The method according to claim 1, characterized in that, The quality control is performed on the average speed of roads matched to the same road segment, the same direction, and the same collection time. Weighted fusion of speed data from multiple platforms is used to obtain the fused speed. Speed ​​reliability is calculated, and the effective speed constrained by upper and lower boundaries is output. The calculation formula is as follows: ; In the formula, Indicates the fusion speed; This represents the average road speed on the p-th map platform; This represents the fusion weight of the p-th platform, which is determined by the platform's historical reliability, multi-platform consistency, and data integrity. ; In the formula, Indicates speed reliability; C range Indicates the reliability of the speed range; C time Indicates the reliability of temporal continuity; C cons Indicates the consistency and reliability across multiple platforms; C plat Indicates the platform's historical reliability; a, b, c, and d are weighted indices. ; In the formula, Indicates effective speed; Indicates the minimum effective speed threshold; This represents the free-flow velocity corresponding to road class r.

5. The method according to claim 1, characterized in that, The initial traffic flow for a road segment, direction, and time period is calculated based on effective speed, road grade, number of lanes, free-flow speed, congestion density, and multi-lane distribution correction factor. The calculation formula is as follows: ; ; In the formula, i represents the road segment; d represents the direction; t represents the time window; and r represents the road class. Indicates the initial density; Indicates the initial traffic flow; This represents the congestion density of road class r; The free-flow velocity represents the road class r; Indicates the shape parameters of the velocity-density curve; Indicates the number of lanes in each direction; This represents the correction factor for multi-lane distribution; This represents the upper limit of the single-lane saturation flow rate for road class r; when When the flow rate exceeds the preset high flow rate threshold and is occurring at night or in a low-demand state, a free-flow low-flow baseline function is used to replace the initial value of the basic graph to avoid high velocity being misjudged as high flow rate; when When the flow rate is below the congestion speed threshold, queuing release capacity or saturation flow rate upper limit constraints are used to avoid excessive density due to low speed, which would lead to unreasonable flow rates.

6. The method according to claim 1, characterized in that, The traffic flow data from the access measured correction section is spatiotemporally matched with map traffic data within the same road segment, direction, and time window to form a window-based correction sample. The correction ratio and sample weight of each correction sample are calculated using the following formula: ; In the formula, This represents the correction ratio of the j-th corrected sample; This represents the measured traffic flow of the j-th sample; This represents the initial traffic flow for the j-th sample; This represents the minimum flow threshold to prevent division by zero or near-zero amplification. ; In the formula, C represents the weight of the j-th corrected sample; m,j Indicates the reliability of the match; C v,j Indicates speed reliability; C o,j Indicates the reliability of the measured count; C s,j Indicates the credibility of spatially similar road segments; C t,j Indicates the reliability of time synchronization; γ1 to γ5 are weighting indices.

7. The method according to claim 1, characterized in that, The correction ratio based on the correction samples is used to generate stratified correction coefficients according to road class, hour, congestion level, date type, and functional area type. When the number of effective samples in the target stratum is insufficient, the correction coefficients of the parent layer or adjacent layers are used for shrinkage and smoothing to obtain the shrunken stratified correction coefficients, calculated as follows: ; In the formula, The correction coefficient for the stratified combination g is given by g = (r, h, c, day, f), where r represents the road class, h represents the hour, c represents the congestion level, day represents the date type, and f represents the functional zone type. RobustWeightedEstimator represents the weighted median, truncated mean, or robust regression estimator. ; In the formula, This represents the stratification correction factor after shrinkage; Indicates the correction factor for the parent layer or adjacent layer; This represents the number of valid samples in the target stratification. This represents the shrinkage strength parameter; The value is dynamically determined based on the dispersion of the parent layer samples; the more dispersed the parent layer samples, the better. The smaller the value, the more stable the parent layer samples. The larger the value, the better.

8. The method according to claim 7, characterized in that, A first sample threshold and a second sample threshold are set. When the number of effective samples in the target stratum is less than the first sample threshold, the parent layer correction coefficient is directly used. When the number of effective samples in the target stratum is between the first sample threshold and the second sample threshold, the shrinkage formula is used. When the number of effective samples in the target stratum is greater than the second sample threshold, the target stratum robust correction coefficient is directly used.

9. The method according to claim 1, characterized in that, The initial traffic flow is corrected using the stratified correction coefficient to obtain the first-corrected traffic flow. When the matching confidence, speed confidence, or stratified sample coverage is lower than a preset threshold, a reference traffic flow is introduced for low-confidence backoff to obtain the backoffed traffic flow, calculated as follows: ; In the formula, This indicates the traffic flow rate after one correction. This represents the contraction correction coefficient for the layer combination to which road segment i, direction d, and time window t belong; ; ; In the formula, ω represents the credibility fusion weight; Indicates stratified sample coverage; This indicates the traffic flow after a low-reliability rollback; This indicates the reference traffic volume.

10. The method according to claim 1, characterized in that, The residual secondary correction is performed on at least one of the special traffic conditions, namely, smooth traffic at night, severe congestion, and high proportion of trucks, to obtain the final corrected traffic flow, calculated as follows: ; ; ; ; In the formula, This indicates the traffic flow after adjustments for nighttime traffic conditions. Indicates nighttime reference flow rate; Indicates the upper limit coefficient for nighttime; This indicates the traffic flow after adjustments for severe congestion. This indicates the queuing release capacity of road class r; This indicates the traffic flow after correction for a high proportion of trucks. This indicates the proportion of trucks on the target road segment during the target time period; Indicates the proportion of the benchmark truck; Indicates the structural strength correction of the truck; , , These indicators respectively represent the status of smooth traffic at night, severe congestion, and high proportion of trucks. This represents the state residual correction function.

Images