A method for obtaining bus corridor passenger flow based on IC card and GPS data

Abstract

The present invention discloses a method for obtaining bus corridor cross-sectional passenger flow based on IC card and GPS data, which relates to the field of data fusion analysis technology. The method specifically includes the following steps: defining a cross-sectional space; the cross-sectional space includes determining a cross-sectional range and determining a direction; calculating the time it takes for a vehicle to pass through the cross-sectional space; constructing spatiotemporal constraints based on the cross-sectional space; using the constructed spatiotemporal constraints as a benchmark, predicting the vehicle arrival time at each station within the cross-sectional space based on IC card data records; calculating a basic cross-sectional passenger flow prediction based on GPS data quality and vehicle arrival time differences; optimizing the basic cross-sectional passenger flow prediction based on geographic location; and controlling errors through long-term iteration. Based on data fusion, the present invention can obtain bus corridor cross-sectional passenger flow for any time period, breaking through the limitations of traditional station statistics methods.

Description

Technical Field

[0001] The present invention relates to the technical field of data fusion analysis, and in particular to a method for obtaining passenger flow in a bus corridor section based on IC card and GPS data. Background Art

[0002] Data fusion technology is an information processing technology that integrates information from multiple data sources. It improves the accuracy, reliability and integrity of data by coordinating, combining and interpreting data from different sources.

[0003] Bus passenger flow analysis is crucial for public transportation planning, optimizing scheduling, and improving passenger travel efficiency. IC card data provides information on passenger boarding times and stops, while GPS data records bus trajectories. The integration of these two types of data helps obtain more accurate bus passenger flow information and provides a basis for optimizing public transportation systems. Traditional bus passenger flow estimation methods mainly rely on passenger surveys, fixed monitoring point data, or empirical models based on historical data. However, these methods suffer from data lag, limited coverage, and difficulty adapting to real-time changes. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data, so as to solve the problems raised in the prior art.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solution: a method for obtaining bus corridor cross-sectional passenger flow based on IC card and GPS data, the method specifically comprising the following steps:

[0006] Step S100, defining a cross-sectional space; the cross-sectional space includes determining a cross-sectional range and determining a direction;

[0007] Step S200, calculating the time it takes for the vehicle to pass through the cross-sectional space;

[0008] Step S300: constructing space-time constraints based on the cross-sectional space;

[0009] Step S400: Based on the constructed spatiotemporal constraints and IC card data records, predict the vehicle arrival time at each station in the cross-sectional space;

[0010] Step S500: Calculate the passenger flow forecast for the basic section based on GPS data quality and vehicle arrival time differences;

[0011] Step S600: Optimizing the passenger flow prediction of the basic section based on the geographical location;

[0012] Step S700: Control error through long-term iteration;

[0013] Define the cross-sectional space, which includes determining the cross-sectional range and direction, specifically:

[0014] The cross-sectional space is established through a GIS coordinate system;

[0015] The cross section range is set as a rectangle; the starting point coordinates (x1, y1) and the end point coordinates (x m ,y m ) determines the length; the width is the actual width of the road; where x1 and x m Represents the horizontal coordinates of the starting point and the end point respectively; y1 and y m Represent the ordinates of the starting point and the end point respectively; m is a positive integer;

[0016] Determine the direction through the continuous displacement coordinates of the vehicle's GPS trajectory points;

[0017] Calculate the time it takes for a vehicle to pass through the cross-sectional space, specifically:

[0018] Step S3-1, obtain the GPS track point sequence of the vehicle, the GPS track point sequence is characterized as ; Among them, (t i ,x i ,y i ) represents the i-th GPS track point; t i Indicates the acquisition time of GPS track points; and They represent the position coordinates of the GPS track points of the vehicle respectively; n represents the number of track points in the GPS track point sequence, n is a positive integer; i represents the number label of the track point, i is a positive integer, 1≤i≤n;

[0019] Step S3-2: Filter the point set S that enters the cross-section space in the trajectory points in , the point set S in ={(t i ,x i ,y i )|(x1≤x i ≤x m )∩(y1≤y i ≤y m )};

[0020] Step S3-3: If the point set S in Not empty, select the first point time t first and the last point time t last ;

[0021] Step S3-4, using linear interpolation to calculate the time of entering and exiting the cross-sectional space;

[0022] Specifically:

[0023] ;

[0024] ;

[0025] Among them, t entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space; L1 indicates the minimum displacement value from the cross-sectional space into the boundary; L2 indicates the minimum displacement value from the cross-sectional space out of the boundary; v1 indicates the clockwise speed when entering the boundary; v2 indicates the clockwise speed when exiting the boundary;

[0026] Furthermore, v1 can be obtained by obtaining the time and position coordinates of the two minimum trajectory points in the GPS trajectory points that are within the cross-section space boundary; v2 can be obtained similarly;

[0027] According to the cross-sectional space, a time-space constraint condition is constructed to obtain GPS trajectory points that meet the time-space constraint condition, specifically:

[0028] Ensure that the vehicle's traveling direction is consistent with the determined direction of the cross-sectional space;

[0029] The vehicle's travel time range is [ , ]; where t entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space;

[0030] Based on the constructed spatiotemporal constraints and IC card data records, the vehicle arrival time at each station in the cross-sectional space is predicted as follows:

[0031] Step S5-1: Obtain the passenger's boarding station and boarding time, and obtain the average travel time matrix of the route based on historical data;

[0032] Furthermore, the route average travel time matrix is ​​obtained in the following manner:

[0033] ;

[0034] in, Indicates a site To the site The historical average travel time, u and v represent station identifiers; N represents the total number of vehicles departing from the station during the statistical period. The number of valid trips to the site, N is a positive integer; Indicates a site Arrival time; Indicates that the slave site The departure time of the vehicle, k represents the valid number of trips of the vehicle, k is a positive integer, 1≤k≤N;

[0035] The effective number of trips is specifically:

[0036] If the trip time exceeds ±3 times the standard deviation of the mean, it is considered abnormal (e.g., traffic congestion caused by a traffic accident) and is not counted as a valid trip.

[0037] Step S5-2: Adjust the historical average travel time according to the real-time traffic conditions, obtain the deviation of the actual travel time, and calculate the correction term;

[0038] The correction term reflects the deviation between the actual travel time on that day and the historical average time, thereby more accurately estimating the passenger disembarkation time;

[0039] According to the vehicle's GPS track points, the actual driving time T' of the vehicle between different stations is obtained;

[0040] The calculation formula of the correction term is specifically represented as follows:

[0041] ;in, represents the correction term; T represents the historical average travel time;

[0042] As a preferred method, T can be obtained by calculating the N valid trips closest to the current time;

[0043] Step S5-3: Based on the boarding time and average travel time matrix obtained in step S5-1 and the correction term obtained in step S5-2, predict the vehicle arrival time at each station in the cross-sectional space;

[0044] The vehicle arrival time is the algebraic sum of the departure time of the previous station in the cross-sectional space, the historical average travel time between the corresponding stations, and the correction term between the corresponding stations;

[0045] The predicted passenger flow for the basic section is calculated based on the GPS data quality and the difference in vehicle arrival time. Specifically:

[0046] Step S6-1, calculating the GPS data quality weight based on the GPS trajectory point sequence of the vehicle;

[0047] Furthermore, the calculation formula of the GPS data quality weight is characterized as follows: β=Q / Q'; wherein β represents the GPS data quality weight; Q represents the number of GPS track points in the cross-sectional space, Q is a positive integer, Q≤n; Q' represents the theoretical number of GPS track points; Q'=c / l; wherein c represents the length of the cross-sectional space; l represents the GPS sampling interval distance;

[0048] Step S6-2: Calculate the vehicle arrival time difference based on historical error calibration;

[0049] Specifically: ;

[0050] in, represents the difference in vehicle arrival times; e represents a natural constant; Indicates the actual arrival time of the corresponding station in history; Indicates the predicted vehicle arrival time; represents the time attenuation coefficient, which is obtained by fitting historical data with natural constants;

[0051] Step S6-3: Calculate the basic cross-sectional passenger flow forecast using the cross-sectional spatial passenger flow aggregation model;

[0052] The expression of the cross-sectional passenger flow aggregation model is:

[0053] ;

[0054] Among them, A represents the passenger flow forecast of the basic section; represents the GPS data quality weight of vehicle h; Represents a set of vehicles with a certain direction and entry and exit time that conforms to the cross-sectional space; represents the arrival time difference of the vehicle taken by passenger p; represents the set of passengers on vehicle h; represents the probability of passenger p getting off the bus in the cross-sectional space, which is obtained based on the frequency G of the number of people getting off the bus between stations; I represents the time-space matching indicator function; represents the current time stamp of passenger p;

[0055] Among them, A is the passenger flow forecast of the basic section, so it can be rounded up or down according to the actual situation;

[0056] The logical judgment condition of the space-time matching indicator function I is: ; Takes 1 if the condition is met, and takes 0 if it is not met; the condition checks whether the time period of passenger p's ride overlaps with the time period of vehicle h passing through the cross-section space; that is, the passenger's boarding time is earlier than the time the vehicle leaves the cross-section, and the passenger's disembarkation time is later than the time the vehicle enters the cross-section; this ensures that the passenger is on the vehicle when the vehicle passes through the cross-section, and is therefore counted in the passenger flow of the cross-section;

[0057] In step S6-3, the statistical frequency of the number of people getting off the bus in history is added to the cross-sectional passenger flow aggregation model, specifically:

[0058] Collect the number of people getting on and off at different stops in the history;

[0059] Based on the number of people getting on and off at different stations, calculate the statistical frequency of the number of people getting off at different stations;

[0060] Based on the influence of geographical location, the passenger flow forecast of the basic section is optimized as follows:

[0061] Among them, the impact of geographical factors on passenger flow in bus corridor sections is mainly reflected in the nature of surrounding land use and spatial accessibility;

[0062] Step S8-1, constructing a geographic influence coefficient function by using the distance between the geographic location and the cross-sectional space, the average annual passenger flow of the geographic location, and the distance attenuation coefficient obtained by fitting historical data;

[0063] Specifically: ;

[0064] in, represents the geographical influence coefficient; represents the number of POIs of the fth category; Indicates the straight-line distance from POI to the cross-section space; represents the weight of the f-th category POI, reflecting the attraction of this type of facility to passenger flow, and is calibrated based on historical passenger flow regression analysis; σ represents the distance attenuation coefficient, which controls the decay rate of POI influence with distance; f represents the category identifier of the POI; F is the total number of POI categories;

[0065] The POI represents a specific location or target in the GIS map, including restaurants, hotels, gas stations, parks, schools, hospitals, etc.

[0066] Step S8-2: Calculate the geographical influence coefficient based on the actual cross-sectional space and optimize the basic cross-sectional passenger flow prediction.

[0067] The specific characterization formula is:

[0068] ;in, It represents the predicted passenger flow of the basic section after optimization;

[0069] The error is controlled through long-term iteration, specifically:

[0070] Update model parameters based on accumulated historical data;

[0071] Minimize the error using gradient descent;

[0072] The gradient descent method is used to minimize the error, specifically

[0073] ;

[0074] Among them, ensure that the error value The model error can be smaller than that of the model composed of parameters before the update.

[0075] Compared with the existing technology, the beneficial effects of the present invention are as follows: based on the fusion of IC card and GPS data, combined with spatiotemporal constraints, geographic information optimization and error control technology, the present invention can achieve accurate acquisition of passenger flow in bus corridor sections, breaking through the limitations of traditional station statistical methods; the method can accurately monitor line passenger flow, improve the accuracy of vehicle arrival time prediction, and improve the accuracy of passenger flow calculation through multi-source data fusion; in addition, the introduction of geographical influence coefficients optimizes passenger flow prediction results, combines error control with long-term optimization mechanisms, improves adaptability to sudden traffic conditions, and provides a scientific basis for bus system optimization, thereby improving the operational efficiency and management level of urban public transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 This is a schematic flow chart of the steps of the method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data of the present invention;

[0077] Figure 2 Schematic diagram of a curve showing changes in POI with passenger flow density in Example 2 of a method for acquiring passenger flow in a bus corridor section based on IC cards and GPS data of the present invention. DETAILED DESCRIPTION

[0078] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0079] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, a method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data, and the bus corridor cross-section passenger flow obtaining method specifically comprises the following steps:

[0080] Step S100, defining a cross-sectional space; the cross-sectional space includes determining a cross-sectional range and determining a direction;

[0081] Step S200, calculating the time it takes for the vehicle to pass through the cross-sectional space;

[0082] Step S300: constructing space-time constraints based on the cross-sectional space;

[0083] Step S400: Based on the constructed spatiotemporal constraints and IC card data records, predict the vehicle arrival time at each station in the cross-sectional space;

[0084] Step S500: Calculate the passenger flow forecast for the basic section based on GPS data quality and vehicle arrival time differences;

[0085] Step S600: Optimizing the passenger flow prediction of the basic section based on the geographical location;

[0086] Step S700: Control error through long-term iteration;

[0087] Define the cross-sectional space, which includes determining the cross-sectional range and direction, specifically:

[0088] The cross-sectional space is established through a GIS coordinate system;

[0089] The cross section range is set as a rectangle; the starting point coordinates (x1, y1) and the end point coordinates (x m ,y m ) determines the length; the width is the actual width of the road; where x1 and x m Represents the horizontal coordinates of the starting point and the end point respectively; y1 and y m Represent the ordinates of the starting point and the end point respectively; m is a positive integer;

[0090] Determine the direction through the continuous displacement coordinates of the vehicle's GPS trajectory points;

[0091] Calculate the time it takes for a vehicle to pass through the cross-sectional space, specifically:

[0092] Step S3-1, obtain the GPS track point sequence of the vehicle, the GPS track point sequence is characterized as ; Among them, (t i ,x i ,y i ) represents the i-th GPS track point; t i Indicates the acquisition time of GPS track points; and They represent the position coordinates of the GPS track points of the vehicle respectively; n represents the number of track points in the GPS track point sequence, n is a positive integer; i represents the number label of the track point, i is a positive integer, 1≤i≤n;

[0093] Step S3-2: Filter the point set S that enters the cross-section space in the trajectory points in , the point set S in ={(t i ,x i ,y i )|(x1≤x i ≤x m )∩(y1≤y i ≤y m )};

[0094] Step S3-3: If the point set S in Not empty, select the first point time t first and the last point time t last ;

[0095] Step S3-4, using linear interpolation to calculate the time of entering and exiting the cross-sectional space;

[0096] Specifically:

[0097] ;

[0098] ;

[0099] Among them, t entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space; L1 indicates the minimum displacement value from the cross-sectional space into the boundary; L2 indicates the minimum displacement value from the cross-sectional space out of the boundary; v1 indicates the clockwise speed when entering the boundary; v2 indicates the clockwise speed when exiting the boundary;

[0100] Furthermore, v1 can be obtained by obtaining the time and position coordinates of the two minimum trajectory points in the GPS trajectory points that are within the cross-section space boundary; v2 can be obtained similarly;

[0101] According to the cross-sectional space, a time-space constraint condition is constructed to obtain GPS trajectory points that meet the time-space constraint condition, specifically:

[0102] Ensure that the vehicle's traveling direction is consistent with the determined direction of the cross-sectional space;

[0103] The vehicle's travel time range is [ , ]; where t entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space;

[0104] Based on the constructed spatiotemporal constraints and IC card data records, the vehicle arrival time at each station in the cross-sectional space is predicted as follows:

[0105] Step S5-1: Obtain the passenger's boarding station and boarding time, and obtain the average travel time matrix of the route based on historical data;

[0106] Furthermore, the route average travel time matrix is ​​obtained in the following manner:

[0107] ;

[0108] in, Indicates a site To the site The historical average travel time, u and v represent station identifiers; N represents the total number of vehicles departing from the station during the statistical period. To the site The effective number of trips, N is a positive integer; Indicates a site Arrival time; Indicates that the slave site The departure time of the vehicle, k represents the valid number of trips of the vehicle, k is a positive integer, 1≤k≤N;

[0109] The effective number of trips is specifically:

[0110] If the trip time exceeds ±3 times the standard deviation of the mean, it is considered abnormal (e.g., traffic congestion caused by a traffic accident) and is not counted as a valid trip.

[0111] Step S5-2: Adjust the historical average travel time according to the real-time traffic conditions, obtain the deviation of the actual travel time, and calculate the correction term;

[0112] The correction term reflects the deviation between the actual travel time on that day and the historical average time, thereby more accurately estimating the passenger disembarkation time;

[0113] According to the vehicle's GPS track points, the actual driving time T' of the vehicle between different stations is obtained;

[0114] The calculation formula of the correction term is specifically represented as follows:

[0115] ;in, represents the correction term; T represents the historical average travel time;

[0116] As a preferred method, T can be obtained by calculating the N valid trips closest to the current time;

[0117] Step S5-3: Based on the boarding time and average travel time matrix obtained in step S5-1 and the correction term obtained in step S5-2, predict the vehicle arrival time at each station in the cross-sectional space;

[0118] The vehicle arrival time is the algebraic sum of the departure time of the previous station in the cross-sectional space, the historical average travel time between the corresponding stations, and the correction term between the corresponding stations;

[0119] The predicted passenger flow for the basic section is calculated based on the GPS data quality and the difference in vehicle arrival time. Specifically:

[0120] Step S6-1, calculating the GPS data quality weight based on the GPS trajectory point sequence of the vehicle;

[0121] Furthermore, the calculation formula of the GPS data quality weight is characterized as follows: β=Q / Q'; wherein β represents the GPS data quality weight; Q represents the number of GPS track points in the cross-sectional space, Q is a positive integer, Q≤n; Q' represents the theoretical number of GPS track points; Q'=c / l; wherein c represents the length of the cross-sectional space; l represents the GPS sampling interval distance;

[0122] Step S6-2: Calculate the vehicle arrival time difference based on historical error calibration;

[0123] Specifically: ;

[0124] in, represents the difference in vehicle arrival times; e represents a natural constant; Indicates the actual arrival time of the corresponding station in history; Indicates the predicted vehicle arrival time; represents the time attenuation coefficient, which is obtained by fitting historical data with natural constants;

[0125] Step S6-3: Calculate the basic cross-sectional passenger flow forecast using the cross-sectional spatial passenger flow aggregation model;

[0126] The expression of the cross-sectional passenger flow aggregation model is:

[0127] ;

[0128] Among them, A represents the passenger flow forecast of the basic section; represents the GPS data quality weight of vehicle h; Represents a set of vehicles with a certain direction and entry and exit time that conforms to the cross-sectional space; represents the arrival time difference of the vehicle taken by passenger p; represents the set of passengers on vehicle h; represents the probability of passenger p getting off the bus in the cross-sectional space, which is obtained based on the frequency G of the number of people getting off the bus between stations; I represents the time-space matching indicator function; represents the current time stamp of passenger p;

[0129] The logical judgment condition of the space-time matching indicator function I is: ; Takes 1 if the condition is met, and takes 0 if it is not met; the condition checks whether the time period of passenger p's ride overlaps with the time period of vehicle h passing through the cross-section space; that is, the passenger's boarding time is earlier than the time the vehicle leaves the cross-section, and the passenger's disembarkation time is later than the time the vehicle enters the cross-section; this ensures that the passenger is on the vehicle when the vehicle passes through the cross-section, and is therefore counted in the passenger flow of the cross-section;

[0130] In step S6-3, the statistical frequency of the number of people getting off the bus in history is added to the cross-sectional passenger flow aggregation model, specifically:

[0131] Collect the number of people getting on and off at different stops in the history;

[0132] Based on the number of people getting on and off at different stations, calculate the statistical frequency of the number of people getting off at different stations;

[0133] Based on the influence of geographical location, the passenger flow forecast of the basic section is optimized as follows:

[0134] Among them, the impact of geographical factors on passenger flow in bus corridor sections is mainly reflected in the nature of surrounding land use and spatial accessibility;

[0135] Step S8-1, constructing a geographic influence coefficient function by using the distance between the geographic location and the cross-sectional space, the average annual passenger flow of the geographic location, and the distance attenuation coefficient obtained by fitting historical data;

[0136] Specifically: ;

[0137] in, represents the geographical influence coefficient; represents the number of POIs of the fth category; Indicates the straight-line distance from POI to the cross-section space; represents the weight of the f-th category POI, reflecting the attraction of this type of facility to passenger flow, and is calibrated based on historical passenger flow regression analysis; σ represents the distance attenuation coefficient, which controls the decay rate of POI influence with distance; f represents the category identifier of the POI; F is the total number of POI categories;

[0138] The POI represents a specific location or target in the GIS map, including restaurants, hotels, gas stations, parks, schools, hospitals, etc.

[0139] Example 1,

[0140] Use ArcGIS Pro to load a city's road network vector data (source: a city's Planning and Natural Resources Commission), and combine it with a German map API to supplement the real-time traffic layer;

[0141] Section space definition:

[0142] Section range:

[0143] Starting point coordinates: (x1, y1) = (116.455°E, 39.908°N); measured from the road centerline, extending 25 meters laterally to cover the bus lane;

[0144] End point coordinates: (x m ,y m)=(116.470°E,39.908°N);

[0145] Length: approximately 1.5 km (span of 0.015° longitude, unchanged latitude);

[0146] Width: The actual width of the road is 50 meters (including the bus lane), generating a rectangular geofence;

[0147] Direction determination:

[0148] Obtain GPS trajectories of 10 sample vehicles from the bus group and verify that the proportion of eastbound (increasing longitude) trajectories is greater than 95%, ensuring accurate direction definition.

[0149] That is, the continuous eastward displacement (increasing longitude) of the vehicle GPS trajectory point is the valid direction;

[0150] Bus onboard GPS terminal (model: Douxingtong**);

[0151] GPS track data (bus route 40* in a certain city, license plate *A72***);

[0152] GPS track point sequence:

[0153]

[0154] ;

[0155] Filter the point set S that enters the section in :

[0156] satisfy and There are 12 trajectory points in total;

[0157] Entry and exit time calculation:

[0158] First point time t first =8:00:30;

[0159] Last point time t last =8:02:45;

[0160] Linear interpolation calculates the in and out times:

[0161] t entry =8:00:30−(50m / 8.33m / s)≈8:00:24;

[0162] t exit =8:02:45+(50m / 7.14m / s)≈8:02:52;

[0163] Calculate the GPS data quality weight based on the GPS trajectory point sequence of the vehicle;

[0164] Furthermore, the calculation formula of the GPS data quality weight is characterized as follows: β=Q / Q'; wherein β represents the GPS data quality weight; Q represents the number of GPS track points in the cross-sectional space, Q is a positive integer, Q≤n; Q' represents the theoretical number of GPS track points; Q'=c / l; wherein c represents the length of the cross-sectional space; l represents the GPS sampling interval distance;

[0165] Q'=c / l=1500m / (8.33m / s*10s)≈18;

[0166] β = Q / Q' ≈ 0.67;

[0167] Based on historical error calibration, calculate the vehicle arrival time difference:

[0168] Extract 1% of the passenger records from the historical data, compare the predicted and actual get-off times, and fit the exponential decay parameter λ;

[0169] Calibration results of a city data: λ=0.1,

[0170] When the error is 3 minutes, α=e −0.1×3 ≈0.74;

[0171] Example 2,

[0172] The cross-section space of a pedestrian street on a certain east road in a certain city (east to west), within a radius of 500 meters:

[0173] Based on the historical passenger flow regression analysis, the weights of different categories of POIs are obtained, as shown in Table 1:

[0174] like Figure 2 As shown, this is the curve of POI changing with passenger flow density;

[0175]

[0176] Table 1 Weight comparison table of different categories of POI

[0177] There are two metro stations (w = 0.8), 0.2 km and 0.4 km away from the section;

[0178] There are three large shopping malls (w=0.5), with distances of 0.1km, 0.3km, and 0.5km;

[0179] 5 office buildings (w=0.6), average distance 0.25km;

[0180] Distance attenuation coefficient: =1.5;

[0181] ; (POI is a subway station);

[0182] Similarly, we get:

[0183] Z2=0.66; (POI is a large shopping mall);

[0184] Z3=1.66; (POI is office building);

[0185] The geographical influence coefficient of the cross-sectional space of a pedestrian street on East Road in a certain city is Z=3.13.

[0186] As a preferred method, ArcGIS Spatial Analyst tool is used to generate POI density heat map through spatial interpolation. The quantification of facility concentration can also construct the geographic influence coefficient function, which is as follows:

[0187] ;

[0188] Among them, peo represents the service population size of the f-th type POI;

[0189] Step S8-2: Calculate the geographical influence coefficient based on the actual cross-sectional space and optimize the basic cross-sectional passenger flow forecast.

[0190] The specific characterization formula is:

[0191] ;in, It represents the predicted passenger flow of the basic section after optimization;

[0192] The error is controlled through long-term iteration, specifically:

[0193] Update model parameters based on accumulated historical data;

[0194] Minimize the error using gradient descent;

[0195] The gradient descent method is used to minimize the error, specifically

[0196] ;

[0197] Among them, ensure the error value The model error can be smaller than that of the model composed of parameters before the update.

[0198] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data, characterized by: The method for obtaining passenger flow in bus corridor sections specifically includes the following steps: Step S100, defining a cross-sectional space; the cross-sectional space includes determining a cross-sectional range and determining a direction; Step S200, calculating the time it takes for the vehicle to pass through the cross-sectional space; Step S300: constructing space-time constraints based on the cross-sectional space; Step S400: Based on the constructed spatiotemporal constraints and according to the IC card data records, predict the vehicle arrival time at each station in the cross-sectional space, specifically including: Step S5-1: Obtain the passenger's boarding station and boarding time, and obtain the average travel time matrix of the route based on historical data; The method for obtaining the average travel time matrix of the route is specifically as follows: ； in, Indicates a site To the site The historical average travel time, u and v represent station identifiers; N represents the total number of vehicles departing from the station during the statistical period. To the site The effective number of trips, N is a positive integer; Indicates a site Arrival time; Indicates that the slave site The departure time of the vehicle, k represents the valid number of trips of the vehicle, k is a positive integer, 1≤k≤N; Step S5-2: Adjust the historical average travel time according to the real-time traffic conditions, obtain the deviation of the actual travel time, and calculate the correction term; According to the vehicle's GPS track points, the actual driving time T' of the vehicle between different stations is obtained; The calculation formula of the correction term is specifically represented as follows: ;in, represents the correction term; T represents the historical average travel time; Step S5-3: Based on the boarding time and average travel time matrix obtained in step S5-1 and the correction term obtained in step S5-2, predict the vehicle arrival time at each station in the cross-sectional space; Step S500: Calculate the passenger flow forecast for the basic section based on GPS data quality and vehicle arrival time differences; Step S600: Optimizing the passenger flow prediction of the basic section based on the geographical location; Step S500 and step S600 are specifically as follows: Step S6-1, calculating the GPS data quality weight based on the GPS trajectory point sequence of the vehicle; Furthermore, the calculation formula of the GPS data quality weight is characterized as follows: β=Q / Q'; wherein β represents the GPS data quality weight; Q represents the number of GPS track points in the cross-sectional space, Q is a positive integer, Q≤n; Q' represents the theoretical number of GPS track points; Q'=c / l; wherein c represents the length of the cross-sectional space; l represents the GPS sampling interval distance; Step S6-2: Calculate the vehicle arrival time difference based on historical error calibration; Specifically: ; in, represents the difference in vehicle arrival times; e represents a natural constant; Indicates the actual arrival time of the corresponding station in history; Indicates the predicted vehicle arrival time; represents the time attenuation coefficient, which is obtained by fitting historical data with natural constants; Step S6-3: Calculate the basic cross-sectional passenger flow forecast using the cross-sectional spatial passenger flow aggregation model; The expression of the cross-sectional passenger flow aggregation model is: ； Among them, A represents the passenger flow forecast of the basic section; represents the GPS data quality weight of vehicle h; Represents a set of vehicles with a certain direction and entry and exit time that conforms to the cross-sectional space; represents the arrival time difference of the vehicle taken by passenger p; represents the set of passengers on vehicle h; represents the probability of passenger p getting off the bus in the cross-sectional space, Obtained based on the frequency G of the number of people getting off at each station; entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space; I indicates the time-space matching indicator function; represents the current time stamp of passenger p; Step S700: Control the error through long-term iteration.

2. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 1, characterized in that: Define the cross-sectional space, which includes determining the cross-sectional range and direction, specifically: The cross-sectional space is established through a GIS coordinate system; The cross section range is set as a rectangle; the starting point coordinates (x1, y1) and the end point coordinates (x m ,y m ) determines the length; the width is the actual width of the road; where x1 and x m Represents the horizontal coordinates of the starting point and the end point respectively; y1 and y m Represent the ordinates of the starting point and the end point respectively; m is a positive integer; The direction is determined by the continuous displacement coordinates of the vehicle's GPS trajectory points.

3. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 2, characterized in that: Calculate the time it takes for a vehicle to pass through the cross-sectional space, specifically: Step S3-1, obtain the GPS track point sequence of the vehicle, the GPS track point sequence is characterized as ; Among them, (t i ,x i ,y i ) represents the i-th GPS track point; t i Indicates the acquisition time of GPS track points; and They represent the position coordinates of the GPS track points of the vehicle respectively; n represents the number of track points in the GPS track point sequence, n is a positive integer; i represents the number label of the track point, i is a positive integer, 1≤i≤n; Step S3-2: Filter the point set S that enters the cross-section space in the trajectory points in , the point set S in ={(t i ,x i ,y i )|(x1≤x i ≤x m )∩(y1≤y i ≤y m )}; Step S3-3: If the point set S in Not empty, select the first point time t first and the last point time t last ; Step S3-4: use linear interpolation to calculate the time of entering and exiting the cross-sectional space.

4. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 3, characterized in that: According to the cross-sectional space, a time-space constraint condition is constructed to obtain GPS trajectory points that meet the time-space constraint condition, specifically: Ensure that the vehicle's traveling direction is consistent with the determined direction of the cross-sectional space; The vehicle's travel time range is [ , ]; where t entry Indicates the time of entering the cross-sectional space; t exit Indicates the time of exiting the cross-sectional space.

5. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 4, characterized in that: In step S6-3, the statistical frequency of the number of people getting off the bus in history is added to the cross-sectional passenger flow aggregation model, specifically: Collect the number of people getting on and off at different stops in the history; Based on the number of people getting on and off the bus at different stations, the statistical frequency of the number of people getting off the bus between different stations is calculated to form a cross-sectional passenger flow aggregation model.

6. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 5, characterized in that: Based on the influence of geographical location, the passenger flow forecast of the basic section is optimized as follows: Step S8-1, constructing a geographic influence coefficient function by using the distance between the geographic location and the cross-sectional space, the average annual passenger flow of the geographic location, and the distance attenuation coefficient obtained by fitting historical data; Step S8-2: Calculate the geographical influence coefficient based on the actual cross-sectional space and optimize the basic cross-sectional passenger flow prediction.

7. The method for obtaining bus corridor cross-section passenger flow based on IC card and GPS data according to claim 6, characterized in that: The error is controlled through long-term iteration, specifically: Update model parameters based on accumulated historical data; Minimize the error using gradient descent.

Images