Method, device and equipment for detecting cross-regional personnel flow volume based on multi-source data

Abstract

The application discloses a cross-regional personnel flow volume detection method and device based on multi-source data and equipment, and relates to the technical field of personnel flow volume detection. The method comprises the following steps: collecting multi-modal traffic data and preprocessing, constructing a space-time correlation database based on the preprocessed multi-modal traffic data; constructing a passenger transport total volume model, a highway passenger transport OD model and a passenger transport channel identification model based on the space-time correlation database, calculating the cross-regional flow volume based on the passenger transport total volume model, and containing business / non-business highway travel sub-items; generating a global origin-destination point matrix by using the highway passenger transport OD model; aggregating high-flow paths through the passenger transport channel identification model; calculating a dynamic calibration passenger carrying coefficient, and combining a parameter optimization module to optimize the parameters of the cross-regional flow volume, the global origin-destination point matrix and the high-flow paths, so as to determine the cross-regional personnel flow report. The application provides a statistical system capable of improving the cross-regional personnel flow volume detection effect.

Description

Technical Field

[0001] This application relates to the field of personnel flow detection technology, and in particular to a method, apparatus and equipment for cross-regional personnel flow detection based on multi-source data. Background Technology

[0002] In recent years, the railway network (including trunk railways, intercity railways, suburban railways, and urban rail transit) has achieved leapfrog development, while the number of private cars has surged and the proportion of self-driving travel has increased significantly. The passenger transport market is showing diversified and fragmented characteristics, and the traditional statistical system based on the volume of operating road passenger transport (including scheduled buses and chartered buses) is no longer able to meet the actual needs.

[0003] To gain a comprehensive understanding of passenger transport operations, it is necessary to focus on the travel volume and direction of private cars and various modes of transportation, with a particular emphasis on the highway passenger transport sector. This requires coordinating commercial and non-commercial passenger transport to establish a more comprehensive and realistic statistical system for personnel movement. Consequently, the representativeness of traditional commercial highway passenger transport volume indicators is declining, and they cannot fully reflect the actual development of highway transport.

[0004] Therefore, providing a statistical system that can improve the detection effect of cross-regional population flow has practical application value and significance. Summary of the Invention

[0005] In order to provide a statistical system that can improve the detection effect of cross-regional population flow, this application provides a method, device and equipment for cross-regional population flow detection based on multi-source data.

[0006] Firstly, the objective of this invention is achieved through the following technical solution:

[0007] Methods for detecting cross-regional population mobility based on multi-source data include:

[0008] Multimodal traffic data is collected and preprocessed. A spatiotemporal correlation database is constructed based on the preprocessed multimodal traffic data. A passenger volume model, a highway passenger OD model, and a passenger corridor identification model are constructed based on the spatiotemporal correlation database. The total passenger volume model is used to calculate the total cross-regional flow, including commercial and non-commercial highway travel items. A global origin-destination matrix is ​​generated using the highway passenger OD model. High-volume paths are aggregated using the passenger corridor identification model.

[0009] The dynamic calibration passenger load factor is calculated, and the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths are optimized in combination with the parameter optimization module to determine the cross-regional personnel flow report.

[0010] By adopting the above technical solutions, multimodal traffic data is obtained through the integration of multimodal data collection channels such as online ticketing, highway toll collection, video surveillance, and mobile phone signaling. This facilitates the full-mode and all-time coverage of cross-regional passenger flow, solving the statistical blind spot problem caused by the reliance on a single data source in traditional methods. By dynamically calibrating the passenger load factor, such as combining video detection technology with the DPR algorithm, the error rate of non-commercial passenger car travel volume is reduced. This application constructs a passenger volume model, a highway passenger OD model, and a passenger channel identification model, and couples them together. This can identify high-flow channels between multiple cities within a specified large area (such as the Guangzhou-Shenzhen Expressway between 21 cities in Guangdong Province), improving the high-flow channel identification rate. Through parameter optimization, the impact of actual policies or extreme weather on cross-regional passenger flow can be analyzed to meet the needs of complex scenarios, thereby assisting traffic management departments in dynamically allocating transport capacity. Thus, this application achieves the goal of improving the detection effect and accuracy of cross-regional passenger flow detection.

[0011] In a preferred embodiment of this application, the multimodal data includes ticketing data from online ticketing devices, vehicle traffic flow data from highway toll collection devices, vehicle cross-sectional flow data from traffic control stations on ordinary national and provincial highways, video surveillance data, mobile phone signaling data, and ticketing and flight schedule data from railways and civil aviation.

[0012] Preprocessing of multimodal traffic data includes data cleaning, deduplication, spatiotemporal alignment, and standardization.

[0013] The calibration methods for the dynamic calibration of passenger load factor include a motorized travel estimation method based on mobile phone signaling, a sampling statistics method based on service area video surveillance, and a long-distance small target detection method optimized based on the DPR algorithm.

[0014] By adopting the above technical solutions, the data complementarity is strong. Among them, online ticketing data fills the gap in commercial passenger transport statistics, video surveillance makes up for the loopholes in non-commercial travel monitoring, and mobile phone signaling data strengthens user behavior profiles, breaking the limitations of traditional single data sources.

[0015] In a preferred embodiment of this application, the mobile travel estimation algorithm based on mobile phone signaling includes:

[0016]

[0017] Among them, private car trips = total motorized trips - urban public transport passenger volume - intra-county trips; total motorized trips = permanent residents × average number of motorized trips per person.

[0018] By adopting the above technical solution, the amount of car travel is estimated based on mobile phone signaling data, which makes up for the shortcomings of traditional statistical methods in covering private traffic flow and provides a new way to more accurately estimate the amount of cross-regional personnel flow. This application further refines the differences between different modes of travel by defining the relationship between the total amount of motorized travel and the passenger volume of urban public transportation and intra-county travel, which helps to more accurately assess the specific composition of cross-regional personnel flow.

[0019] In a preferred embodiment, this application includes:

[0020] In the aforementioned passenger volume model, cross-regional passenger flow = ∑(commercial road passenger volume + non-commercial road travel volume) + railway passenger volume + civil aviation passenger volume + waterway passenger volume; where non-commercial road travel volume is the sum of passenger car travel volume on expressways and passenger car travel volume on ordinary national and provincial highways, and passenger car travel volume on ordinary national and provincial highways is calculated using the formula:

[0021]

[0022] Among them, the number of passenger cars traveling on ordinary national and provincial highways = ∑(passenger cars on ordinary national and provincial highways × distance traveled on ordinary national and provincial highways) = average cross-sectional traffic volume of passenger cars on ordinary national and provincial highways × mileage of ordinary national and provincial highways;

[0023] Urban passenger volume = Urban passenger volume of public buses and trolleybuses + Urban rail transit passenger volume + Urban passenger volume of taxis + Ferry passenger volume.

[0024] By adopting the above technical solutions, the passenger volume of various transportation modes (road, rail, civil aviation, waterway) is comprehensively considered, providing a comprehensive perspective for understanding cross-regional population flow. This application can better reflect the actual traffic situation by distinguishing between passenger car travel volume on expressways and ordinary national and provincial roads and adjusting it in conjunction with the passenger load factor. At the same time, by classifying the passenger volume of different modes of transportation within the city in detail, it helps to understand the characteristics of population flow within the city.

[0025] In a preferred embodiment of this application, the number of passenger car trips on the highway is calculated using the following formula:

[0026]

[0027] Among them, the number of passenger cars traveling on highways = ∑(passenger cars traveling on highways × distance traveled on highways).

[0028] By adopting the above technical solution, the calculation of passenger car travel volume on highways can more accurately capture the distance traveled by vehicles on highways and the corresponding number of passengers; by specifically defining variables and parameters, the calculation process becomes transparent and easy to verify, enhancing the credibility of the passenger volume model.

[0029] In a preferred embodiment of this application: the generation of a global origin-destination matrix using the highway passenger transport OD model includes: YYOD ij =YYOD i +FYYOD i YYOD i =α∑x j FYYOD i =∑y j Among them, YYOD ij YYOD represents the OD of commercial and non-commercial passenger traffic in multiple cities, corresponding to each other. i For commercial highway passenger traffic; FYYOD i For non-commercial highway passenger traffic; i is the pairwise correlation coefficient between cities, ranging from 1 to 144; α is the expansion coefficient; x j This refers to individual OD information generated through online ticketing coefficient analysis; y j This refers to individual OD information generated from highway toll data analysis.

[0030] By adopting the above technical solutions, the global origin-destination matrix is ​​constructed by including both commercial and non-commercial highway passenger traffic. This matrix can comprehensively reflect the flow of people between different cities, covering not only officially operating passenger traffic but also taking into account the impact of non-commercial transportation modes such as private cars. By using data from networked ticketing devices and highway toll collection devices for sample expansion and combining it with specific city-level correlation coefficients, the calculated OD information becomes more accurate.

[0031] In a preferred embodiment of this application, the calibration method for the dynamically calibrated passenger load factor further includes a multiple linear regression analysis prediction method incorporating correction factors, comprising:

[0032] Y chunyun =β0+β1X1+β2X2+β3X3+e

[0033]

[0034] Among them, Y chunyun Y is the passenger load factor during the Spring Festival travel rush; Y is the passenger load factor for a specified time period. β0 is the adjustment coefficient for a specified time period, representing the ratio of operational passenger volume to operational passenger volume during the Spring Festival travel season; X1, X2, and X3 are independent variables, where X1 refers to the operational passenger volume parameter, X2 refers to the parameter for Class I passenger vehicles on the expressway network, and X3 refers to the parameter for the flow of small and medium-sized passenger vehicles on ordinary national and provincial highways; β0 is a constant term; β1, β2, and β3 are regression coefficients, where β1 is the regression coefficient for the operational passenger volume parameter; e is the error term, where β2 is the regression coefficient for the parameter for Class I passenger vehicles on the expressway network, and β3 is the regression coefficient for the parameter for the flow of small and medium-sized passenger vehicles on ordinary national and provincial highways; δ is a correction factor.

[0035] By adopting the above technical solution, a multiple linear regression model with integrated correction factors is used to dynamically calibrate the passenger load factor, which improves the adaptability and accuracy of the model. It can flexibly adjust the passenger load factor according to the changes of different independent variables (such as operational passenger volume, parameters of type I buses, etc.) to ensure that the prediction results in different time periods are closer to the actual situation. This application introduces a specified time period adjustment coefficient, which can effectively capture the population flow characteristics during special periods, especially for data during the Spring Festival travel season.

[0036] Secondly, the objective of this invention is achieved through the following technical solution:

[0037] A cross-regional population flow detection device based on multi-source data, characterized in that the device comprises:

[0038] A multimodal traffic data acquisition module is used to acquire multimodal traffic data and preprocess the multimodal traffic data; a spatiotemporal correlation database construction module is used to construct a spatiotemporal correlation database based on the preprocessed multimodal traffic data;

[0039] The model building and calculation module is used to build a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on the spatiotemporal correlation database; it is also used to calculate the total cross-regional flow based on the total passenger volume model, which includes commercial / non-commercial highway travel items; it uses the highway passenger OD model to generate a global origin-destination matrix; and it aggregates high-volume paths through the passenger corridor identification model.

[0040] The parameter optimization module is used to calculate the dynamically calibrated passenger load factor and, in combination with the dynamically calibrated passenger load factor, optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report.

[0041] Thirdly, the objective of this invention is achieved through the following technical solution:

[0042] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for detecting cross-regional population flow based on multi-source data.

[0043] Fourthly, the objective of this invention is achieved through the following technical solution:

[0044] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for detecting cross-regional population flow based on multi-source data.

[0045] In summary, this application includes at least one of the following beneficial technical effects:

[0046] 1. Significantly enhanced the ability to detect cross-regional passenger flow. Whether from the diversity of data sources, the rigor of the processing procedures, or the accuracy of model construction, it demonstrates its great potential in improving traffic management efficiency; 2. The use of a multiple linear regression model with integrated correction factors to dynamically calibrate the passenger load factor improves the model's adaptability and accuracy, and can flexibly adjust the passenger load factor according to changes in different independent variables (such as operational passenger volume, parameters of type I buses, etc.). Attached Figure Description

[0047] Figure 1 This is a flowchart of a cross-regional population flow detection method based on multi-source data in one embodiment of this application;

[0048] Figure 2 This is a schematic diagram of a device according to one embodiment of this application. Detailed Implementation

[0049] The present application will be further described in detail below with reference to the accompanying drawings.

[0050] In one embodiment, such as Figure 1 As shown, this application discloses a method for detecting cross-regional population flow based on multi-source data, which specifically includes the following steps:

[0051] S1: Collect multimodal traffic data and preprocess it, and build a spatiotemporal correlation database based on the preprocessed multimodal traffic data.

[0052] In this embodiment, the multimodal data includes ticketing data from the online ticketing system, vehicle traffic flow data from the highway toll system, vehicle cross-sectional flow data from traffic control stations on ordinary national and provincial highways, video surveillance data, mobile phone signaling data, and ticketing and flight schedule data from railways and civil aviation.

[0053] Specifically, the online ticketing system includes electronic ticketing systems for passenger transport systems such as high-speed rail, long-distance buses, and ferries. Ticketing data includes information such as passenger purchase time, travel time, departure point, and destination. Highway toll data includes data from ETC gantry systems and toll station license plate recognition equipment, including vehicle passage timestamps, entrance / exit stations (accurate to the toll station number), license plate numbers, and vehicle type (e.g., passenger cars, trucks). Data from traffic control stations on ordinary national and provincial highways comes from traffic flow monitoring equipment set up at key nodes on national and provincial highways, including cross-sectional traffic flow (categorized by vehicle type), average vehicle speed, and timestamps. Video surveillance data comes from camera networks at service areas, highway interchanges, and city entrances / exits. Mobile phone signaling data comes from mobile communication operator base station logs, including user mobile phone numbers (anonymized), base station handover timestamps, Location Area Codes (LACs), and Cell Identifiers (CIDs). Railway and civil aviation data comes from the 12306 railway system and airport departure systems, including train schedules, passenger load factors, flight departure and arrival times, and passenger throughput.

[0054] Furthermore, the preprocessing of multimodal traffic data includes data cleaning, deduplication, spatiotemporal alignment, and standardization. Data cleaning includes filtering overbooked orders (such as records of overcrowded seating) and duplicate ticket purchases (multiple tickets from the same user within the same time period) from online ticketing data; removing vehicles without license plate information or with abnormal travel routes (such as records of mismatched ETC entrances and MTC exits); and removing blurry frames (resolution <480P) and data from low-light nighttime periods (illuminance <5 lux).

[0055] Spatiotemporal alignment includes: first, establishing a unified spatiotemporal reference for all data, where the time reference is the UTC timestamp and the spatial reference is the WGS84 coordinate system.

[0056] Specifically, the steps for constructing a spatiotemporal relational database include:

[0057] First, establish data fusion rules: using "license plate number + timestamp" as the key field, link highway toll data with video surveillance data (e.g., matching the duration a vehicle spends at a service area); using "user mobile phone number + base station ID" as the key field, link mobile signaling data with railway and civil aviation data (e.g., locating a passenger's connecting route from home to the high-speed rail station). The database indexing mechanism is a spatiotemporal index, supporting fast data retrieval by region and time period.

[0058] S2: Construct a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on a spatiotemporal correlation database. Calculate the total cross-regional flow based on the passenger volume model, including commercial / non-commercial highway travel items. Generate a global origin-destination matrix using the highway passenger OD model. Aggregate high-flow paths through the passenger corridor identification model.

[0059] In this embodiment, the model formula for the total passenger volume model includes:

[0060] Cross-regional passenger flow = ∑(commercial road passenger volume + non-commercial road travel volume) + railway passenger volume + civil aviation passenger volume + waterway passenger volume; where non-commercial road travel volume is the sum of passenger car travel volume on expressways and passenger car travel volume on ordinary national and provincial highways, and passenger car travel volume on ordinary national and provincial highways is calculated using the formula:

[0061]

[0062] Among them, the number of passenger cars traveling on ordinary national and provincial highways = ∑(passenger cars on ordinary national and provincial highways × distance traveled on ordinary national and provincial highways) = average cross-sectional traffic volume of passenger cars on ordinary national and provincial highways × mileage of ordinary national and provincial highways;

[0063] Urban passenger volume = Urban passenger volume of public buses and trolleybuses + Urban rail transit passenger volume + Urban passenger volume of taxis + Ferry passenger volume.

[0064] The number of passenger car trips on highways is calculated using the following formula:

[0065]

[0066] Among them, the number of passenger cars traveling on highways = ∑(passenger cars traveling on highways × distance traveled on highways).

[0067] Specifically, the volume of people moving on highways is divided into passenger traffic on commercial highways and travel on non-commercial highways; passenger traffic on commercial highways includes buses and chartered buses; travel on non-commercial highways includes passenger cars with 9 seats or less.

[0068] In this embodiment, the global origin-destination matrix (also known as the global OD matrix) is generated using the highway passenger transport OD model, including:

[0069] YYOD ij =YYOD i +FYYOD i YYOD i =α∑x j FYYOD i =∑y j Among them, YYOD ij YYOD represents the OD of commercial and non-commercial passenger traffic in multiple cities, corresponding to each other. i For commercial highway passenger traffic; FYYOD i For non-commercial highway passenger traffic; i is the pairwise correlation coefficient between cities, taking the pairwise correlation coefficient between the 21 cities in Guangdong Province as an example, i ranges from 1 to 144; α is the expansion coefficient; x jThis refers to individual OD information generated through online ticketing coefficient analysis; y j This refers to individual OD information generated from highway toll data analysis.

[0070] In this embodiment, the passenger transport corridor identification model uses the DBSCAN clustering algorithm to obtain video surveillance traffic data (road segment cross-section vehicle flow ≥ 1000 vehicles / hour) and the global origin-destination matrix output by the highway passenger transport OD model. Taking the cross-regional passenger flow detection in Guangdong Province as an example: the road network of Guangdong Province is abstracted as a graph G = (V, E), where nodes are transportation hubs, such as highway entrances and exits, and service areas, and edges are road or rail connections. Node features include historical traffic flow, spatiotemporal coordinates, and surrounding facility density, while edge features include road grade, speed limit, and real-time traffic flow. The spatiotemporal neighborhood information of nodes is aggregated through a graph convolutional network (GCN) to capture the spatiotemporal correlation of traffic flow. The dynamic evolution process of traffic flow is modeled using gated cyclic units (GRUs), and a dynamic weight matrix Wt is output. Then, based on the dynamic weight matrix Wt and video surveillance traffic data, high-confidence paths are selected through an attention mechanism. The formula for calculating the corridor importance score is: S channal =σ(Wt×Concat(Q, K, V)), where Q, K, and V are the query, key, and value vectors, respectively, and σ is the activation function. The clustering threshold in the DBSCAN clustering algorithm is set to paths with an OD flow density ≥ 80% (e.g., sections of the Guangzhou-Shenzhen Expressway with a daily average traffic volume ≥ 80% are identified as core channels). A personnel flow threshold based on the detection time period is set to determine and aggregate high-flow paths.

[0071] S3: Calculate the dynamic calibration passenger load factor, and combine it with the parameter optimization module to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report.

[0072] In this embodiment, the calibration method for dynamically calibrating the passenger load factor includes a motorized travel estimation method based on mobile phone signaling, a sampling statistics method based on service area video surveillance, and a long-distance small target detection method optimized based on the DPR algorithm.

[0073] In this embodiment, the mobile travel estimation algorithm based on mobile phone signaling includes:

[0074]

[0075] Among them, private car trips = total motorized trips - urban public transport passenger volume - intra-county trips; total motorized trips = permanent residents × average number of motorized trips per person.

[0076] Specifically, the sampling statistical methods for service area video surveillance include:

[0077] Extract surveillance videos from 10% of service areas across the province (e.g., 10 service areas along the Guangzhou-Shenzhen Expressway); detect the number of people in vehicles using the YOLOv5 algorithm (accuracy ≥ 90%); and calculate the average passenger load factor within a given time period (e.g., a passenger load factor of 2.9 during the morning peak hours).

[0078] Specifically, the DPR algorithm optimization method includes:

[0079] Image reconstruction of small targets at a distance (>50 meters) is performed using a diffusion model (CDM); the localization accuracy of small target bounding boxes is enhanced by a Transformer network (mAP≥8.9%). For example, a truck illegally carrying two passengers had a 40% false negative rate with the traditional algorithm, while the DPR algorithm reduced the false negative rate to 8%.

[0080] Furthermore, parameter optimization includes multivariate linear regression calibration and real-time DPR optimization. The calibration method for dynamically calibrating the passenger load factor also includes a multivariate linear regression analysis prediction method incorporating correction factors, including:

[0081] Y chunyun =β0+β1X1+β2X2+β3X3+e

[0082]

[0083] Among them, Y chunyun Y is the passenger load factor during the Spring Festival travel rush (empirical value 1.3); Y is the passenger load factor for a specified time period. β0 is the adjustment coefficient for a specified time period, representing the ratio of operational passenger volume to operational passenger volume during the Spring Festival travel season; X1, X2, and X3 are independent variables, where X1 refers to the operational passenger volume parameter, X2 refers to the parameter for Class I passenger vehicles on the expressway network, and X3 refers to the parameter for the flow of small and medium-sized passenger vehicles on ordinary national and provincial highways; β0 is a constant term; β1, β2, and β3 are regression coefficients, where β1 is the regression coefficient for the operational passenger volume parameter; e is the error term, where β2 is the regression coefficient for the parameter for Class I passenger vehicles on the expressway network, and β3 is the regression coefficient for the parameter for the flow of small and medium-sized passenger vehicles on ordinary national and provincial highways; δ is a correction factor.

[0084] Furthermore, the parameter-optimized real-time DPR optimization includes feature enhancement of the video stream via a Transformer network to adjust the small target detection box coordinate error in real time.

[0085] In this embodiment, the cross-regional personnel flow report is a cross-regional personnel flow volume report that includes a spatiotemporal distribution map, a channel heat map, and a sensitivity analysis. Taking Guangdong Province as an example, the spatiotemporal distribution map is a flow heat map of 21 cities in Guangdong Province displayed on a GIS platform (such as the red high-flow area in the Guangzhou-Foshan-Zhaoqing metropolitan area). The channel heat map is used to mark the real-time traffic efficiency of high-flow channels such as the Guangzhou-Shenzhen Expressway and the Humen Bridge (such as triggering an early warning when the congestion index is ≥2.0). The sensitivity analysis is as follows: if the construction of a certain road section causes a 20% decrease in traffic capacity, the system automatically simulates its diversion effect on surrounding channels (such as a 15% increase in traffic on the Guangzhou-Shenzhen Riverside Expressway).

[0086] In one embodiment, the calibration method for dynamically calibrating the passenger load factor further includes multi-source data fusion calibration, specifically comprising the following steps:

[0087] S10: Integration of mobile signaling data (S 信令 ), video surveillance data (S 视频 ) and DPR algorithm detection data (S DPR Construct a multidimensional feature matrix:

[0088] S 融合 =ω1×S 信令 +ω2×S 视频 +ω3×S DPR Where ω1, ω2, and ω3 are spatiotemporal dynamic weighting coefficients, based on the regional congestion index (C). 拥堵 Dynamic adjustment: Where C 阈值 The preset congestion index threshold is defined, and k is the sensitivity coefficient; assuming threshold C = 7.0, k = 1.0 (default value). For example, when the congestion index C on the Guangzhou-Shenzhen Expressway is... 拥堵 When the congestion level is 8.5 (severe), ω1 = 0.4, ω2 = 0.4, and ω3 = 0.2, video surveillance data should be used first.

[0089] In this embodiment, the DPR algorithm enhancement calibration method includes: for missed samples of small targets (>50 meters) at a distance in video surveillance, image reconstruction is performed using a diffusion model (CDM), and the loss function is defined as: Wherein, λ1 is a pixel-level error constraint function with a value of 0.7, which is used to constrain the pixel-level error between the reconstructed image and the real image; To reconstruct the image; I GT The image is the real image; λ2 is 0.3, used to optimize the intersection-union ratio between the target bounding box and the real bounding box; B represents the target detection bounding box. GT This is a true bounding box.

[0090] In this embodiment, after obtaining the dynamic passenger load factor Y, the parameter optimization module is used to optimize the total cross-regional flow, the global origin-destination matrix, and high-flow paths. Specifically, the optimization methods include the following:

[0091] In this embodiment, the adjustment of the dynamic passenger load factor Y adopts a spatiotemporal weighted average optimization method: Among them, Y final The optimized passenger capacity coefficient; ω4, ω5, and ω6 are weighting coefficients, and the congestion index (C) is also included. 拥堵 ) Dynamic adjustment, such as when there is severe congestion, ω5 increases.

[0092] Specifically, the global origin-endpoint matrix YYOD ij The correction is as follows: Obtain the original global origin-end point matrix. Real-time road network speed data v ij Optimization is achieved by introducing a velocity decay factor: in, Here is the optimized global origin-end point matrix; k1 is the velocity attenuation factor; v 阈值 The speed is 60 km / h (when the speed in congested sections is below the threshold, the OD flow weight is reduced).

[0093] In this embodiment, the high-traffic path weight update method includes: obtaining the path weight W output by the channel identification model. path Real-time GPS trajectory density ρ GPS And it adopts dynamic adjustment based on reinforcement learning: in, The updated and optimized path weights are defined by η, which is the learning rate, taking a value of 0.5; ρ 基准 The preset trajectory density threshold is set to 100 vehicles / minute (the path weight is enhanced when the traffic exceeds the baseline).

[0094] The above parameter optimization process is configured with constraints: physical constraints include that the total cross-regional flow must not exceed the total population of the region × the average daily travel rate (to prevent overfitting); the passenger load factor Y is limited to the range of [1.0, 4.0] (complying with the passenger capacity limit of cars); and real-time constraints require the optimization algorithm to complete the optimization within <5 seconds, which can be achieved using a quantized gradient descent method, such as AdamW. The dynamic parameter linkage mechanism in this embodiment deeply binds environmental parameters (congestion, weather) with model parameters (passenger load factor, OD matrix), realizing a closed loop of "environmental perception, parameter adaptation, and report output," which is beneficial to improving the accuracy and efficiency of cross-regional population flow prediction.

[0095] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0096] In one embodiment, a cross-regional population flow detection device based on multi-source data is provided, which corresponds to the cross-regional population flow detection method based on multi-source data in the above embodiments.

[0097] The cross-regional population flow detection device based on multi-source data includes a multimodal traffic data acquisition module, a spatiotemporal correlation database construction module, a model construction and calculation module, and a parameter optimization module. Detailed descriptions of each functional module are as follows: The multimodal traffic data acquisition module is used to collect multimodal traffic data and preprocess it; the spatiotemporal correlation database construction module is used to construct a spatiotemporal correlation database based on the preprocessed multimodal traffic data.

[0098] The model building and calculation module is used to build a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on a spatiotemporal correlation database; it is also used to calculate the total cross-regional flow based on the passenger volume model, which includes commercial / non-commercial highway travel items; it generates a global origin-destination matrix using the highway passenger OD model; and it aggregates high-volume paths through the passenger corridor identification model.

[0099] The parameter optimization module is used to calculate the dynamically calibrated passenger load factor and combine it with the dynamic calibration passenger load factor to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths in order to determine the cross-regional personnel flow report.

[0100] For specific limitations regarding the cross-regional population flow detection device based on multi-source data, please refer to the limitations of the cross-regional population flow detection method based on multi-source data mentioned above, which will not be repeated here. Each module in the aforementioned cross-regional population flow detection device based on multi-source data can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in the processor of the computer device in hardware form or independent of it, or it can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0101] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 2As shown, the computer device includes a processor, memory, network interface, and database connected via a device bus. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores operating devices, computer programs, and the database. The internal memory provides an environment for the operation of the operating devices and computer programs stored in the non-volatile storage medium. The database stores multimodal traffic data, cross-regional population flow reports, etc. The network interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for detecting cross-regional population flow based on multi-source data.

[0102] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:

[0103] S1: Collect and preprocess multimodal traffic data, and construct a spatiotemporal correlation database based on the preprocessed multimodal traffic data; S2: Construct a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on the spatiotemporal correlation database; calculate the total cross-regional flow based on the passenger volume model, including commercial / non-commercial highway travel items; generate a global origin-destination matrix using the highway passenger OD model; and aggregate high-volume paths using the passenger corridor identification model;

[0104] S3: Calculate the dynamic calibration passenger load factor, and combine it with the parameter optimization module to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report.

[0105] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0106] S1: Collect and preprocess multimodal traffic data, and construct a spatiotemporal correlation database based on the preprocessed multimodal traffic data; S2: Construct a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on the spatiotemporal correlation database; calculate the total cross-regional flow based on the passenger volume model, including commercial / non-commercial highway travel items; generate a global origin-destination matrix using the highway passenger OD model; and aggregate high-volume paths using the passenger corridor identification model;

[0107] S3: Calculate the dynamic calibration passenger load factor, and combine it with the parameter optimization module to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report.

[0108] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0109] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0110] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for detecting cross-regional population flow based on multi-source data, characterized in that, include: Collect and preprocess multimodal traffic data, and construct a spatiotemporal correlation database based on the preprocessed multimodal traffic data; Based on the aforementioned spatiotemporal correlation database, a passenger volume model, a highway passenger OD model, and a passenger corridor identification model are constructed. The total passenger volume model is used to calculate the total cross-regional flow, including commercial and non-commercial highway travel items. The highway passenger OD model is used to generate a global origin-destination matrix. High-volume paths are aggregated through the passenger corridor identification model. Cross-regional passenger flow = ∑(commercial road passenger volume + non-commercial road travel volume) + railway passenger volume + civil aviation passenger volume + waterway passenger volume; Calculate the dynamic calibration passenger load factor, and combine it with the parameter optimization module to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report; The calibration method for the dynamically calibrated passenger load factor includes a multiple linear regression analysis prediction method with fused correction factors, comprising: in, Y is the passenger load factor during the Spring Festival travel rush; Y is the passenger load factor for a specified time period. X1 is the adjustment coefficient for a specified time period, which refers to the ratio of the volume of commercial passenger traffic to the volume of commercial passenger traffic during the Spring Festival travel season; X1, X2, and X3 are independent variables, where X1 refers to the parameter of commercial passenger traffic, X2 refers to the parameter of Class I passenger vehicles on the expressway network, and X3 refers to the parameter of traffic flow of small and medium-sized passenger vehicles on ordinary national and provincial highways. For constant terms; is the regression coefficient, where The regression coefficients for operational passenger volume parameters; For the error term, The regression coefficients are the parameters of a type of passenger vehicle in the highway network. The regression coefficients are for the traffic flow parameters of small and medium-sized passenger vehicles on ordinary national and provincial highways. This is a correction factor.

2. The method for detecting cross-regional population flow based on multi-source data according to claim 1, characterized in that, The multimodal traffic data includes ticketing data from networked ticketing devices, vehicle traffic flow data from highway toll collection devices, vehicle cross-sectional flow data from traffic control stations on ordinary national and provincial highways, video surveillance data, mobile phone signaling data, and ticketing and flight schedule data from railways and civil aviation. Preprocessing of multimodal traffic data includes data cleaning, deduplication, spatiotemporal alignment, and standardization. The calibration method for dynamically calibrating the passenger load factor also includes a motorized travel estimation method based on mobile phone signaling, a sampling statistics method based on service area video surveillance, and a long-distance small target detection method optimized based on the DPR algorithm.

3. The method for detecting cross-regional population flow based on multi-source data according to claim 2, characterized in that, The mobile travel estimation algorithm based on mobile phone signaling includes: Passenger load factor = Among them, private car travel volume = total motorized travel volume - urban public transport passenger volume - intra-county travel volume; total motorized travel volume = permanent resident population × average number of motorized trips per person.

4. The method for detecting cross-regional population flow based on multi-source data according to claim 1, characterized in that, include: In the aforementioned passenger volume model, the non-commercial road travel volume is the sum of passenger car travel volume on expressways and passenger car travel volume on ordinary national and provincial highways. The passenger car travel volume on ordinary national and provincial highways is calculated using the following formula: Passenger car trips on ordinary national and provincial highways = ; in, = Average traffic volume of passenger cars on ordinary national and provincial highways × Mileage of ordinary national and provincial highways; Urban passenger volume = Urban passenger volume of public buses and trolleybuses + Urban rail transit passenger volume + Urban passenger volume of taxis + Ferry passenger volume.

5. The method for detecting cross-regional population flow based on multi-source data according to claim 4, characterized in that, The number of passenger car trips on the expressway is calculated using the following formula: Highway passenger car travel volume = ;in, .

6. The method for detecting cross-regional population flow based on multi-source data according to claim 1 or 4, characterized in that, The process of generating a global origin-destination matrix using the highway passenger transport OD model includes: ,in , ;in, The OD (Original Directed Traffic) and non-operating passenger traffic are related to each pair of commercial highway passenger traffic volumes in multiple cities; This refers to commercial highway passenger traffic volume; , represents non-commercial highway passenger traffic; i is the pairwise correlation coefficient between cities, ranging from 1 to 144; This is the expansion coefficient; This refers to individual OD information generated through online ticketing coefficient analysis; This refers to individual OD information generated from highway toll data analysis.

7. A cross-regional population flow detection device based on multi-source data, characterized in that, The device includes: A multimodal traffic data acquisition module is used to acquire multimodal traffic data and preprocess the multimodal traffic data; The spatiotemporal correlation database construction module is used to build a spatiotemporal correlation database based on preprocessed multimodal traffic data; The model building and calculation module is used to construct a passenger volume model, a highway passenger OD model, and a passenger corridor identification model based on the spatiotemporal correlation database; it is also used to calculate the total cross-regional flow based on the total passenger volume model, which includes commercial / non-commercial highway travel items; it uses the highway passenger OD model to generate a global origin-destination matrix; and it aggregates high-volume paths through the passenger corridor identification model; the cross-regional passenger flow = ∑(commercial highway passenger volume + non-commercial highway travel volume) + railway passenger volume + civil aviation passenger volume + waterway passenger volume; The parameter optimization module is used to calculate the dynamically calibrated passenger load factor and combine the dynamically calibrated passenger load factor to optimize the parameters of the total cross-regional flow, the global origin-destination matrix, and high-volume paths to determine the cross-regional personnel flow report. The calibration method for the dynamically calibrated passenger load factor includes a multiple linear regression analysis prediction method with fused correction factors, comprising: in, Y is the passenger load factor during the Spring Festival travel rush; Y is the passenger load factor for a specified time period. X1 is the adjustment coefficient for a specified time period, which refers to the ratio of the volume of commercial passenger traffic to the volume of commercial passenger traffic during the Spring Festival travel season; X1, X2, and X3 are independent variables, where X1 refers to the parameter of commercial passenger traffic, X2 refers to the parameter of Class I passenger vehicles on the expressway network, and X3 refers to the parameter of traffic flow of small and medium-sized passenger vehicles on ordinary national and provincial highways. For constant terms; is the regression coefficient, where The regression coefficients for operational passenger volume parameters; For the error term, The regression coefficients are the parameters of a type of passenger vehicle in the highway network. The regression coefficients are for the traffic flow parameters of small and medium-sized passenger vehicles on ordinary national and provincial highways; This is a correction factor.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the cross-regional population flow detection method based on multi-source data as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the cross-regional population flow detection method based on multi-source data as described in any one of claims 1 to 6.

Images