A method to improve the resolution of traffic space-time graph

Abstract

The present invention discloses a method for improving the resolution of a traffic space-time map. The method comprises: collecting data and drawing an original traffic space-time map; dividing the space-time map into grids, projecting the collected data according to the grids, and calculating the average speed of each grid; further refining the grids and constructing a fine-grained grid speed prediction model; using the fine-grained grid speed prediction model to perform predictions and generate a high-resolution traffic space-time map. It can be seen that the present application divides the original "rough" traffic space-time map into grids, completes the sub-grid speed prediction using a fine-grained grid speed prediction model, and reproduces the "beautified" refined traffic space-time map, thereby achieving the function of improving the resolution of the traffic space-time map. In this way, more accurate basic analysis data for traffic status analysis and traffic event identification can be provided to traffic managers, supporting the formulation of traffic decisions.

Description

Technical Field

[0001] The invention relates to the field of traffic technology, and in particular to a method for improving the resolution of a traffic spatiotemporal graph. Background Art

[0002] Traffic spatiotemporal graph is one of the most important analysis and visualization tools in the field of transportation. The key to drawing a traffic spatiotemporal graph is to measure the speed corresponding to each spatiotemporal grid. The smaller the grid size of the spatiotemporal graph and the higher the resolution, the more detailed road traffic changes can be displayed. Ideally, the speed of the grid can be accurately calculated by using 100% high-fidelity vehicle trajectories. However, current information technology cannot support the collection of such a wide range of 100% high-fidelity vehicle trajectory data, and thus cannot support the acquisition of refined and high-resolution traffic spatiotemporal graphs. In addition, in real-time traffic systems, 5 minutes is a commonly used information release time interval. Therefore, in order to obtain detailed traffic speeds in a specific spatiotemporal grid, it is usually necessary to supplement daily sampling data with traffic state estimation.

[0003] By searching existing patents and papers, it is found that most studies use deep learning methods to improve the resolution of imaging systems, and there are few methods to improve the resolution of traffic spatiotemporal graphs. The Chinese patent document "A method for improving underwater image resolution based on an improved generative adversarial network" (CN114037843A, 2022-02-11) constructs an improved generative adversarial network model by improving the structure of the generator module and introducing a dense jump connection structure, but the process calculation cost is very high, which hinders the real-time application of this method; the Chinese patent document "A method and system for real-time reconstruction of spatiotemporal super-resolution" (CN114170085A, 2022-03-11) extracts features from the input frame, compensates for the feature map offset field, and fuses features to generate super-resolution reconstructed images, but this process requires a large number of images for training. In the field of transportation, high-fidelity and high-sampling rate traffic data are difficult to collect, so the above methods are difficult to implement. Therefore, the present invention provides a method for improving the resolution of traffic spatiotemporal graphs, which does not require a large number of high-resolution spatiotemporal graphs for model training, but only requires sparse trajectory data. The method is simple and practical, and the computational complexity is low. Summary of the invention

[0004] In view of the widespread problem that low-resolution spatiotemporal graphs cannot display detailed traffic dynamics, the present invention provides a method for improving the resolution of traffic spatiotemporal graphs. By gridding the original "rough" traffic spatiotemporal graph and using a fine-grained grid speed prediction model to complete sub-grid speed prediction, the "beautified" refined traffic spatiotemporal graph can be reproduced to achieve the function of improving the resolution of the traffic spatiotemporal graph. High-resolution spatiotemporal graphs can show more traffic details, help identify traffic bottlenecks, understand traffic characteristics, predict travel times, and estimate traffic emissions, and have broad application prospects in the field of transportation.

[0005] To solve the above technical problems, the present invention provides a method for improving the resolution of traffic spatiotemporal graphs, which can reproduce refined traffic spatiotemporal graphs and help provide more traffic information for identifying traffic bottlenecks, understanding traffic characteristics, predicting travel time, and estimating traffic emissions. In order to achieve the above objectives, the technical solutions provided in the embodiments of the present application are as follows:

[0006] Step 1: Collect data and draw the original traffic time-space diagram;

[0007] Step 2: Grid the original traffic space-time map, project the collected data into grids, and calculate the average speed of each grid;

[0008] Step 3: Further refine the grid division and build a fine-grained grid velocity prediction model;

[0009] Step 4: Use the fine-grained grid speed prediction model to generate a high-resolution traffic spatiotemporal map.

[0010] Furthermore, in step 1, the collected data includes the vehicle's longitude, latitude, speed, lane, and time; the original traffic space-time diagram is a vehicle trajectory-speed space-time diagram, in which the x-axis represents time, the y-axis represents space, and the color in the diagram represents the average speed.

[0011] Furthermore, in step 2, the shape of the grid division is uniform square i, with sizes of 15s×25m, 30s×50m, 60s×100m, 120s×200m and 240s×400m respectively; the 8 neighboring cells of the central grid i are UL (upper left), Up (upper), UR (upper right), Lf (left), Rt (right), LL (lower left), Lw (lower), and LR (lower right).

[0012] Furthermore, in step 2, projecting the collected data into grids and calculating the average speed of each grid means projecting according to the longitude, latitude and time coordinates of each vehicle and the grid range coordinates; the calculation method of the average speed of the grid is: Where x is the average speed of a grid, in km / h; n is the number of trajectory points in a grid, in units; v p represents the instantaneous speed of the pth trajectory point in km / h; the average speed of grid i is represented by x i .

[0013] Furthermore, in step 3, further refining the grid division means further dividing the single grid i into four equal-sized square sub-grids j, j∈J={LL,LR,UR,UL}.

[0014] Furthermore, in step 3, constructing a fine-grained grid speed prediction model means using the average speed x of multiple grids surrounding the small grid j to predict the speed of the subgrid j.

[0015] Furthermore, in step 3, constructing a fine-grained grid speed prediction model requires constructing model training samples, which are divided into free-flow samples and congestion samples according to their traffic status; the traffic status k is determined by discriminating the speed value of grid i, and the speed discrimination threshold is selected as 60km / h; when the average speed of grid i is greater than 60km / h, it is free flow (ff), otherwise it is congested (cg).

[0016] Furthermore, in step 3, a fine-grained grid velocity prediction model is constructed by using a multivariate linear regression method, and the model parameters p are estimated by using the least squares method. jk , and obtain the constant ε jk The model is in is the estimated speed of subgrid j in grid i, in km / h; p jk are model parameters, in, are the regression constants of grid i, LL, LW, LR, Rt, UR, Up, UL, and Lf respectively; x is the average speed of the grid in km / h, x=[x i ,x LL ,x Lw ,x LR ,x Rt ,x UR ,x Up ,x UL ,x Lf ], where x i 、x LL 、x Lw 、x LR 、x Rt 、x UR 、x Up 、x UL 、x LfRespectively represent the average speed of grid i, LL, LW, LR, Rt, UR, Up, UL, Lf; ε jk is a partial regression constant, where k∈{ff,cg}.

[0017] Furthermore, in step 4, using the fine-grained grid speed prediction model for prediction means inputting the original traffic spatiotemporal graph, dividing the grids according to the optimal model, and calculating the average speed x of grid i. i And judge the size. If x i >60km / h, the free-flow model is used to predict the average speed of the subgrid. i If the speed is less than or equal to 60 km / h, the model under congestion conditions is used to predict the average speed of the subgrid.

[0018] Furthermore, in step 4, generating a high-resolution traffic spatiotemporal map means that steps 1-4 can be repeatedly used until a traffic spatiotemporal map of a desired resolution is obtained.

[0019] The present invention provides a method for improving the resolution of a traffic spatiotemporal graph, which has the following advantages:

[0020] 1. The calculation process is simple, which can quickly and effectively improve the resolution of the rough space-time map and show more traffic details;

[0021] 2. It is highly practical and can be widely used in the spatiotemporal analysis of daily collected data. The refined spatiotemporal map helps to identify traffic bottlenecks, understand traffic characteristics, predict travel time and estimate traffic emissions. It has broad application prospects in the field of transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 A flowchart of a method for improving the resolution of a traffic spatiotemporal graph provided in an embodiment of the present application.

[0024] Figure 2 A schematic diagram of grid division provided in an embodiment of the present application.

[0025] Figure 3 A diagram of the traffic space-time graph conversion process provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0027] See also Figure 1 , which is a flow chart of a method for improving the resolution of a traffic spatiotemporal graph provided in an embodiment of the present application.

[0028] like Figure 1 As shown, a method for improving the resolution of a traffic spatiotemporal graph provided in an embodiment of the present application includes the following steps:

[0029] Step 1: Collect data and draw the original traffic space-time diagram.

[0030] Step 2: Grid the original traffic space-time map, project the collected data according to the grid, and calculate the average speed of each grid.

[0031] Step 3: Further refine the grid division and construct a fine-grained grid velocity prediction model.

[0032] Step 4: Use the fine-grained grid speed prediction model to generate a high-resolution traffic spatiotemporal map.

[0033] As a possible implementation method, the collected data provided in the embodiment of the present application includes the vehicle's longitude, latitude, speed, lane, and time; the original traffic space-time diagram is a vehicle trajectory-speed space-time diagram, in which the x-axis represents time, the y-axis represents space, and the average speed is represented by color in the diagram.

[0034] As a possible implementation manner, the grid division provided in the embodiment of the present application is in the shape of uniformly sized squares i, with sizes of 15s×25m, 30s×50m, 60s×100m, 120s×200m and 240s×400m respectively; the eight neighboring cells of the center grid i are UL (upper left), Up (upper), UR (upper right), Lf (left), Rt (right), LL (lower left), Lw (lower), and LR (lower right).

[0035] As a possible implementation mode, the method of projecting the collected data into grids and calculating the average speed of each grid provided in the embodiment of the present application refers to performing projection according to the longitude, latitude and time coordinates of each vehicle and the grid range coordinates; the calculation method of the average speed of the grid is: Where x is the average speed of a grid, in km / h; n is the number of trajectory points in a grid, in units; v p represents the instantaneous speed of the pth trajectory point in km / h; the average speed of grid i is represented by x i .

[0036] As a possible implementation manner, the further refined grid division provided in the embodiment of the present application refers to further dividing a single grid i into four equal-sized square sub-grids j, j∈J={LL,LR,UR,UL}.

[0037] As a possible implementation method, the construction of a fine-grained grid speed prediction model provided in the embodiment of the present application refers to using the average speed x of multiple grids surrounding a small grid j to predict the speed of a subgrid j.

[0038] As a possible implementation method, the construction of a fine-grained grid speed prediction model provided in the embodiment of the present application requires the construction of model training samples, which are divided into free flow samples and congestion samples according to their traffic status; the traffic state k is determined by distinguishing the speed value of grid i, and the speed distinction threshold is selected as 60km / h; when the average speed of grid i is greater than 60km / h, it is free flow (ff), otherwise it is congested (cg).

[0039] As a possible implementation method, the fine-grained grid velocity prediction model provided in the embodiment of the present application adopts a multivariate linear regression method to construct the model, and the least squares method is used to estimate the model parameters p jk , and obtain the constant ε jk The model is in is the estimated speed of subgrid j in grid i, in km / h; p jk are model parameters, in, are the regression constants of grid i, LL, LW, LR, Rt, UR, Up, UL, and Lf respectively; x is the average speed of the grid in km / h, x=[x i ,x LL ,x Lw ,x LR ,x Rt ,x UR ,x Up ,x UL ,x Lf ], where x i 、x LL 、x Lw 、x LR 、x Rt 、x UR 、x Up 、x UL 、x Lf Respectively represent the average speed of grid i, LL, LW, LR, Rt, UR, Up, UL, Lf; ε jk is a partial regression constant, where k∈{ff,cg}.

[0040] As a possible implementation method, the prediction using the fine-grained grid speed prediction model provided in the embodiment of the present application refers to inputting the original traffic spatiotemporal graph, dividing the grids according to the optimal model, and calculating the average speed x of the grids. i And judge the size. If x i >60km / h, the free-flow model is used to predict the average speed of the subgrid. i If the speed is less than or equal to 60 km / h, the model under congestion conditions is used to predict the average speed of the subgrid.

[0041] As a possible implementation, the generation of a high-resolution traffic spatiotemporal graph provided in the embodiment of the present application means that steps 1-4 can be repeatedly used until a traffic spatiotemporal graph of a desired resolution is obtained.

[0042] In order to better understand the method for improving the resolution of a traffic spatiotemporal graph provided in the embodiment of the present application, the method is introduced below through several specific embodiments in conjunction with the accompanying drawings.

[0043] See also Figure 2 , which is a schematic diagram of the grid division provided in an embodiment of the present application.

[0044] like Figure 2 As shown, the eight neighboring cells of the central grid i are UL (upper left), Up (upper right), UR (upper right), Lf (left), Rt (right), LL (lower left), Lw (lower right), and LR (lower right); the sub-grids of grid i are further divided into j∈J={LL,LR,UR,UL}.

[0045] See Table 1, which shows the model parameter fitting results provided in the embodiments of the present application.

[0046] As shown in Table 1, it can be seen that all R values ​​of the estimated parameters under congestion conditions 2 are all greater than 0.9; moreover, the R 2 is larger than the parameter under free flow conditions, indicating that the regression method is more suitable for refining the spatiotemporal grid under congested conditions.

[0047] Using the model in Table 1, we can predict the subgrid speed and draw a high-resolution traffic spatiotemporal map, such as Figure 3 shown. Figure 3 The figure shows the traffic space-time graph conversion process of the selected examples 1-4 test (red dashed box) and 1-4-16 test (blue dashed box). The 1-4-16 test refers to refining the space-time graph from a 240s×400m grid to a 60s×100m grid, which means the resolution is increased by 16 times. The 1-4 test refers to refining the space-time graph from a 120s×200m grid to a 60s×100m grid, which means the resolution is increased by 4 times.

[0048] like Figure 3 As shown, a fairly rough spatiotemporal diagram of traffic patterns ( Figure 3 (b1)) can be refined into a spatiotemporal graph, and even detailed stop-and-go waves can be observed ( Figure 3 (b3)). In addition, the flow propagation direction in the low-resolution spatiotemporal image is almost vertical ( Figure 3 (b1)); after refinement, the propagation direction with slope is Figure 3 (b3) and 3(c2), which is more consistent with the real-world traffic pattern ( Figure 3 (a) and 3(d)). Compared with the real image ( Figure 3 (d)) Compared with the refined spatiotemporal graphs in the 1-4 and 1-4-16 tests ( Figure 3 Both (c2) and 3(b3)) satisfactorily reproduce the traffic detail patterns, such as the stop-and-go waves and their wave directions.

[0049] In summary, the embodiment of the present application divides the original "rough" traffic time-space map into grids, uses a fine-grained grid speed prediction model to complete sub-grid speed prediction, reproduces the "beautified" refined traffic time-space map, and realizes the function of improving the resolution of the traffic time-space map. In this way, the method provided by the embodiment of the present application can provide traffic managers with more accurate basic analysis data for traffic status analysis and traffic event discrimination, and support the formulation of traffic decisions.

[0050] It can be known from the description of the above implementation mode that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for enabling a computer device (which can be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0051] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the method disclosed in the embodiment, since it corresponds to the system disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the system part description.

[0052] It should also be noted that, in this article, the term "includes", "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0053] The above description of the disclosed embodiments will enable professionals in the field to implement or use the various modifications of these embodiments to be apparent to professionals in the field, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0054] Table 1

[0055]

Claims

1. A method for improving the resolution of a traffic spatiotemporal graph, characterized in that: The following steps are involved: Step 1: Collect data and draw the original traffic time-space diagram; Step 2: Grid the original traffic space-time map, project the collected data into grids, and calculate the average speed of each grid; Step 3: Further refine the grid division and build a fine-grained grid velocity prediction model; Step 4: Use the fine-grained grid speed prediction model to generate a high-resolution traffic spatiotemporal map; The shape of the grid in step 2 is a square i, with sizes of 15s×25m, 30s×50m, 60s×100m, 120s×200m and 240s×400m respectively; the eight neighboring cells of the center grid i are upper left UL, upper Up, upper right UR, left Lf, right Rt, lower left LL, lower Lw and lower right LR respectively; In step 2, the collected data is projected into grids and the average speed of each grid is calculated, which means that the projection is performed according to the longitude, latitude and time coordinates of each vehicle and the grid range coordinates; the calculation method of the average speed of the grid is: Where x is the average speed of a grid, in km / h; n is the number of trajectory points in a grid, in units; v p represents the instantaneous speed of the pth trajectory point in km / h; the average speed of grid i is represented by x i ; In step 3, the construction of the fine-grained grid speed prediction model requires the construction of model training samples, which are divided into free flow samples and congestion samples according to their traffic status; the traffic status k is determined by judging the speed value of grid i, and the speed judgment threshold is selected as 60km / h; when the average speed of grid i is greater than 60km / h, it is free flow ff, otherwise it is congestion cg; In step 3, a fine-grained grid velocity prediction model is constructed using the multivariate linear regression method, and the least squares method is used to estimate the model parameters p jk , and obtain the constant ε jk The model is in is the estimated speed of subgrid j in grid i, in km / h; p jk are model parameters, in, are the regression constants of grid i, LL, LW, LR, Rt, UR, Up, UL, and Lf respectively; x is the average speed of the grid in km / h, x=[x i ,x LL ,x Lw ,x LR ,x Rt ,x UR ,x Up ,x UL ,x Lf ], where x i 、x LL 、x Lw 、x LR 、x Rt 、x UR 、x Up 、x UL 、x Lf Respectively represent the average speed of grid i, LL, LW, LR, Rt, UR, Up, UL, Lf; ε jk is a partial regression constant, where k∈{ff,cg}; Further refining the grid in step 3 means further dividing the single grid i into four equal-sized square subgrids j, j∈J={LL,LR,UR,UL}; Constructing a fine-grained grid velocity prediction model in step 3 means using the average velocity x of multiple grids surrounding small grid j to predict the velocity of subgrid j 2. The method according to claim 1, characterized in that: The data collected in step 1 include the vehicle's longitude, latitude, speed, lane, and time; the original traffic space-time diagram is a vehicle trajectory-speed space-time diagram, where the x-axis represents time, the y-axis represents space, and the color in the diagram represents the average speed.

3. The method according to claim 1, characterized in that: In step 4, the prediction using the fine-grained grid speed prediction model refers to inputting the original traffic time-space map, dividing the grid according to the optimal model, and calculating the average speed x of the grid. i And judge the size; if x i >60km / h, the model under free flow conditions is used to predict the average speed of the subgrid; if x i If the speed is less than or equal to 60 km / h, the model under congestion conditions is used to predict the average speed of the subgrid.

4. The method according to claim 1, characterized in that: Generating a high-resolution traffic spatiotemporal map in step 4 means that steps 1-4 can be repeatedly used until a traffic spatiotemporal map of a desired resolution is obtained.

Images