An intersection queue length estimation method based on sparse networked vehicle trajectory data

Abstract

The application discloses an intersection queuing length estimation method based on sparse networked vehicle trajectory data, which comprises the following steps: step 1, constructing a speed matrix under a red light state, a speed matrix under a green light state and a standard deviation matrix of speed by using sparse networked vehicle trajectory data; step 2, obtaining a possible influence range matrix of queuing vehicles by using the speed matrix under the red light state, the speed matrix under the green light state and the standard deviation matrix of speed; step 3, obtaining an actual influence range matrix of queuing vehicles by using an integer programming model and constraint conditions based on the obtained possible influence range matrix of queuing vehicles, so that the queuing length of vehicles in a signal cycle is obtained; and step 4, repeating steps 1 to 3 to dynamically obtain the queuing length of vehicles in each signal cycle. The application can effectively deduce the queuing length in the intersection by using limited sparse networked vehicle trajectory data and an integer programming model and constraint conditions.

Description

Technical Field

[0001] This invention belongs to the field of traffic information estimation technology, specifically relating to a method for estimating the queue length at intersections based on sparse connected vehicle trajectory data. Background Technology

[0002] With the continuous increase in the number of vehicles in cities, urban traffic congestion is becoming increasingly severe. Surveys show that 17%-35% of residents' commuting time is caused by vehicle delays at signalized intersections. Queue length at intersections can promptly reflect the congestion situation and is an important indicator for evaluating the efficiency of signalized intersections. Furthermore, queue length information at intersections is crucial for traffic signal optimization, travel time estimation, and traffic management.

[0003] With the improvement of detection methods in recent years, queue length estimation methods, as one of the important parameters of traffic information, have also been continuously developed. Currently, there are three common queue length estimation methods. The first is based on traffic sensors such as loop detectors. These sensors are buried a certain distance behind the stop line in the approach lane, and the queue length is estimated by combining traffic flow density with traffic flow theory. Its disadvantages include high installation and maintenance costs, limited coverage, and the inability to detect vehicles exceeding the detector's range if the queue length at the intersection is too long. Furthermore, this method cannot describe the spatial distribution of vehicles in the queue. The second method utilizes video surveillance data, estimating queue length through image-based or video-based methods. Image-based methods extract information such as license plate numbers and timestamps from the video to infer vehicle travel time and then estimate the queue length. This method still uses static information, and image recognition is time-consuming and labor-intensive, with accuracy affected by video quality. Video-based methods analyze the position of stationary vehicles by identifying and tracking them to determine the queue length in the video. However, this method is limited by the camera's field of view; it becomes ineffective when the queue exceeds the camera's monitoring range. Recent research has proposed an improved video-based approach, utilizing vehicle information from surveillance videos to reconstruct vehicle trajectories outside the video's range and estimate the evolution of queue lengths beyond the monitored area. However, this method relies on neural network models for training, which are complex, labor-intensive, and require high-quality and abundant video data. Furthermore, it demands specific hardware support, leading to high costs. A third approach combines video and GPS trajectory data, employing multi-data fusion and a random forest regression model to obtain real-time queue length estimates. This method offers high accuracy, but requires diverse data sources. Firstly, it necessitates the fusion of multiple data sources; secondly, the training process for the random forest regression model is complex.

[0004] The main drawbacks of existing technologies are: (1) the coverage of traffic sensors and surveillance videos is limited, making it impossible to estimate the length of queued vehicles exceeding the coverage area; (2) the novel methods using neural networks and random forest regression models require massive amounts of data for training, otherwise the results are poor. Therefore, it is necessary to propose a dynamic estimation method for queue length that is low-cost and easy to implement, so as to provide timely feedback on the congestion situation at intersections and assist traffic management departments in optimizing traffic signals and managing traffic flow. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a method for estimating the queue length at intersections based on sparse connected vehicle trajectory data. Based on sparse connected vehicle trajectory data within the intersection area, an integer programming model is used to effectively infer the queue length within the intersection.

[0006] The technical solution provided by this invention is as follows:

[0007] A method for estimating intersection queue length based on sparse connected vehicle trajectory data includes the following steps:

[0008] Step 1: Construct a speed matrix for intersection segments under red light conditions using sparse connected vehicle trajectory data. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity ;

[0009] Step 2: Utilize the constructed speed matrix of the intersection segment under red light conditions. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity Obtain the matrix of potential impact ranges of queued vehicles. ;

[0010] Step 3: Based on the obtained matrix of potential impact ranges of queuing vehicles Using an integer programming model and constraints, the matrix of the actual impact range of queuing vehicles is obtained. This allows us to obtain the vehicle queue length within the signal cycle. ;

[0011] Step 4: Repeat steps 1 to 3 to dynamically obtain the vehicle queue length within each signal cycle.

[0012] According to the above technical solution, step 1 specifically includes:

[0013] Step 11: For a given intersection approach lane, divide it from upstream to the stop line into... Each of the following equal road sections is: Each segment is [length] Meters; the red phase of the traffic light corresponding to this road segment is divided into... Each of the following is an equal time interval: ;

[0014] Step 12, use Indicates road segment Time interval The speed, this value belongs to the road segment Time interval The average speed of all connected vehicle trajectory points is used to obtain the speed matrix under red light conditions. As shown in equation (1):

[0015] (1)

[0016] Step 13: Similarly, obtain the speed matrix of this road segment under green light conditions. The standard deviation matrix of the velocity As shown in equations (2) and (3) below:

[0017] (2)

[0018] (3)

[0019] In the formula, Road section under green light status Time interval The average speed of all connected vehicle trajectory points. Road section under green light status Time interval The standard deviation of the average speed of all connected vehicle trajectory points.

[0020] According to the above technical solution, step 2 specifically includes:

[0021] Step 21: Construct a matrix of the potential impact range of queuing vehicles. As shown in equation (4):

[0022] (4)

[0023] In the formula, For threshold parameters; if This indicates the red light status. Compared to the green light status Much smaller, then that is, road section In time interval There are vehicles queuing in the middle.

[0024] According to the above technical solution, step 2 also includes:

[0025] Step 22: Matrix the potential impact range of queuing vehicles Visualization.

[0026] According to the above technical solution, the threshold parameter The speed limit is set according to the design speed of the road section.

[0027] According to the above technical solution, step 3 specifically includes:

[0028] Step 31: Assume the matrix of the actual impact range of the queuing vehicles is as follows. As shown in formula (5):

[0029] (5)

[0030] Step 32: Using the following integer programming model (6) and constraints (7) and (8), obtain the following... :

[0031] (6)

[0032] (7)

[0033] (8)

[0034] By minimizing the objective function (6), we obtain: when hour, ;when hour, ;

[0035] Step 34: Based on the matrix of actual impact range of queued vehicles Find the maximum queue length As shown in formula (9):

[0036] (9)

[0037] In the formula, j In the matrix of actual impact range of queuing vehicles The smallest j value.

[0038] According to the above technical solution, step 3 also includes:

[0039] Step 33: Matrix the actual impact range of queuing vehicles Visualization.

[0040] According to the above technical solution, the trajectory data of connected vehicles includes GPS trajectory data of connected vehicles.

[0041] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0042] This invention can utilize limited sparse GPS trajectory data of connected vehicles to construct speed matrices for red and green light states, and use an integer programming model and constraints to fully construct the actual influence range of queuing vehicles, thereby obtaining the queue length in each signal cycle. This method does not require data for all vehicles, and is scientific, effective, and highly practical. Attached Figure Description

[0043] Figure 1 This is a flowchart of the intersection queue length estimation method based on sparse connected vehicle trajectory data in this invention;

[0044] Figure 2 This is a schematic diagram of road segment division in this invention;

[0045] Figure 3 This is the matrix of the potential impact range of queuing vehicles in this invention. Visualization chart;

[0046] Figure 4 This is the matrix of the actual impact range of queuing vehicles in this invention. Visualization chart;

[0047] Figure 5 This is the intersection diagram used in the example of the present invention;

[0048] Figure 6 This is a visualization of the potential impact range matrix of queuing vehicles in an example of the present invention;

[0049] Figure 7 This is a visualization of the actual impact range matrix of queuing vehicles in an example of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0051] This invention proposes a dynamic estimation method for intersection queue length based on sparse connected vehicle GPS trajectory data. This method innovatively proposes to use sparse connected vehicle GPS trajectory data within the intersection range and utilizes an integer programming model to effectively infer the queue length within the intersection.

[0052] The present invention provides a dynamic estimation method for intersection queue length based on sparse connected vehicle GPS trajectory data, such as... Figure 1 As shown, it includes the following steps:

[0053] Step 1: Construct a speed matrix for intersection segments under red light conditions using sparse connected vehicle trajectory data. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity ;

[0054] Step 2: Use the velocity matrix to obtain the potential impact range matrix of the queuing vehicles. ;

[0055] Step 3: Based on the queuing vehicle potential impact range matrix obtained in Step 2 Using an integer programming model and constraints, the matrix of the actual impact range of queuing vehicles is obtained. This allows us to obtain the vehicle queue length within this signal period. ;

[0056] Step 4: Repeat steps 1-3 repeatedly to dynamically obtain the vehicle queue length within each signal cycle.

[0057] According to the above technical solution, in step 1, a speed matrix of the intersection segment is constructed using sparse connected vehicle trajectory data. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity The specific steps include:

[0058] Step 11, as follows Figure 2 As shown, for a given intersection approach lane, it is divided from upstream to the stop line into... Each of the following equal road sections is: Each segment is [length] Meters. The red phase of the traffic light corresponding to this road segment is divided into... Each of the following is an equal time interval: ;

[0059] Step 12, use Indicates road segment Time interval The speed, this value belongs to the road segment Time interval The average speed of all GPS track points is used to obtain the speed matrix under red light conditions. As shown in equation (1):

[0060] (1)

[0061] Step 13: To eliminate the influence of other factors on the queue length at the intersection, it is necessary to obtain the speed matrix of this road segment under green light conditions. The standard deviation matrix of the velocity As shown in equations (2) and (3) below:

[0062] (2)

[0063] (3)

[0064] According to the above technical solution, in step 2, the potential impact range matrix of queuing vehicles is obtained using the velocity matrix. The specific steps include:

[0065] Step 21: Construct a matrix of the potential impact range of queuing vehicles. As shown in formula (4):

[0066] (4)

[0067] In the formula, This is a threshold parameter, set according to the design speed of the road segment. If This indicates the red light status. Compared to the green light status Much smaller, then Explain the road section In time interval There are vehicles queuing in the middle.

[0068] Step 22: Matrix the potential impact range of queuing vehicles Visualization, such as Figure 3 As shown.

[0069] According to the above technical solution, in step 3, the potential impact range matrix of queuing vehicles obtained in step 2 is used. Using an integer programming model and constraints, the matrix of the actual impact range of queuing vehicles is obtained. This allows us to obtain the length of the vehicle queue. The specific steps include:

[0070] Step 31: Assume the matrix of the actual impact range of the queuing vehicles is as follows. As shown in formula (5):

[0071] (5)

[0072] Step 32: Using the following integer programming model (6) and constraints (7) and (8), obtain the following... :

[0073] (6)

[0074] (7)

[0075] (8)

[0076] By minimizing the objective function (6), we can obtain that when hour, ,when hour, And based on experience, the first section of road near the parking line... .

[0077] Step 33: Matrix the actual impact range of queuing vehicles Visualization, such as Figure 4 As shown;

[0078] Step 34: Visualize the actual impact range matrix of queuing vehicles. Find the maximum queue length As shown in formula (9):

[0079] (9)

[0080] Specifically, refer to Figures 5 to 7 This embodiment of the dynamic estimation method for intersection queue length based on sparse connected vehicle GPS trajectory data uses the south entrance of an intersection in a certain city as a specific case to illustrate the dynamic estimation method for queue length of the present invention, including the following steps:

[0081] Step 1: Construct a speed matrix for intersection segments under red light conditions using sparse connected vehicle trajectory data. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity .

[0082] Step 11, as follows Figure 5 As shown, for the south entrance road, it is divided into 20 equal segments from upstream to the parking line. Each segment is 8 meters long. The red phase of the traffic light corresponding to that segment starts at 8:00:00 AM and continues until 8:01:39 AM. This is divided into 11 equal time intervals. ;

[0083] Step 12, use Indicates road segment Time interval The speed, this value belongs to the road segment Time interval The average speed of all GPS track points is used to obtain the speed matrix under red light conditions. As shown below:

[0084]

[0085] Step 13: To eliminate the influence of other factors on the queue length at the intersection, it is necessary to obtain the speed matrix of this road segment under green light conditions. and the standard deviation matrix of velocity As shown below:

[0086]

[0087]

[0088] Step 2: Use the velocity matrix to obtain the potential impact range matrix of the queuing vehicles. .

[0089] Step 21: Construct a matrix of the potential impact range of queuing vehicles. In the formula, Set to 3.2:

[0090]

[0091] Step 22: Matrix the potential impact range of queuing vehicles Visualization, such as Figure 6 As shown.

[0092] Step 3: Based on the queuing vehicle potential impact range matrix obtained in Step 2 Using an integer programming model and constraints, the matrix of the actual impact range of queuing vehicles is obtained. This allows us to obtain the vehicle queue length within this signal period. .

[0093] Step 31: Assume the matrix of the actual impact range of the queuing vehicles is as follows. As shown below:

[0094]

[0095] Step 32: Using the following integer programming model and constraints, obtain the solution. :

[0096]

[0097]

[0098]

[0099] And based on experience, the first section of road near the parking line, .

[0100] Step 33: Matrix the actual impact range of queuing vehicles Visualization, such as Figure 7 As shown;

[0101] Step 34: Visualize the actual impact range matrix of queuing vehicles. Find the maximum queue length :

[0102]

[0103] Step 4: Repeat steps 1-3 repeatedly to dynamically obtain the vehicle queue length within each signal cycle.

[0104] It is evident that the method of this invention can scientifically and reasonably estimate the queue length within a signal period. Compared with other methods, it requires only a single data source, is simple and rapid, has a rigorous and scientific model, and has significant practical application value.

[0105] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.

[0106] Those skilled in the art will readily understand that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for estimating the queue length at intersections based on sparse connected vehicle trajectory data, characterized in that, Includes the following steps: Step 1: Construct a speed matrix for intersection segments under red light conditions using sparse connected vehicle trajectory data. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity Specifically, it includes: Step 11: For a given intersection approach lane, divide it from upstream to the stop line into... Each of the following equal road sections is: Each segment is [length] Meters; the red phase of the traffic light corresponding to this road segment is divided into... Each of the following is an equal time interval: ; Step 12, use Indicates road segment Time interval The speed, this value belongs to the road segment Time interval The average speed of all connected vehicle trajectory points is used to obtain the speed matrix under red light conditions. As shown in equation (1): (1) Step 13: Similarly, obtain the speed matrix of this road segment under green light conditions. The standard deviation matrix of the velocity As shown in equations (2) and (3) below: (2) (3) In the formula, Road section in green light state Time interval The average speed of all connected vehicle trajectory points. Road section under green light status Time interval The standard deviation of the average speed of all connected vehicle trajectory points; Step 2: Utilize the constructed speed matrix of the intersection segment under red light conditions. Speed ​​matrix under green light conditions The standard deviation matrix of the velocity Obtain the matrix of potential impact ranges of queued vehicles. Specifically, this includes: Step 21: Construct a matrix of the potential impact range of queuing vehicles. As shown in equation (4): (4) In the formula, For threshold parameters; if This indicates the red light status. Compared to the green light status Much smaller, then that is, road section In time interval There are vehicles queuing in the middle; Step 3: Based on the obtained matrix of potential impact ranges of queuing vehicles Using an integer programming model and constraints, the matrix of the actual impact range of queuing vehicles is obtained. This allows us to obtain the vehicle queue length within the signal cycle. Specifically, it includes: Step 31: Assume the matrix of the actual impact range of the queuing vehicles is as follows. As shown in formula (5): (5) Step 32: Using the following integer programming model (6) and constraints (7) and (8), obtain the following... : (6) (7) (8) By minimizing the objective function (6), we obtain: when hour, ;when hour, ; Step 34: Based on the matrix of actual impact range of queuing vehicles Find the maximum queue length As shown in formula (9): (9) In the formula, j In the matrix of actual impact range of queuing vehicles The smallest j value; Step 4: Repeat steps 1 to 3 to dynamically obtain the vehicle queue length within each signal cycle.

2. The intersection queue length estimation method based on sparse connected vehicle trajectory data according to claim 1, characterized in that, Step 2 also includes: Step 22: Matrix the potential impact range of queuing vehicles Visualization.

3. The intersection queue length estimation method based on sparse connected vehicle trajectory data according to claim 1 or 2, characterized in that, Threshold parameter The speed limit is set according to the design speed of the road section.

4. The intersection queue length estimation method based on sparse connected vehicle trajectory data according to claim 1, characterized in that, Step 3 also includes: Step 33: Matrix the actual impact range of queuing vehicles Visualization.

5. The intersection queue length estimation method based on sparse connected vehicle trajectory data according to claim 1, characterized in that, Connected vehicle trajectory data includes connected vehicle GPS trajectory data.

Images