Random traffic flow simulation method based on graph space-time neural network

Abstract

The invention relates to a random traffic flow simulation method based on a graph space-time neural network, and the method comprises the following steps: S1, carrying out the correction, detection and vehicle tracking through a bridge deck video collected by a camera, and forming the effective trajectory data of a traffic flow; s2, based on a multi-layer Monte Carlo model and a graph space-time neural network, establishing a departure model for the vehicle to get on the bridge; and S3, establishing a vehicle driving microscopic simulation model based on the graph space-time neural network model. Compared with the prior art, the vehicle flow simulation method conforming to actual speed change and lane change behaviors is realized, and meanwhile, a vehicle speed change-lane change coupled vehicle simulation model is established; the random traffic flow simulation method based on the graph space-time neural network, provided by the invention, has relatively high transportability and important reference significance on vehicle track history analysis of bridges and other roads.

Description

Technical Field

[0001] The present invention relates to a random traffic flow simulation method, in particular to a random traffic flow simulation method based on a graph spatiotemporal neural network. Background Art

[0002] With the acceleration of urbanization, the continuous increase in traffic volume has brought tremendous pressure to infrastructure such as bridges. Bridge design and maintenance require accurate simulation of vehicle loads to evaluate its fatigue reliability.

[0003] Traditional traffic flow simulation methods, such as models based on cellular automata, can simulate the basic dynamics of traffic flow, but often have difficulty in accurately capturing the microscopic behaviors of vehicles, such as speed and lane changes.

[0004] In recent years, deep learning, especially graph neural networks, has shown great potential in traffic flow prediction. Models such as Graph Spatio-Temporal Graph Neural Network (STGCN) can use graph structures to simulate complex interactions in traffic networks and improve the accuracy of prediction.

[0005] Although existing deep learning models have made some progress in traffic prediction, how to apply these models to actual traffic flow simulation, especially in specific scenarios such as bridges, remains a challenge. Summary of the invention

[0006] The purpose of the present invention is to provide a random traffic flow simulation method based on a graph spatiotemporal neural network in order to overcome the defects of the above-mentioned prior art, realize a traffic flow simulation method that conforms to actual speed change and lane change behaviors, and at the same time establish a vehicle simulation model of vehicle speed change-lane change coupling; the proposed random traffic flow simulation method based on a graph spatiotemporal neural network has strong portability and important reference significance for the vehicle trajectory history analysis of bridges and other roads.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] The present invention provides a random traffic flow simulation method based on a graph spatiotemporal neural network, comprising the following steps:

[0009] S1: The bridge video collected by the camera is corrected, detected, and tracked to form effective trajectory data of the vehicle flow;

[0010] S2: Using the effective trajectory data of the vehicle flow formed in S1, a vehicle departure model for going onto the bridge is established based on the multi-layer Monte Carlo model and graph spatiotemporal neural network;

[0011] S3: Using the effective trajectory data of the traffic flow formed in S1 and the vehicle departure model on the bridge established in S2, a vehicle driving micro-simulation model is established based on the graph spatiotemporal neural network model to realize random traffic flow simulation.

[0012] Furthermore, S1 specifically includes the following steps:

[0013] Video rectification: Correct image distortion through photogrammetric projection transformation using ground control points and the global coordinate station;

[0014] Vehicle detection and tracking: The corrected video is processed based on the target detection and target tracking algorithm, and the current timestamp, vehicle number, geometric center position, and vehicle type are recorded. The vehicle speed and vehicle acceleration are calculated, and the acquired data together form the effective trajectory data of the vehicle flow.

[0015] Furthermore, in the process of video correction, the ground control points and the total coordinate station are used to perform photogrammetric projection transformation to correct image distortion, and the original image coordinates are converted into image coordinates of physical space relationship, specifically including:

[0016] Through photogrammetry, the original image is distorted using ground control points (GCPs) and the total coordinate station;

[0017] in:

[0018] Collecting Ground Control Points (GCPs): First, control points need to be collected on the ground at known geographical locations. These points can be clearly identified in the image and their precise coordinates are obtained through ground measurement equipment (such as GPS).

[0019] Distortion Correction: These control points are then used to correct image distortion using a photogrammetric projective transformation. This usually involves the use of specialist software such as ENVI or ArcGIS Pro, which are able to calculate the image's distortion parameters based on the known coordinates of the control points and their location in the image, and apply the appropriate correction algorithm.

[0020] Coordinate transformation: After distortion correction, the original image coordinates are transformed into image coordinates associated with physical space. This process may include mapping pixels in the image coordinate system to the geographic coordinate system to achieve geographic reference of the image.

[0021] Resampling and output: Finally, according to the geometric relationship between the rectified image and the original image, a new rectified image is generated by a resampling method (such as nearest neighbor, bilinear interpolation, or cubic convolution interpolation), and the output is saved.

[0022] Convert the rectified image coordinates into physical space coordinates.

[0023] Furthermore, S2 specifically includes the following steps:

[0024] Using the effective trajectory data of the traffic flow formed in S1, the time interval and headway between vehicles are simulated through a multi-level Monte Carlo model to predict the departure model of vehicles on the bridge.

[0025] A graph spatiotemporal neural network is used to predict vehicle information and abstract the vehicle interaction relationship in the scene into a spatiotemporal undirected graph, where nodes represent vehicles and edges represent the interaction behaviors between vehicles.

[0026] By constructing a multi-layer Monte Carlo model and graph spatiotemporal neural network, inputting vehicle information in the area and inferring the speed and model of newly arrived vehicles, a departure model that can reflect actual traffic conditions is obtained.

[0027] Furthermore, the specific process of simulating the time interval between vehicle arrivals and the headway between vehicles through the multi-level Monte Carlo model includes:

[0028] A multi-level Monte Carlo model is used to simulate the time interval and headway of vehicles arriving at the bridge deck. Statistical methods and random sampling techniques are used to generate a series of possible vehicle arrival times and headways based on historical traffic data, thereby constructing a vehicle departure model that can reflect actual traffic conditions. The departure model can predict the probability distribution and time characteristics of vehicles arriving at the bridge deck under different traffic flows and different time periods, providing accurate starting conditions for subsequent traffic simulation.

[0029] Furthermore, the specific process of abstracting the vehicle interaction relationship in the scene into a spatiotemporal undirected graph includes: after obtaining the distance between the newly arrived vehicles, the graph neural network is used to predict the vehicle information of the target vehicle, and the vehicle interaction relationship in the scene is abstracted into a spatiotemporal undirected graph G(V,E), where V represents different vehicles and the graph edge E represents the interaction behavior between different vehicles, including the spatial edge set E s and the inter-frame edge set E f .

[0030] Furthermore, in the spatiotemporal undirected graph, the graph structure is represented by an adjacency matrix A={A0,A1}, both of which are of size V×V, where A0 is a self-connected unit matrix at the temporal level, and A1 is an adjacency matrix at the spatial level. Whether there is spatial adjacency is determined by whether the Euclidean distance between nodes is less than a set threshold.

[0031] Furthermore, S3 specifically includes the following steps:

[0032] Construction of trajectory prediction model: Using the effective vehicle trajectory data formed in S1 and the vehicle departure model established in S2, a trajectory prediction model based on graph spatiotemporal neural network is constructed;

[0033] Implementation of random traffic flow simulation: The traffic flow data generated by the departure model is used as the initial input and put into the trajectory prediction model. The model predicts the vehicle data in the driving area and merges the vehicle data with the data of newly arrived vehicles to form the input data for the next moment, thus realizing the microscopic simulation of random traffic flow in the target area.

[0034] Furthermore, the trajectory prediction model unifies the vehicle's following and lane-changing behaviors into a trajectory prediction problem, using a network structure consistent with the departure module. The input features of the model include the vehicle's lateral position x, longitudinal position y, speed, acceleration and vehicle type, and the output is the vehicle's lateral position x' and longitudinal position y' at the next moment.

[0035] Furthermore, during the simulation, in the driving area, the vehicles that are still in the driving area in the prediction results are filtered out at each time step, and together with the vehicles that enter the driving area from the departure area in the new frame, they are used as the input of the next prediction model. This process is repeated until the simulation time reaches the specified time length.

[0036] Compared with the prior art, the present invention has the following technical advantages:

[0037] 1) This invention provides a novel random traffic simulation method by combining graph spatiotemporal neural network and multi-level Monte Carlo model, which can more accurately predict vehicle driving behavior and traffic flow pattern. This method generates traffic flow data that conforms to actual traffic conditions by simulating vehicle following and lane changing behaviors, which has important practical value for bridge design verification, traffic planning and road safety maintenance.

[0038] 2) High portability and adaptability. Since the model is built based on actual vehicle trajectory data, it can be easily applied to different bridge and road conditions, providing customized traffic simulation services for various traffic projects. In addition, the method can simulate extreme traffic conditions and provide a more comprehensive safety assessment for engineering design.

[0039] 3) Through refined driving simulation, the present invention not only improves the accuracy of vehicle trajectory prediction, but also enhances the understanding of the dynamic changes of traffic flow. This has a significant promoting effect on optimizing traffic flow, reducing congestion and improving road use efficiency. At the same time, this method provides a powerful tool for traffic management departments to help them better understand traffic flow characteristics and formulate more effective traffic management strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1It is a schematic diagram of the overall process of the random traffic flow simulation method based on graph spatiotemporal neural network;

[0041] Figure 2 This is a schematic diagram of surveillance video acquisition mode;

[0042] Figure 3 This is a schematic diagram of perspective projection;

[0043] Figure 4 for multi-level Monte Carlo models and graph neural networks;

[0044] Figure 5 It is a vehicle driving flow chart;

[0045] Figure 6 This is a schematic diagram of the process of acquiring traffic data;

[0046] Figure 7 The following is an example of simulation results. DETAILED DESCRIPTION

[0047] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms and other features not clearly described in this technical solution are all considered to be common technical features disclosed in the prior art.

[0048] Example 1

[0049] In general, constructing a vehicle load model for bridge fatigue reliability analysis is an important means of bridge design verification and safety maintenance. Problems that need to be solved in vehicle load model construction include traffic simulation, response analysis, model inference, etc., among which random traffic simulation is a key issue. However, traditional traffic micro-models cannot obtain traffic that is more in line with actual speed change and lane change behaviors. The construction of a following model or lane change model based on a neural network alone ignores the connection between speed change and lane change behaviors. For this reason, the present invention proposes a data-driven random traffic simulation method based on a graph spatiotemporal neural network, see Figure 1 , the method of the present invention includes the following steps: 1) correcting, detecting, and tracking vehicles through the bridge deck video collected by the camera to form effective trajectory data of the traffic flow; 2) establishing a vehicle departure model for getting on the bridge based on the multi-layer Monte Carlo model and the graph space-time neural network; 3) establishing a vehicle driving micro-simulation model based on the graph space-time neural network model. The present invention implements a traffic simulation method that conforms to actual speed change and lane change behaviors, and at the same time establishes a vehicle simulation model of vehicle speed change-lane change coupling. In addition, the proposed random traffic simulation method based on the graph space-time neural network has strong portability and important reference significance for the vehicle trajectory history analysis of bridges and other roads.

[0050] For random traffic simulation in a specific range, in order to effectively consider the relationship between vehicles changing lanes and vehicle distribution, the first thing is to obtain the historical data of traffic trajectories within the range. Specifically, several cameras are deployed in the area of ​​concern to shoot vehicles within the bridge deck. The form of monitoring video collection is shown in Figure 2 .

[0051] After obtaining the video, the image distortion is corrected by photogrammetric projection transformation. Specifically, the original image coordinates are converted into image coordinates that can correspond to the physical space relationship using ground control points (GCPs) and the total coordinate station. The conversion form is shown in Figure 3 This transformation process includes correcting distortion and aligning the image with reality, which significantly improves accuracy and perspective. The corrected video is subjected to target detection and target tracking algorithms, recording the current timestamp, vehicle number, geometric center position, and vehicle type, and on this basis, calculating the vehicle speed and acceleration.

[0052] After obtaining the trajectory information of the vehicle queue, a multi-level Monte Carlo model and graph neural network method are used for training, see Figure 4 , predicting the vehicles that are about to arrive at the area. The traditional departure model uses Poisson distribution to obtain the headway distance, and the vehicle speed and vehicle type are randomly assigned according to the probability density distribution of statistical data. Considering that the headway distance between two vehicles in a fleet will be affected by the degree of traffic congestion, the present invention introduces a multi-layer Monte Carlo model to obtain the headway distance data.

[0053] After obtaining the distance between the newly arrived vehicles, the graph neural network is used to predict the vehicle information of the vehicle. The vehicle interaction relationship in the scene is abstracted into a spatiotemporal undirected graph G(V,E), where V represents different vehicles and the graph edge E represents the interaction behavior between different vehicles, including the spatial edge set E s and the inter-frame edge set E f The graph structure is represented by the adjacency matrix A = {A0, A1}, both of which are of size V × V. A0 is the self-connected unit matrix at the temporal level, and A1 is the adjacency matrix at the spatial level. Whether the nodes are adjacent in space is determined by whether the Euclidean distance between the nodes is less than the set threshold. Based on the graph neural network, the vehicle information in the input area is used to infer the speed and model of the newly arrived vehicle.

[0054] The vehicle driving module unifies the vehicle's following and lane-changing behaviors into trajectory prediction. The prediction model is also based on the graph spatiotemporal neural network, and the network structure is consistent with the departure module, but when processing the input data, there is no need to process the model and speed of the newly appeared vehicle. The input features become the lateral position x, longitudinal position y, speed, acceleration, and model, and the output becomes the lateral position x and longitudinal position y of the vehicle at the next moment.

[0055] The traffic flow data generated by the departure module is input into the trajectory prediction model. The predicted vehicle data in the driving area and the data of newly arrived vehicles together constitute the input data for the next moment. The final traffic flow simulation data is obtained by continuous looping. Figure 5 .

[0056] Based on the above multi-level Monte Carlo model, the departure graph neural network model, and the driving graph neural network model, a random traffic flow simulation is performed on the target area. After determining the initial state, the initial state is input into the departure model to obtain several sets of departure data. The departure data is placed in the buffer position of the departure area, and the departure data is input into the driving model according to the time step to simulate the random traffic flow in the area. In the driving area, the vehicles that are still in the driving area in the prediction results are filtered out at each time step, and the vehicles that enter the driving area from the departure area in the new frame together constitute the input of the next prediction model, and the cycle continues until the simulation time reaches the specified time.

[0057] Application Example 1

[0058] In order to verify and analyze the proposed random traffic flow simulation method, the bridge deck video data of a large-span suspension bridge is taken as an example. The data is processed according to the above process. Through result analysis and efficiency calculation, the feasibility of the method in this paper and its advantages of rapidity and accuracy are verified.

[0059] Data processing

[0060] Based on the video data of the bridge deck of a long-span suspension bridge, the surveillance camera is placed in the middle of the bridge tower beam and shoots 1920x1080 pixel video at 25 frames per second from a height of 85 meters. Image distortion is corrected by photogrammetric projection transformation. The coordinates are converted into three-dimensional frames using ground control points (GCPs) and a total coordinate station. This homologous transformation process includes correcting distortion and aligning the image with reality, significantly improving accuracy and perspective. The corrected video is then subjected to Yolov8-based target detection and Deepsort-based target tracking, recording the current timestamp, vehicle number, geometric center position, and vehicle type, and on this basis calculating vehicle speed and vehicle acceleration. For a schematic diagram of the traffic data acquisition process, see Figure 6 .

[0061] Random traffic generation

[0062] This part uses 400 minutes of actual traffic data collected to train the spatiotemporal graph neural network for predicting vehicle model and speed, and 100 minutes of data is used as the validation set and test set. Features include the vehicle horizontal coordinate X and vertical coordinate Y, the adjacent distance is 20 meters, the maximum number of nodes is 40, the input step is 4 frames, the future step is 1 frame, and the interval between each frame is 0.5 seconds. For the frame at time t in the actual traffic data, if there is a new car in the frame, the data from t-3 to t+1 are extracted and added to the training data, and the vehicle model or speed of each vehicle when it first appears is set to 0 to predict the actual vehicle model or speed. The average inference time for sample speed is 3.6ms, and the average inference time for vehicle model is 3.4ms.

[0063] Then, the headway distribution is obtained through various methods, converted into vehicle positions, and input into the model and speed spatiotemporal graph neural network to obtain the simulated initial vehicle data including model, speed, and position. Two methods are used to obtain the headway, namely, Monte Carlo sampling through the fitted headway probability density function to obtain the headway, and constructing a multi-layer Monte Carlo model to obtain the headway. The headway of each vehicle at a certain moment in the multi-layer Monte Carlo model is correlated, which is more in line with the actual situation. The 2nth layer is the lateral distance node between the nth vehicle and the n+1th vehicle, and the 2n+1th layer is the longitudinal distance node between the nth vehicle and the n+1th vehicle. The number of nodes in each layer of the lateral distance layer is 10, and the spacing between adjacent nodes is 200 pixels. The kth node in the lateral distance layer indicates that the lateral position is between (k-1)*200 and k*200 pixels. The number of nodes in each layer of the longitudinal spacing layer is 3, and the kth node indicates that it is in the kth lane.

[0064] Four consecutive frames are randomly selected as the initial input. Each departure lasts for 800 frames. The two methods simulate the departure of about 30,000 vehicles each. The results are shown in Table 1.

[0065] Table 1 Statistics of vehicle speeds and vehicle types for different departure methods

[0066]

[0067] As shown in Table 1, the proportion of each vehicle type, mean speed, and speed variance obtained by Multi-Layer Monte Carlo are closer to the actual situation.

[0068] Random traffic flow

[0069] The vehicle trajectory spatiotemporal graph neural network is trained using actual traffic data. The learning rate is set to 1×10 -4, the batch size is 256. The neighbor distance is 20m, the maximum number of nodes is 100, the input step is 8 frames, the future step is 1 frame, and the interval between each frame is 0.5s. Multiple input features are tried, and the training results are shown in Table 2. When the features are horizontal coordinate X, vertical coordinate Y, vehicle type Class, vehicle speed Speed, and acceleration Acceleration, Loss, MAE, and RMSE are all the smallest, so they are used as the final features. The average inference time of the model sample is 4.5ms.

[0070] Table 1 Minimum training loss and accuracy of different features

[0071]

[0072] Random traffic flow simulation

[0073] The 50 sets of departure data obtained by the departure model are input into the driving model. The departure area is 200 meters, the driving area is 700 meters, and the simulation is repeated 50 times, each lasting 800 frames, with an interval of 0.5 seconds between each frame. The value of X in the departure section is set to 1800. Vehicles with a horizontal coordinate greater than this value in the departure data are considered to have entered the departure area. Vehicles in the departure area travel at a constant speed and enter the driving area when the lateral displacement is greater than 200 meters. Initially, the 1st to 8th frames of the departure data are obtained and input into the trajectory prediction model to obtain the prediction results. Then, the vehicles in the driving area that are still in the prediction results are screened, and the vehicles that enter the driving area from the departure area in the new frame together form the input of the next prediction model, and the cycle continues until the simulation time reaches the specified time.

[0074] Figure 7 This is an example of simulation results. The green blocks are cars and the blue blocks are trucks. Some vehicles show obvious lane changing behavior, which shows that the system is capable of realizing micro-operating condition simulation.

[0075] The statistical characteristics of the headway distance of the measured data and the simulated data are shown in Table 2. The mean and standard deviation of the headway distance of each lane are relatively close.

[0076] Table 2 Random traffic flow simulation headway table

[0077]

[0078] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A random traffic flow simulation method based on graph spatiotemporal neural network, characterized in that: The following steps are involved: S1: The bridge video collected by the camera is corrected, detected, and tracked to form effective trajectory data of the vehicle flow; S2: Using the effective trajectory data of the vehicle flow formed in S1, a vehicle departure model for going onto the bridge is established based on the multi-layer Monte Carlo model and graph spatiotemporal neural network; S3: Using the effective trajectory data of the traffic flow formed in S1 and the vehicle departure model on the bridge established in S2, a vehicle driving micro-simulation model is established based on the graph spatiotemporal neural network model to realize the random traffic flow simulation on the bridge deck.

2. According to claim 1, a random traffic flow simulation method based on graph spatiotemporal neural network is characterized in that: S1 specifically includes the following steps: Video rectification: Correct image distortion through photogrammetric projection transformation using ground control points and the global coordinate station; Vehicle detection and tracking: The corrected video is processed based on the target detection and target tracking algorithm, and the current timestamp, vehicle number, geometric center position, and vehicle type are recorded. The vehicle speed and vehicle acceleration are calculated, and the acquired data together form the effective trajectory data of the vehicle flow.

3. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 2 is characterized in that: The video correction process uses ground control points and a total coordinate station to perform photogrammetric projection transformation to correct image distortion, and converts the original image coordinates into image coordinates of physical space relations, specifically including: Through photogrammetry, the original image is distorted using ground control points and a total coordinate station; Convert the rectified image coordinates into physical space coordinates.

4. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 1 is characterized in that: S2 specifically includes the following steps: Using the effective trajectory data of the traffic flow formed in S1, the time interval and headway between vehicles are simulated through a multi-level Monte Carlo model to predict the departure model of vehicles on the bridge. A graph spatiotemporal neural network is used to predict vehicle information and abstract the vehicle interaction relationship in the scene into a spatiotemporal undirected graph, where nodes represent vehicles and edges represent the interaction behaviors between vehicles. By constructing a multi-layer Monte Carlo model and graph spatiotemporal neural network, inputting vehicle information in the area and inferring the speed and model of newly arrived vehicles, a departure model that can reflect actual traffic conditions is obtained.

5. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 4 is characterized in that: The specific process of simulating the time interval and headway between vehicles through the multi-level Monte Carlo model includes: A multi-level Monte Carlo model is used to simulate the time interval and headway of vehicles arriving at the bridge deck. Statistical methods and random sampling techniques are used to generate a series of possible vehicle arrival times and headways based on historical traffic data, thereby constructing a vehicle departure model that can reflect actual traffic conditions. The departure model can predict the probability distribution and time characteristics of vehicles arriving at the bridge deck under different traffic flows and different time periods, providing accurate starting conditions for subsequent traffic simulation.

6. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 4 is characterized in that: The specific process of abstracting the vehicle interaction relationship in the scene into a spatiotemporal undirected graph includes: after obtaining the distance between the newly arrived vehicles, the graph neural network is used to predict the vehicle information of the target vehicle, and the vehicle interaction relationship in the scene is abstracted into a spatiotemporal undirected graph G(V,E), where V represents different vehicles and the graph edge E represents the interaction behavior between different vehicles, including the spatial edge set E s and the inter-frame edge set E f .

7. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 6 is characterized in that: In the spatiotemporal undirected graph, the graph structure is represented by an adjacency matrix A={A0,A1}, both of which are of size V×V. A0 is the self-connected unit matrix at the temporal level, and A1 is the adjacency matrix at the spatial level. Whether there is spatial adjacency is determined by whether the Euclidean distance between nodes is less than a set threshold.

8. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 4 is characterized in that: S3 specifically includes the following steps: Construction of trajectory prediction model: Using the effective vehicle trajectory data formed in S1 and the vehicle departure model established in S2, a trajectory prediction model based on graph spatiotemporal neural network is constructed; Implementation of random traffic flow simulation: The traffic flow data generated by the departure model is used as the initial input and put into the trajectory prediction model. The model predicts the vehicle data in the driving area and merges the vehicle data with the data of newly arrived vehicles to form the input data for the next moment, thus realizing the microscopic simulation of random traffic flow in the target area.

9. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 8 is characterized in that: The trajectory prediction model unifies the vehicle's following and lane-changing behaviors into a trajectory prediction problem, using a network structure consistent with the departure module. The input features of the model include the vehicle's lateral position x, longitudinal position y, speed, acceleration and vehicle type, and the output is the vehicle's lateral position x' and longitudinal position y' at the next moment.

10. The random traffic flow simulation method based on graph spatiotemporal neural network according to claim 8 is characterized in that: During the simulation process, in the driving area, the vehicles that are still in the driving area in the prediction results are screened out at each time step, and together with the vehicles entering the driving area from the departure area in the new frame, they are used as the input of the next prediction model, and the cycle continues until the simulation time reaches the specified time length.

Images