Vehicle-road cooperative intelligent traffic control method and system based on millimeter wave target perception

Abstract

The present application relates to the field of vehicle-road collaboration technology, and in particular to a vehicle-road collaboration intelligent traffic control method and system based on millimeter wave target perception, wherein the method and system sequentially obtain traffic detection information according to a preset vehicle-road collaboration platform; obtain vehicle basic information and road basic information, and build a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information; generate traffic control information according to the traffic management model and a preset graph neural network model; send the traffic control information to the corresponding vehicle through the vehicle-road collaboration platform; thereby achieving accurate prediction and real-time regulation of emergencies and complex congestion situations, and improving overall response capabilities and operating efficiency.

Description

Technical Field

[0001] The present application relates to the field of vehicle-road collaboration technology, and in particular to a vehicle-road collaboration intelligent traffic control method and system based on millimeter wave target perception. Background Art

[0002] In existing intelligent transportation systems, vehicle-road cooperative technology has gradually been applied, realizing real-time traffic monitoring and basic traffic control functions through information sharing between vehicles and road facilities. Traditional vehicle-road cooperative systems usually rely on sensors such as millimeter-wave radars and cameras to obtain traffic information, including vehicle location, speed, driving trajectory, etc., in order to monitor traffic flow, identify traffic accidents and perform basic congestion relief. However, these systems still have significant deficiencies in processing traffic information and executing control decisions.

[0003] Existing vehicle-road cooperative systems are usually limited to the collection of basic information and simple traffic signal control. They lack in-depth analysis of complex traffic networks and find it difficult to accurately predict and control emergencies and complex congestion situations in real time. Summary of the invention

[0004] Based on this, it is necessary to provide a vehicle-road cooperative intelligent traffic control method and system based on millimeter wave target perception to address the above technical problems, so as to solve the technical problem that the existing traffic control methods based on vehicle-road cooperative platforms are difficult to achieve accurate prediction and real-time regulation of emergencies and complex congestion situations.

[0005] The present invention provides a vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception, comprising:

[0006] Obtain traffic detection information based on the preset vehicle-road collaboration platform;

[0007] Obtain vehicle basic information and road basic information, and build a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information;

[0008] Generate traffic control information according to the traffic management model and a preset graph neural network model;

[0009] The traffic control information is sent to the corresponding vehicle through the vehicle-road cooperative platform.

[0010] As a preferred solution, traffic detection information is obtained according to the preset vehicle-road cooperative platform, including:

[0011] Acquire vehicle detection information based on a vehicle detection device of a preset vehicle-road cooperative platform, wherein the vehicle detection information includes vehicle millimeter wave sensing data and vehicle image detection data;

[0012] Acquiring road detection information based on a road detection device of a preset vehicle-road cooperative platform, wherein the road detection information includes road millimeter wave sensing data and road image detection data;

[0013] Generate millimeter wave sensing information based on a target detection algorithm according to the vehicle millimeter wave sensing data and the road millimeter wave sensing data, and generate image auxiliary information based on a Canny edge detection algorithm according to the vehicle image detection data and the road image detection data;

[0014] The millimeter wave sensing information and the image auxiliary information are data-fused, and a Kalman filter algorithm is used to generate traffic environment sensing information;

[0015] Extracting features through a principal component analysis algorithm based on the traffic environment perception information, and generating vehicle driving information and road traffic status information;

[0016] Traffic detection information is generated according to the vehicle driving information and the traffic status information.

[0017] As a preferred solution, traffic control information is generated according to the traffic management model and the preset graph neural network model, including:

[0018] Extracting data from the traffic management model to generate extracted data, and normalizing and standardizing the extracted data to generate standardized feature data;

[0019] Constructing traffic graph structure data according to the standardized feature data, wherein the traffic graph structure data includes a node feature matrix, an adjacency matrix and an edge feature matrix;

[0020] Inputting the traffic graph structure data into a preset graph neural network model to generate traffic accident information and congestion prediction information;

[0021] Annotating the traffic management model according to the traffic accident information and the traffic forecast information to generate traffic visualization information and traffic warning information;

[0022] Traffic control information is generated according to the visualization information and the traffic warning information.

[0023] As a preferred solution, the traffic graph structure data is input into a preset graph neural network model to generate traffic accident information and traffic prediction information, including:

[0024] Inputting the traffic graph structure data into the input layer of a preset graph neural network model;

[0025] Performing multi-layer convolution on the node feature matrix based on the graph neural network according to the adjacency matrix and the edge feature matrix to generate an updated node feature matrix;

[0026] Transforming the update node feature matrix through a preset activation function to generate a transformed node feature matrix;

[0027] Generate graph-level features based on the pooling layer of the graph neural network according to the transformed node feature matrix, and generate traffic accident information and congestion prediction information according to the graph-level features.

[0028] As a preferred solution, the node feature matrix includes corresponding node features of multiple nodes, and the node features in the updated node feature matrix are transformed to generate corresponding transformed node features based on the following formula;

[0029]

[0030] Among them, v is one of the nodes in the node feature matrix, h v (k+1) is the transformed node feature of node v in the k+1th convolutional layer, σ is the activation function, u is the neighbor node of node v, N (v) is the set of neighbor nodes of node v, d v is the degree of node v, d u is the degree of node u, W is the preset weight matrix, h u (k) is the updated node feature of node u in the kth convolutional layer, and b is the preset bias item.

[0031] As a preferred solution, the traffic accident information includes the accident coordinates, accident type, accident severity and accident impact range;

[0032] The traffic management model is annotated according to the traffic accident information and the congestion prediction information to generate traffic visualization information and traffic warning information, including:

[0033] Creating an accident mark based on the traffic management model according to the accident coordinates and the accident type;

[0034] Creating a dynamic impact area according to the accident mark, the accident severity and the accident impact range;

[0035] generating traffic visualization information according to the congestion prediction information and the dynamic impact area;

[0036] Traffic warning information is generated according to the vehicle driving information and the dynamic impact area.

[0037] As a preferred solution, basic vehicle information and basic road information are obtained, and a traffic management model is constructed based on the Unity engine according to the basic vehicle information, the basic road information and the traffic detection information, including:

[0038] Acquiring basic vehicle information, wherein the basic vehicle information includes vehicle type data, vehicle size data, and vehicle performance data;

[0039] Acquiring basic road information, wherein the basic road information includes road structure data, lane data, traffic rule data and traffic node data;

[0040] Importing the road basic information into the Unity engine, generating a 3D road model according to the road structure data and the lane data, and generating a road network model according to the traffic rule data, the traffic node data and the 3D road model;

[0041] Importing the basic vehicle data into the road network model to generate a traffic network model;

[0042] The traffic network model is updated according to the traffic detection information, and a traffic management model is generated.

[0043] The present invention also provides a vehicle-road cooperative intelligent traffic control system based on millimeter wave target perception, comprising:

[0044] Traffic information detection module, used to obtain traffic detection information according to the preset vehicle-road cooperative platform;

[0045] A management model generation module is used to obtain basic vehicle information and basic road information, and to build a traffic management model based on the Unity engine according to the basic vehicle information, the basic road information and the traffic detection information;

[0046] A control information generation module, used to generate traffic control information according to the traffic management model and a preset graph neural network model;

[0047] The control information sending module is used to send the traffic control information to the corresponding vehicle through the vehicle-road cooperative platform.

[0048] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned vehicle-road collaborative intelligent traffic control method based on millimeter wave target perception are implemented.

[0049] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned vehicle-road collaborative intelligent traffic control method based on millimeter wave target perception are implemented.

[0050] The technical effects achieved by the present invention are as follows:

[0051] The above-mentioned vehicle-road cooperative intelligent traffic control method and system based on millimeter wave target perception obtains traffic detection information according to a preset vehicle-road cooperative platform in sequence; obtains vehicle basic information and road basic information, and builds a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information; generates traffic control information according to the traffic management model and a preset graph neural network model; sends the traffic control information to the corresponding vehicle through the vehicle-road cooperative platform; and then realizes accurate prediction and real-time regulation of emergencies and complex congestion situations, thereby improving the overall response capability and operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a process flow of a vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception in one embodiment;

[0053] Figure 2 This is a structural block diagram of a vehicle-road cooperative intelligent traffic control system based on millimeter wave target perception in one embodiment. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0055] In one embodiment, a terminal is provided, which obtains traffic detection information according to a preset vehicle-road collaborative platform; obtains basic vehicle information and basic road information, and builds a traffic management model based on the Unity engine according to the basic vehicle information, the basic road information and the traffic detection information; generates traffic control information according to the traffic management model and a preset graph neural network model; and sends the traffic control information to the corresponding vehicle through the vehicle-road collaborative platform.

[0056] The terminal may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices.

[0057] In this embodiment, Figure 1 As shown, a vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception is provided, including:

[0058] Step S100: Obtaining traffic detection information according to a preset vehicle-road cooperative platform;

[0059] In this step, real-time traffic information is obtained through the preset vehicle-road collaboration platform. The vehicle-road collaboration platform collects, transmits and analyzes multi-source data of vehicles and roads in real time through pre-set millimeter-wave radars, cameras, etc., to achieve efficient communication and collaboration between vehicles and road infrastructure. This real-time information can better display, predict and control traffic conditions.

[0060] Step S200: acquiring vehicle basic information and road basic information, and constructing a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information;

[0061] In this step, a highly realistic traffic simulation model was built using the Unity engine by integrating basic vehicle information, basic road information, and real-time traffic detection data. The Unity engine has powerful three-dimensional visualization and interactive functions, which can intuitively display information such as traffic network structure, vehicle movement trajectory, and signal light status changes, enabling the model to accurately simulate actual traffic conditions, including details such as vehicle driving behavior, traffic flow characteristics, and road usage. With this visual simulation model, traffic managers can observe and analyze traffic conditions more intuitively and comprehensively, and discover potential problems in a timely manner.

[0062] Step S300: generating traffic control information according to the traffic management model and a preset graph neural network model;

[0063] In this step, the graph neural network model can effectively process complex traffic network structures and capture the associations between road nodes and the dynamic changes of traffic flows. By using the pre-trained graph neural network model to infer the processed traffic data, potential problems such as traffic accidents and congestion can be predicted. Based on these prediction results, corresponding traffic control strategies are generated, such as adjusting the timing of traffic lights, optimizing lane settings, etc. In this embodiment, vehicles are directly dispatched. With the help of this traffic control information, traffic managers can implement effective intervention measures to alleviate traffic pressure in a timely manner and improve road traffic efficiency and traffic safety.

[0064] Step S400: Send the traffic control information to the corresponding vehicle through the vehicle-road cooperative platform.

[0065] In this step, the real-time communication capability of the vehicle-road collaboration platform is used to accurately transmit traffic control instructions to relevant vehicles. These instructions include but are not limited to speed adjustment, lane change suggestions, avoidance reminders, etc., which are intended to guide vehicles to adopt appropriate driving strategies and optimize traffic flow. Through this real-time information interaction, vehicles can quickly respond to the control of the traffic management system, reduce traffic congestion, and reduce the risk of accidents. Efficient collaboration between vehicles and road infrastructure is achieved.

[0066] In this embodiment, this solution is implemented based on the vehicle-road collaborative platform. Traffic detection information including roads and vehicles on the roads is obtained through the vehicle-road collaborative platform, and basic vehicle information and road basic information are integrated. A visual traffic management model is built based on the Unity engine to facilitate real-time monitoring and analysis. Graph neural networks are used to conduct in-depth analysis of traffic data to generate traffic control information. Finally, this information is sent to the corresponding vehicles in real time through the vehicle-road collaborative platform to achieve efficient collaboration and real-time response, thereby improving traffic safety and traffic efficiency.

[0067] In one embodiment, step S100: obtaining traffic detection information according to a preset vehicle-road cooperative platform includes:

[0068] Step S110: acquiring vehicle detection information based on a vehicle detection device of a preset vehicle-road cooperative platform, wherein the vehicle detection information includes vehicle millimeter wave sensing data and vehicle image detection data;

[0069] Step S120: Acquiring road detection information based on a road detection device of a preset vehicle-road cooperative platform, wherein the road detection information includes road millimeter wave sensing data and road image detection data;

[0070] Step S130: generating millimeter wave sensing information based on a target detection algorithm according to the vehicle millimeter wave sensing data and the road millimeter wave sensing data, and generating image auxiliary information based on a Canny edge detection algorithm according to the vehicle image detection data and the road image detection data;

[0071] In this step, the millimeter wave is not affected by light or weather and can work stably under adverse conditions such as rain, fog, and snow. High-precision position, speed, and trajectory information of the vehicle is provided to support fine-grained traffic monitoring. At the same time, edge information of the road environment, such as lane lines, traffic signs, and road boundaries, is extracted through image detection, thereby supplementing the shortcomings of millimeter wave perception information in identifying environmental details. In this way, image-assisted information can provide richer environmental features and enhance the understanding of complex traffic scenes.

[0072] Step S140: fusing the millimeter wave sensing information and the image auxiliary information, and using a Kalman filter algorithm to generate traffic environment sensing information;

[0073] Step S150: extracting features through a principal component analysis algorithm based on the traffic environment perception information, and generating vehicle driving information and road traffic status information;

[0074] Step S160: Generate traffic detection information according to the vehicle driving information and the traffic status information.

[0075] In this embodiment, vehicle millimeter wave perception data, vehicle image detection data, road millimeter wave perception data and road image detection data are obtained through the preset vehicle detection device and road detection device of the vehicle-road collaborative platform. Millimeter wave perception information is generated by the target detection algorithm, and image auxiliary information is generated based on the Canny edge detection algorithm. The advantages of millimeter wave sensors and image sensors are combined to make up for the perception deficiencies that may occur in a single sensor under specific environments, ensuring that the system can operate normally under various complex conditions. The above information is fused and the Kalman filter algorithm is used to generate traffic environment perception information to improve the accuracy and stability of traffic environment perception. Feature extraction is then performed through the principal component analysis algorithm to generate vehicle driving information and road traffic status information, and finally traffic detection information is generated to provide a basis for the subsequent formulation of traffic control strategies.

[0076] In one embodiment, step S200: obtaining vehicle basic information and road basic information, and building a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information, including:

[0077] Step S210: Acquire basic vehicle information, wherein the basic vehicle information includes vehicle type data, vehicle size data, and vehicle performance data;

[0078] In this step, by obtaining basic information such as vehicle type, size, and performance, the characteristics and behaviors of different types of vehicles can be accurately simulated, the accuracy of traffic simulation can be improved, the traffic management model can reflect the real vehicle dynamics, and data support can be provided for the formulation of traffic control strategies for different vehicles.

[0079] Step S220: Acquire basic road information, wherein the basic road information includes road structure data, lane data, traffic rule data and traffic node data;

[0080] In this step, detailed information such as road structure, lane configuration, traffic rules and traffic nodes is obtained, so that the traffic management model can simulate traffic flow and vehicle behavior more accurately and reflect road conditions accurately.

[0081] Step S230: importing the basic road information into the Unity engine, generating a 3D road model according to the road structure data and the lane data, and generating a road network model according to the traffic rule data, the traffic node data and the 3D road model;

[0082] In this step, 3D road models and road network models are generated in the Unity engine, and its three-dimensional visualization and interactive functions are used to intuitively display the traffic network structure and road details, allowing traffic managers to observe traffic conditions more intuitively.

[0083] Step S240: importing the basic vehicle data into the road network model to generate a traffic network model;

[0084] In this step, the basic vehicle data is imported into the road network model, and the generated traffic network model can simulate the operation of vehicles in the road network, which is helpful for analyzing traffic flow characteristics, vehicle behavior and road usage.

[0085] Step S250: updating the traffic network model according to the traffic detection information, and generating a traffic management model.

[0086] In this step, the traffic network model is updated using real-time traffic detection information to ensure that the model always reflects the current traffic conditions.

[0087] In this embodiment, by obtaining basic vehicle information (including vehicle type data, vehicle size data and vehicle performance data) and basic road information (including road structure data, lane data, traffic rule data and traffic node data), these information are imported into the Unity engine, a three-dimensional road model is generated according to the road structure data and lane data, and then a road network model is generated by combining the traffic rule data, traffic node data and the three-dimensional road model. Subsequently, the basic vehicle data is imported into the road network model to generate a traffic network model, and the traffic network model is updated according to real-time traffic detection information, and finally a traffic management model is generated. By constructing a traffic management model in the Unity engine, the solution realizes a highly realistic traffic simulation, accurately simulating details such as vehicle driving behavior, traffic flow characteristics and road usage. The model has powerful three-dimensional visualization and interactive functions, which is convenient for traffic managers to observe and analyze traffic conditions intuitively and comprehensively, and to discover potential problems in a timely manner, providing a reliable basis for optimizing traffic control strategies (such as signal light optimization, traffic flow regulation, etc.), thereby improving the efficiency and effectiveness of traffic management, helping to alleviate traffic congestion, reduce the incidence of accidents, and improve the safety and operation efficiency of the traffic system.

[0088] In one embodiment, step S300: generating traffic control information according to the traffic management model and a preset graph neural network model includes:

[0089] Step S310: extracting data from the traffic management model to generate extracted data, and normalizing and standardizing the extracted data to generate standardized feature data;

[0090] In this step, normalization and standardization are used to eliminate the dimensional differences between different features, prevent certain features from having too much impact on model training, improve data consistency and comparability, and reduce the risk of overfitting of the model.

[0091] Step S320: constructing traffic graph structure data according to the standardized feature data, wherein the traffic graph structure data includes a node feature matrix, an adjacency matrix and an edge feature matrix;

[0092] In this step, the node feature matrix is ​​used to describe the basic information of each traffic entity (e.g., vehicle, traffic node, etc.). The graph neural network understands the spatial relationship between vehicles, between vehicles and roads, and between vehicles and traffic lights through the node feature matrix, so as to effectively predict traffic accidents, plan paths, and optimize traffic flows. The adjacency matrix represents the interactions and connections between traffic entities. The graph neural network understands the spatial relationship between vehicles, between vehicles and roads, and between vehicles and traffic lights through the adjacency matrix, so as to effectively predict traffic accidents, plan paths, and optimize traffic flows.

[0093] The edge feature matrix is ​​used to represent the attributes of the connected edges in the graph, that is, the interactive information between nodes. Through the edge feature matrix, the graph neural network can transmit information between multiple nodes, update the node status, and generate more accurate traffic prediction and accident warning information.

[0094] Step S330: inputting the traffic graph structure data into a preset graph neural network model to generate traffic accident information and congestion prediction information;

[0095] Step S340: annotating the traffic management model according to the traffic accident information and the traffic forecast information to generate traffic visualization information and traffic warning information;

[0096] In this step, accident and congestion information is marked in the traffic management model, and a visual traffic status map is generated to facilitate managers to intuitively understand the current and predicted traffic conditions. At the same time, traffic warning information is generated to promptly notify relevant departments and vehicle drivers, so that response measures can be taken in advance to reduce the impact of accidents and congestion.

[0097] Step S350: Generate traffic control information according to the visualization information and the traffic warning information.

[0098] In this embodiment, the data of the traffic management model is extracted and preprocessed to ensure the consistency between the features, and standardized feature data is generated. Based on these standardized data, traffic graph structure data is constructed, including node feature matrix, adjacency matrix and edge feature matrix. Through these matrices, the graph neural network can understand the relationship and interaction between traffic entities, capture the complex traffic network topology and dynamic characteristics, and thus more accurately predict traffic accidents and congestion. The traffic graph structure data is input into the preset graph neural network model to generate traffic accident information and congestion prediction information. Based on these prediction results, the traffic management model is annotated to generate intuitive traffic visualization information and warning information to help traffic managers understand and respond to traffic conditions in a timely manner. Using these visualization and warning information, traffic control information is generated to improve traffic efficiency and reduce congestion and accidents.

[0099] In one embodiment, step S330: inputting the traffic graph structure data into a preset graph neural network model to generate traffic accident information and traffic prediction information includes:

[0100] Step S331: inputting the traffic graph structure data into the input layer of a preset graph neural network model;

[0101] Step S332: performing multi-layer convolution on the node feature matrix based on the graph neural network according to the adjacency matrix and the edge feature matrix to generate an updated node feature matrix;

[0102] In this step, multi-layer convolution operations are performed on the adjacency matrix, edge feature matrix, and node feature matrix through the graph neural network, so that the node can aggregate its own information and that of its neighboring nodes, extract deeper features, and generate an updated node feature matrix.

[0103] Step S333: transforming the update node feature matrix through a preset activation function to generate a transformed node feature matrix;

[0104] In this step, a preset ReLU activation function is applied to the updated node feature matrix for nonlinear transformation to generate a transformed node feature matrix.

[0105] Step S334: Generate graph-level features based on the pooling layer of the graph neural network according to the transformed node feature matrix, and generate traffic accident information and congestion prediction information according to the graph-level features.

[0106] In this embodiment, by inputting the traffic graph structure data into the preset graph neural network model, the generation of traffic accident information and traffic prediction information is realized. Specifically, the constructed traffic graph structure data (including node feature matrix, adjacency matrix and edge feature matrix) is first input into the input layer of the model. Then, based on the adjacency matrix and edge feature matrix, multi-layer convolution operation is performed on the node feature matrix to aggregate the information of the node itself and the neighboring nodes to extract deeper features. Then, the preset activation function is applied to perform nonlinear transformation on the updated node feature matrix to capture complex traffic patterns. Subsequently, the transformed node feature matrix is ​​pooled through the pooling layer of the graph neural network to generate a graph-level feature representation, and finally traffic accident information and congestion prediction information are generated based on these features. By utilizing the advantages of graph neural networks in processing graph structure data, it is possible to effectively capture the complex spatial and topological relationships in the traffic network, improve the accuracy of traffic accident and congestion predictions, provide timely and reliable decision-making basis for traffic managers, and help take measures in advance to reduce accident risks, alleviate traffic pressure, optimize traffic control strategies, and improve road traffic efficiency and traffic safety levels.

[0107] In one embodiment, the node feature matrix includes corresponding node features of a plurality of nodes, and the node features in the updated node feature matrix are transformed to generate corresponding transformed node features based on the following formula;

[0108]

[0109] Among them, v is one of the nodes in the node feature matrix, h v (k+1) is the transformed node feature of node v in the k+1th convolutional layer, σ is the activation function, u is the neighbor node of node v, N (v) is the set of neighbor nodes of node v, d v is the degree of node v, d u is the degree of node u, W is the preset weight matrix, h u (k) is the updated node feature of node u in the kth convolutional layer, and b is the preset bias item.

[0110] In this embodiment, through the sequence of neighbor node feature aggregation, weighting, normalization and nonlinear activation, the graph neural network extracts deep information in the graph structure layer by layer, thereby learning the complex relationship between nodes in the traffic network, and ultimately providing more accurate basic data for traffic accident and congestion prediction. Specifically, the feature update of node v first obtains information about the nodes around node v by aggregating the features of neighbor node u, thereby capturing the local structural relationship in the traffic network. Then, the feature h of each neighbor node u (k)The preset weight matrix W is used for weighting to ensure that nodes can adjust the influence of information according to different neighbor characteristics and improve the expressiveness of the model. is the degree normalization factor, which is normalized by the square root of the node degree to ensure that feature aggregation is not affected by the number of node connections, thereby ensuring the balance and stability of information propagation and preventing feature deviation caused by different numbers of node connections. Next, the aggregated features are added with the bias term b and nonlinearly transformed by the activation function σ to generate new node features h v (k+1) ,Introducing nonlinearity enables the model to capture complex traffic patterns.

[0111] In one embodiment, the traffic accident information includes the accident coordinates, the accident type, the accident severity, and the accident impact range. Step S340: annotating the traffic management model according to the traffic accident information and the congestion prediction information to generate traffic visualization information and traffic warning information includes:

[0112] Step S341: creating an accident mark based on the traffic management model according to the accident coordinates and the accident type;

[0113] Step S342: creating a dynamic impact area according to the accident mark, the accident severity and the accident impact range;

[0114] Step S343: generating traffic visualization information according to the congestion prediction information and the dynamic impact area;

[0115] Step S344: Generate traffic warning information according to the vehicle driving information and the dynamic impact area.

[0116] In this embodiment, firstly, an accident mark is created in the traffic management model according to the accident coordinates and the type of accident, so that the traffic manager can intuitively understand the location and nature of the accident, and take countermeasures quickly. Then, according to the accident mark, the severity of the accident and the scope of the accident, a dynamic impact area is created to accurately display the scope of the accident's impact on the surrounding traffic and help formulate an effective traffic diversion plan. Then, according to the congestion prediction information and the dynamic impact area, traffic visualization information is generated to present the traffic flow and congestion in an intuitive way, supporting managers to adjust traffic control strategies in a timely manner. Finally, according to the vehicle driving information and the dynamic impact area, traffic warning information is generated to provide real-time warning prompts for vehicle drivers, guide them to reasonably plan routes, avoid accident areas, and improve driving safety and traffic smoothness.

[0117] In one embodiment, the vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception further includes:

[0118] Step S510: obtaining current traffic flow information;

[0119] Step S520: acquiring the traffic accident information and the congestion prediction information to generate a traffic signal change instruction, and sending the traffic signal change instruction to the traffic light and lane sign;

[0120] In this step, based on the traffic accident information and congestion prediction information generated in the previous stage, the potential risks and traffic bottlenecks of the current road conditions are analyzed, and traffic signal change instructions are generated, such as adjusting the traffic light cycle, opening or closing variable lanes, etc., in order to optimize traffic flow. Then, these change instructions are sent to traffic facilities such as traffic lights and lane signs to achieve real-time traffic control. By generating and implementing traffic signal change instructions, traffic can be diverted in a timely manner, congestion can be alleviated, and emergency channels can be provided in the event of an accident, effectively alleviating traffic pressure, improving road traffic efficiency, and reducing the further spread of accidents.

[0121] Step S530: updating the traffic management model according to the traffic signal change instruction, and generating control effect prediction information;

[0122] In this step, after executing the traffic signal change command, the change information is fed back to the traffic management model to update the signal light status, lane usage and other contents in the model. At the same time, based on these adjustments, the system predicts the potential impact of these changes on traffic flow and generates control effect prediction information to evaluate the effectiveness of the adjustment measures. By reflecting the latest signal light and lane status in the model, the system can evaluate the potential effect of the current adjustment measures in real time, provide a reference for subsequent optimization, and ensure that traffic control measures are more accurate and further improve management efficiency.

[0123] Step S540: obtaining adjusted traffic flow information, and generating actual control effect information according to the current traffic flow information and the adjusted traffic flow information;

[0124] In this step, after the traffic signal change has been implemented for a period of time, the latest traffic flow information is collected again to evaluate the actual effect of the adjustment measures. By comparing the traffic flow information before the adjustment with the data after the adjustment, the actual control effect information is generated to determine whether the change measures have effectively alleviated congestion or improved traffic efficiency. This step helps the system accurately evaluate the actual impact of the current traffic control measures and provide effective feedback data to further optimize the traffic signal adjustment strategy to ensure that traffic control is always effective and dynamically adapts to actual traffic conditions.

[0125] Step S550: Generate an optimized signal change instruction based on the control effect prediction information and the actual control effect information, and send the optimized change instruction to the traffic light and lane sign.

[0126] In this step, the predicted control effect information is compared with the actual control effect information to identify possible deviations or optimization space. Based on the analysis results, the system generates optimized traffic signal change instructions to further adjust the signal light cycle or lane indication information. The optimization instructions are sent to the signal lights and lane signs again to ensure continuous improvement of traffic control measures.

[0127] In this embodiment, the current traffic flow information is first obtained, and combined with the traffic accident and congestion prediction data, a traffic signal change instruction is generated and sent to the signal light and lane sign, so as to immediately adjust the traffic light cycle or lane indication information to optimize traffic flow. Then, these change information is fed back to the traffic management model to predict the potential impact of these adjustments on traffic conditions, and updated traffic flow data is obtained after a period of time. By comparing the traffic conditions before and after the adjustment, the actual control effect information is generated to evaluate the effectiveness of the current measures. Finally, the predicted effect is compared and analyzed with the actual effect, the optimization space is identified, and further optimized signal change instructions are generated to ensure the continuous improvement of the traffic control strategy. Through this dynamic closed-loop control process, accurate signal adjustments can be made according to the real-time traffic conditions, timely congestion relief, and traffic efficiency can be improved. At the same time, the strategy is continuously optimized through feedback to ensure the adaptability and efficiency of traffic control measures, thereby effectively reducing the risk of accidents and enhancing the overall flexibility and responsiveness of the transportation system.

[0128] In one embodiment, Figure 2 As shown, a vehicle-road cooperative intelligent traffic control system based on millimeter wave target perception is provided, including a traffic information detection module, a management model generation module, a control information generation module and a control information sending module.

[0129] The traffic information detection module is used to obtain traffic detection information according to a preset vehicle-road cooperative platform.

[0130] The management model generation module is used to obtain vehicle basic information and road basic information, and build a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information.

[0131] The control information generation module is used to generate traffic control information based on the traffic management model and a preset graph neural network model.

[0132] The control information sending module is used to send the traffic control information to the corresponding vehicle through the vehicle-road cooperative platform.

[0133] In one embodiment, the traffic information detection module is also used to: obtain vehicle detection information based on a vehicle detection device of a preset vehicle-road collaborative platform, wherein the vehicle detection information includes vehicle millimeter-wave perception data and vehicle image detection data; obtain road detection information based on a road detection device of a preset vehicle-road collaborative platform, wherein the road detection information includes road millimeter-wave perception data and road image detection data; generate millimeter-wave perception information based on a target detection algorithm according to the vehicle millimeter-wave perception data and the road millimeter-wave perception data, and generate image auxiliary information based on a Canny edge detection algorithm according to the vehicle image detection data and the road image detection data; perform data fusion on the millimeter-wave perception information and the image auxiliary information, and generate traffic environment perception information using a Kalman filtering algorithm; perform feature extraction using a principal component analysis algorithm based on the traffic environment perception information, and generate vehicle driving information and road traffic status information; generate traffic detection information based on the vehicle driving information and the traffic status information.

[0134] In one embodiment, the management model generation module is also used to: obtain basic vehicle information, wherein the basic vehicle information includes vehicle type data, vehicle size data and vehicle performance data; obtain basic road information, wherein the basic road information includes road structure data, lane data, traffic rule data and traffic node data; import the basic road information into the Unity engine, generate a 3D road model according to the road structure data and the lane data, and generate a road network model according to the traffic rule data, the traffic node data and the 3D road model; import the basic vehicle data into the road network model to generate a traffic network model; update the traffic network model according to the traffic detection information, and generate a traffic management model.

[0135] In one embodiment, the control information generation module is also used to: extract data from the traffic management model to generate extracted data, normalize and standardize the extracted data to generate standardized feature data; construct traffic graph structure data based on the standardized feature data, wherein the traffic graph structure data includes a node feature matrix, an adjacency matrix and an edge feature matrix; input the traffic graph structure data into a preset graph neural network model to generate traffic accident information and congestion prediction information; annotate the traffic management model based on the traffic accident information and the traffic prediction information to generate traffic visualization information and traffic warning information; and generate traffic control information based on the visualization information and the traffic warning information.

[0136] In one embodiment, the control information generation module is also used to: input the traffic graph structure data into the input layer of a preset graph neural network model; perform multi-layer convolution on the node feature matrix based on the graph neural network according to the adjacency matrix and the edge feature matrix to generate an updated node feature matrix; transform the updated node feature matrix through a preset activation function to generate a transformed node feature matrix; generate graph-level features based on the pooling layer of the graph neural network according to the transformed node feature matrix, and generate traffic accident information and congestion prediction information according to the graph-level features.

[0137] In one embodiment, the control information generation module is further used for: the node feature matrix includes corresponding node features of a plurality of nodes, and the node features in the update node feature matrix are transformed to generate corresponding transformed node features based on the following formula;

[0138]

[0139] Among them, v is one of the nodes in the node feature matrix, h v (k+1) is the transformed node feature of node v in the k+1th convolutional layer, σ is the activation function, u is the neighbor node of node v, N (v) is the set of neighbor nodes of node v, d v is the degree of node v, d u is the degree of node u, W is the preset weight matrix, h u (k) is the updated node feature of node u in the kth convolutional layer, and b is the preset bias item.

[0140] In one embodiment, the control information generation module is also used to: create an accident mark based on the traffic management model according to the accident coordinates and the accident type; create a dynamic impact area according to the accident mark, the accident severity and the accident impact range; generate traffic visualization information according to the congestion prediction information and the dynamic impact area; generate traffic warning information according to the vehicle driving information and the dynamic impact area.

[0141] In one embodiment, the vehicle-road cooperative intelligent traffic control based on millimeter wave target perception also includes a change instruction generation module.

[0142] The change instruction generation module is used to obtain current traffic flow information; obtain the traffic accident information and the congestion prediction information to generate a traffic signal change instruction, and send the traffic signal change instruction to the traffic light and lane sign; update the traffic management model according to the traffic signal change instruction, and generate regulation effect prediction information; obtain adjusted traffic flow information, and generate actual regulation effect information according to the current traffic flow information and the adjusted traffic flow information; generate optimized signal change instructions according to the regulation effect prediction information and the actual regulation effect information, and send the optimized change instructions to the traffic light and lane sign.

[0143] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps described in the above-mentioned vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception when executing the computer program.

[0144] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps described in the above-mentioned vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception are implemented.

[0145] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database 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. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0146] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0147] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception, characterized in that: include: Obtain traffic detection information based on the preset vehicle-road collaboration platform; Obtain vehicle basic information and road basic information, and build a traffic management model based on the Unity engine according to the vehicle basic information, the road basic information and the traffic detection information; Generate traffic control information according to the traffic management model and a preset graph neural network model; The traffic control information is sent to the corresponding vehicle through the vehicle-road cooperative platform.

2. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to claim 1 is characterized in that: Obtain traffic detection information based on the preset vehicle-road collaboration platform, including: Acquire vehicle detection information based on a vehicle detection device of a preset vehicle-road cooperative platform, wherein the vehicle detection information includes vehicle millimeter wave sensing data and vehicle image detection data; Acquiring road detection information based on a road detection device of a preset vehicle-road cooperative platform, wherein the road detection information includes road millimeter wave sensing data and road image detection data; Generate millimeter wave sensing information based on a target detection algorithm according to the vehicle millimeter wave sensing data and the road millimeter wave sensing data, and generate image auxiliary information based on a Canny edge detection algorithm according to the vehicle image detection data and the road image detection data; The millimeter wave sensing information and the image auxiliary information are data-fused, and a Kalman filter algorithm is used to generate traffic environment sensing information; Extracting features through a principal component analysis algorithm based on the traffic environment perception information, and generating vehicle driving information and road traffic status information; Traffic detection information is generated according to the vehicle driving information and the traffic status information.

3. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to claim 2 is characterized in that: Generate traffic control information according to the traffic management model and the preset graph neural network model, including: Extracting data from the traffic management model to generate extracted data, and normalizing and standardizing the extracted data to generate standardized feature data; Constructing traffic graph structure data according to the standardized feature data, wherein the traffic graph structure data includes a node feature matrix, an adjacency matrix and an edge feature matrix; Inputting the traffic graph structure data into a preset graph neural network model to generate traffic accident information and congestion prediction information; Annotating the traffic management model according to the traffic accident information and the traffic forecast information to generate traffic visualization information and traffic warning information; Traffic control information is generated according to the visualization information and the traffic warning information.

4. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to claim 3 is characterized in that: The traffic graph structure data is input into a preset graph neural network model to generate traffic accident information and traffic prediction information, including: Inputting the traffic graph structure data into the input layer of a preset graph neural network model; Performing multi-layer convolution on the node feature matrix based on the graph neural network according to the adjacency matrix and the edge feature matrix to generate an updated node feature matrix; Transforming the update node feature matrix through a preset activation function to generate a transformed node feature matrix; Generate graph-level features based on the pooling layer of the graph neural network according to the transformed node feature matrix, and generate traffic accident information and congestion prediction information according to the graph-level features.

5. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to claim 4 is characterized in that: The node feature matrix includes corresponding node features of a plurality of nodes, and the node features in the update node feature matrix are transformed to generate corresponding transformed node features based on the following formula; Among them, v is one of the nodes in the node feature matrix, h v (k+1) is the transformed node feature of node v in the k+1th convolutional layer, σ is the activation function, u is the neighbor node of node v, N (v) is the set of neighbor nodes of node v, d v is the degree of node v, d u is the degree of node u, W is the preset weight matrix, h u (k) is the updated node feature of node u in the kth convolutional layer, and b is the preset bias item.

6. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to claim 3 is characterized in that: The traffic accident information includes the accident coordinates, accident type, accident severity and accident impact range; The traffic management model is annotated according to the traffic accident information and the congestion prediction information to generate traffic visualization information and traffic warning information, including: Creating an accident mark based on the traffic management model according to the accident coordinates and the accident type; Creating a dynamic impact area according to the accident mark, the accident severity and the accident impact range; generating traffic visualization information according to the congestion prediction information and the dynamic impact area; Traffic warning information is generated according to the vehicle driving information and the dynamic impact area.

7. The vehicle-road cooperative intelligent traffic control method based on millimeter wave target perception according to any one of claims 1 to 6, characterized in that: Obtaining basic vehicle information and basic road information, and building a traffic management model based on the Unity engine according to the basic vehicle information, the basic road information and the traffic detection information, including: Acquiring basic vehicle information, wherein the basic vehicle information includes vehicle type data, vehicle size data, and vehicle performance data; Acquiring basic road information, wherein the basic road information includes road structure data, lane data, traffic rule data and traffic node data; Importing the road basic information into the Unity engine, generating a 3D road model according to the road structure data and the lane data, and generating a road network model according to the traffic rule data, the traffic node data and the 3D road model; Importing the basic vehicle data into the road network model to generate a traffic network model; The traffic network model is updated according to the traffic detection information, and a traffic management model is generated.

8. A vehicle-road cooperative intelligent traffic control system based on millimeter wave target perception, characterized in that: include: Traffic information detection module, used to obtain traffic detection information according to the preset vehicle-road cooperative platform; A management model generation module is used to obtain basic vehicle information and basic road information, and to build a traffic management model based on the Unity engine according to the basic vehicle information, the basic road information and the traffic detection information; A control information generation module, used to generate traffic control information according to the traffic management model and a preset graph neural network model; The control information sending module is used to send the traffic control information to the corresponding vehicle through the vehicle-road cooperative platform.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images