Cooperative intelligent road condition monitoring method for Internet of Vehicles

Abstract

The invention provides a cooperative intelligent road condition monitoring method for the Internet of Vehicles, and relates to the technical field of intelligent traffic, and the method comprises the steps: carrying out the data transmission and fusion of a data collection result through a data transmission mode, and obtaining the global road condition evaluation; performing anomaly recognition on the global road condition evaluation to obtain an abnormal traffic behavior; road condition evaluation and intelligent scheduling are carried out according to the abnormal traffic behaviors, and an intelligent traffic scheduling result is obtained; and generating a lane scheduling report by using an intelligent traffic scheduling result. According to the invention, the technical problems of low road condition monitoring precision and slow abnormal traffic event response caused by dependence on a single data source and insufficient data real-time performance in the prior art are solved, and the technical effects of improving the accuracy and real-time performance of road condition monitoring and further improving the traffic scheduling response speed are achieved.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and particularly to a collaborative intelligent road condition monitoring method for the vehicle-to-everything (V2X) network. Background Art

[0002] Road condition monitoring technology plays an important role in improving road traffic efficiency, reducing traffic accidents, and optimizing urban traffic management. Existing road condition monitoring methods mainly rely on fixed sensors (such as road cameras, inductive loop detectors), vehicle-mounted sensors (such as GPS, millimeter-wave radar), and mobile terminal data (such as traffic flow information provided by mobile navigation software). On the one hand, the data sources are relatively single, mainly relying on fixed sensors on the road or limited sensors on the vehicle itself to obtain information, and only basic data such as traffic flow and vehicle speed on local sections can be obtained, making it difficult to comprehensively reflect the status of the entire traffic network. On the other hand, there is a lack of an efficient architecture for data processing, mostly simple local processing or centralized cloud processing, resulting in insufficient real-time data processing. Due to the lag and incompleteness of the data, accurate judgments and responses cannot be made in a timely manner when abnormal traffic events occur, such as traffic accidents or sudden traffic jams, which affects traffic dispatching, leading to problems such as increased traffic congestion and reduced travel efficiency. Summary of the Invention

[0003] This application provides a collaborative intelligent road condition monitoring method for the vehicle-to-everything (V2X) network, which solves the technical problems in the prior art that due to relying on a single data source and insufficient data real-time performance, the accuracy of road condition monitoring is low and the response to abnormal traffic events is slow, and achieves the technical effect of improving the accuracy and real-time performance of road condition monitoring, and further improving the response speed of traffic dispatching.

[0004] In view of the above problems, this application provides a collaborative intelligent road condition monitoring method for the vehicle-to-everything (V2X) network, and the method includes: performing data transmission and fusion on the data acquisition results through a data transmission mode to obtain a global road condition assessment; performing abnormal identification on the global road condition assessment to obtain abnormal traffic behaviors; performing road condition assessment and intelligent dispatching according to the abnormal traffic behaviors to obtain an intelligent traffic dispatching result; and generating a lane dispatching report by using the intelligent traffic dispatching result.

[0005] Preferably, the data transmission mode includes a first transmission mode, a second transmission mode, and a third transmission mode, where the first transmission mode is a vehicle-to-vehicle communication transmission mode, the second transmission mode is a vehicle-to-infrastructure communication transmission mode, and the third transmission mode is a vehicle-to-cloud communication transmission mode.

[0006] Preferably, real-time road information is collected based on in-vehicle sensors, real-time road facilities are sensed through infrastructure perception devices, vehicle driving status and external environment data are collected by using intelligent mobile terminals, and the real-time road information, the real-time road facilities, the vehicle driving status and the external environment data are added to the data collection result.

[0007] Preferably, data transmission and fusion are performed on the data collection result through a data transmission mode to obtain a global road condition assessment, including: performing data transmission on the data collection result through the data transmission mode to obtain transmission data; performing data denoising, feature extraction and compressed transmission on the transmission data at the edge end to obtain preprocessed data; inputting the preprocessed data into a road condition assessment model obtained by BP neural network learning in the cloud, and analyzing to obtain the global road condition assessment.

[0008] Preferably, abnormal identification is performed on the global road condition assessment to obtain abnormal traffic behaviors, including: performing visual analysis on the global road condition assessment to identify traffic anomalies, where the traffic anomalies include traffic accidents, traffic jams, road surface damage and bad weather; performing spatio-temporal data modeling on the traffic accidents, traffic jams, road surface damage and bad weather to obtain a predicted traffic flow trend; using an isolation forest to analyze the predicted traffic flow trend to identify the abnormal traffic behaviors.

[0009] Preferably, vehicle violation behaviors are judged based on the vehicle driving trajectory and added to the abnormal traffic behaviors, where the vehicle violation behaviors include speeding, sudden braking and occupying the emergency lane.

[0010] Preferably, road condition assessment and intelligent scheduling are performed according to the abnormal traffic behaviors to obtain an intelligent traffic scheduling result, including: generating a dynamic road condition map through the abnormal traffic behaviors, where the dynamic road condition map includes visual road condition information; performing reinforcement learning on the dynamic road condition map to obtain optimized signal light timing and adjusted traffic flow directions; performing path optimization according to the optimized signal light timing and adjusted traffic flow directions to obtain a path optimization result; using an autonomous driving system to execute vehicle-road collaborative control of the path optimization result to obtain the intelligent traffic scheduling result.

[0011] Preferably, the road condition assessment is optimized through a feedback mechanism, where the feedback mechanism includes navigation recommendations for user behavior analysis and a closed-loop optimization mechanism for signal light regulation.

[0012] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0013] Through the integration of multi-modal data collection and transmission, this application utilizes vehicle-to-vehicle, vehicle-to-infrastructure, and vehicle-to-cloud communications to enrich data sources and enhance the real-time and comprehensiveness of monitoring data. Data preprocessing, feature extraction, and compression are performed at the edge to reduce transmission latency and improve data utilization efficiency. Deep learning is carried out using a BP neural network model in the cloud to enhance the accuracy of global road condition assessment. Visual analysis and spatio-temporal data modeling are performed on the global road condition assessment, and an isolation forest is used to identify abnormal traffic behaviors, enabling precise monitoring and early warning of traffic anomaly events. A dynamic road condition map is generated based on abnormal behaviors, and signal timing and traffic flow are optimized in combination with reinforcement learning. A dynamic road condition map is generated and path optimization is executed, achieving intelligent scheduling through vehicle-road collaborative control. In addition, user behavior analysis and closed-loop optimization of signal regulation are introduced as feedback mechanisms to continuously optimize road condition assessment and scheduling strategies. Overall, the solution improves the accuracy of road condition monitoring, increases the response speed of traffic dispatching, and realizes more intelligent and efficient traffic management.

[0014] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specifically illustrates the specific implementation manners of this application. Brief Description of the Drawings

[0015] Figure 1 It is a schematic flowchart of a collaborative intelligent road condition monitoring method for the Internet of Vehicles provided by an embodiment of this application.

[0016] Figure 2 It is a schematic flowchart of obtaining abnormal traffic behaviors in the collaborative intelligent road condition monitoring method for the Internet of Vehicles provided by an embodiment of this application. Detailed Description of the Invention

[0017] By providing a collaborative intelligent road condition monitoring method for the Internet of Vehicles, the embodiment of this application solves the technical problems of low accuracy of road condition monitoring and slow response to abnormal traffic events in the prior art due to relying on a single data source and insufficient data real-time performance, and achieves the technical effect of improving the accuracy and real-time performance of road condition monitoring, thereby increasing the response speed of traffic dispatching.

[0018] As Figure 1 shown, the embodiment of this application provides a collaborative intelligent road condition monitoring method for the Internet of Vehicles, and the method includes:

[0019] Step S1: Perform data transmission and fusion on the data collection results through a data transmission mode to obtain a global road condition assessment.

[0020] Specifically, the data transmission mode refers to the way data is transmitted between different devices or systems. In the vehicle networking environment, vehicles collect data through their own sensors (such as speed sensors, cameras, etc.) and infrastructure (such as traffic flow monitoring devices on the roadside). The collected data results are transmitted through data transmission modes such as vehicle-to-vehicle, vehicle-to-infrastructure, and vehicle-to-cloud. In cloud or local edge computing devices, data fusion algorithms are used to fuse and process this data from different sources, comprehensively considering the data information of each section and each vehicle, so as to obtain a global road condition assessment. This global road condition assessment is a comprehensive evaluation of the entire traffic network situation, including a comprehensive consideration of various aspects of information such as road congestion, vehicle speed distribution, and traffic accident conditions.

[0021] Through the integration of multi-source data, the limitations of a single data source are broken, enabling a more comprehensive and accurate understanding of the entire traffic network situation, providing a rich data basis for subsequent road condition analysis and decision-making.

[0022] Step S2: Perform anomaly identification on the global road condition assessment to obtain abnormal traffic behaviors.

[0023] Specifically, analyze the global road condition assessment result obtained in step S1, and use the established anomaly identification model or algorithm to identify behaviors that do not conform to the normal traffic flow pattern, including traffic accidents, vehicle violations (such as speeding, running red lights), sudden road congestion, etc., to obtain abnormal traffic behaviors. Exemplarily, an image recognition algorithm based on deep learning can be used to analyze the images collected by cameras to identify traffic accidents or vehicle violations; abnormal situations such as sudden congestion points can also be discovered through spatio-temporal analysis of traffic flow, vehicle speed, etc.

[0024] By performing anomaly identification on the global road condition assessment, abnormal situations in the traffic network can be identified in a timely and accurate manner, providing targeted inputs for subsequent intelligent traffic dispatching and improving the response speed to traffic anomaly events.

[0025] Step S3: Perform road condition assessment and intelligent dispatching based on the abnormal traffic behaviors to obtain an intelligent traffic dispatching result.

[0026] Specifically, based on the abnormal traffic behaviors identified in step S2, perform road condition assessment and intelligent dispatching through intelligent traffic dispatching algorithms and models to generate an intelligent traffic dispatching result. For example, if it is found that a traffic accident has occurred at a certain intersection, causing congestion, the green light duration of the intersection signal can be adjusted to reduce the green light time in the congested direction and increase the green light time in other directions; at the same time, path planning algorithms can be used to re-plan the driving routes for surrounding vehicles to guide the vehicles to bypass the congested area.

[0027] Through targeted scheduling of abnormal traffic behaviors, the efficiency of traffic flow is improved, traffic congestion is reduced, and the operation of the entire traffic network is optimized.

[0028] Step S4: Generate a lane scheduling report using the intelligent traffic scheduling result.

[0029] Specifically, according to the intelligent traffic scheduling result obtained in step S3, using report generation software or templates, the lane-related information in the intelligent traffic scheduling result is sorted and formatted to form a report that is easy to understand and execute, namely the lane scheduling report. This lane scheduling report is a report after adjusting the lane usage situation, including information such as scheduling measures and optimization effects. For example, if the intelligent traffic scheduling result is to set a certain lane as a temporary emergency lane or adjust the driving direction of the lane, the lane scheduling report will detail these adjustments, including information such as the functional changes of each lane and speed limits. These information can be conveyed to vehicle drivers and traffic management personnel through in-vehicle terminals, roadside displays, and other means.

[0030] The lane scheduling report provides clear lane usage guidance for traffic management personnel and drivers, helps to better implement the traffic scheduling plan, and improves traffic safety and efficiency.

[0031] Furthermore, the data transmission mode described in step S1 includes a first transmission mode, a second transmission mode, and a third transmission mode. Among them, the first transmission mode is the transmission mode of vehicle-to-vehicle communication, the second transmission mode is the transmission mode of vehicle-to-infrastructure communication, and the third transmission mode is the transmission mode of vehicle-to-cloud communication.

[0032] Specifically, the transmission modes described in the embodiments of the present application include three types: The first transmission mode (V2V, Vehicle-to-Vehicle) is the direct communication between vehicles without relying on infrastructure. For example, when a vehicle detects a sudden brake ahead, it can immediately send this information to the vehicle behind to reduce the risk of rear-end collisions; The second transmission mode (V2I, Vehicle-to-Infrastructure) is the data exchange between vehicles and traffic infrastructure (such as traffic lights, cameras, roadside units RSU). For example, traffic lights can send the remaining countdown time to approaching vehicles so that drivers or autonomous driving systems can adjust their speeds in advance. The third transmission mode (V2C, Vehicle-to-Cloud) is for vehicles to communicate with cloud servers through wireless networks (such as 4G, 5G). For example, vehicles upload their driving trajectories and fuel consumption information to the cloud, and the cloud gives optimal route planning suggestions after analysis.

[0033] The first transmission mode (V2V) mainly relies on DSRC (Dedicated Short Range Communications) or C-V2X (Cellular Vehicle-to-Everything) for communication, allowing vehicles to directly exchange information within a range of 300 meters to 500 meters without the support of a base station or the Internet. The exchanged information includes the dynamic data of the vehicle, such as vehicle speed, acceleration, braking status, steering information, blind spot alerts, etc. V2V communication has low latency and strong real-time performance, and is suitable for short-distance safety warnings, such as avoiding rear-end collisions, blind spot detection, and cooperative lane changes.

[0034] The second transmission mode (V2I) mainly uses RSU (Road Side Unit) and DSRC / C-V2X to achieve the interaction between vehicles and roadside infrastructure (such as traffic lights, roadside sensors, electronic road signs, etc.). Vehicles can obtain data from the infrastructure side (such as signal light status, speed limit information, weather conditions, construction alerts, etc.) or provide their own data to it. V2I communication enables vehicles to better perceive the status of the infrastructure, improves the intelligent level of traffic management, and facilitates the infrastructure (such as signal lights) to better adjust traffic management strategies.

[0035] The third transmission mode (V2C) mainly uses 4G / 5G cellular networks, satellite communications, etc. to achieve remote data exchange between vehicles and the cloud. Vehicles can upload a large amount of data collected by themselves (such as vehicle location information, driving trajectory, sensor data, etc.) to the cloud. The cloud stores, analyzes, and processes these data, and then returns the processing results (such as road condition analysis results, navigation suggestions, etc.) to the vehicles. V2C communication can provide global information and improve the path optimization ability of vehicles.

[0036] Through these three transmission modes, various data scattered in different vehicles, different infrastructure, and the cloud are aggregated together. For example, vehicle A learns about the speed and distance of surrounding vehicles through V2V, obtains the signal light status at the front intersection through V2I, and gets the road condition trend of the entire area from the cloud through V2C. These data are fused and processed locally or in the cloud. The combination of multiple transmission modes enables the collected data to be comprehensively and timely aggregated, providing a rich data basis for obtaining an accurate global road condition assessment, thereby improving the accuracy and real-time performance of road condition assessment.

[0037] Furthermore, real-time road information is collected based on in-vehicle sensors, real-time road facilities are perceived through infrastructure sensing devices, and vehicle driving status and external environment data are collected using intelligent mobile terminals, and the real-time road information, the real-time road facilities, the vehicle driving status, and the external environment data are added to the data collection result.

[0038] Specifically, in the entire data collection system, the vehicle collects its own driving status and surrounding environment information in real time through on-vehicle sensors to obtain real-time road information. The on-vehicle sensors include, but are not limited to, cameras, radars, GPS or Beidou positioning systems, IMU (Inertial Measurement Unit) speed sensors, brake sensors, etc. For example, cameras are used to identify lane lines, traffic signs, vehicles ahead, etc. Radars can detect the distance and relative speed between the vehicle and surrounding objects (such as other vehicles, obstacles, etc.). The speed sensor and IMU (Inertial Measurement Unit) can measure the vehicle's driving speed, acceleration, and angular velocity. The GPS or Beidou positioning system provides the vehicle's real-time position information. The brake sensor detects the braking state of the vehicle to determine whether it is in an emergency braking situation. At the same time, infrastructure sensing devices perform real-time sensing of road facilities. Infrastructure sensing devices are devices installed on road infrastructure (such as traffic signal poles, roadside base stations, etc.) for real-time sensing of the status of road facilities and traffic flow information, including traffic cameras, geomagnetic sensors, lidar, intelligent traffic signals, etc. Traffic cameras are used to capture real-time images of the road and analyze traffic flow and congestion conditions. Geomagnetic sensors are used to detect the number of vehicles passing through and their speeds to assist in analyzing traffic flow. Lidar is used to measure the three-dimensional structure of the road and detect road surface damage. Intelligent traffic signals detect the traffic flow at intersections through sensors and dynamically adjust the signal timing. In addition, intelligent mobile terminals (such as smartphones, tablets, etc.) serve as auxiliary tools for data collection, collecting vehicle driving status and external environment data. For example, it can determine whether the vehicle is accelerating or braking suddenly through an accelerometer, determine the vehicle's driving trajectory and position through a positioning system, and at the same time collect external environment data through the network, including the current weather conditions, accidents, construction, congestion, etc.

[0039] Fuse these collected real-time road information, real-time road facility information, vehicle driving status, and external environment data and add them to the data collection results for subsequent road condition assessment and intelligent scheduling. Through multi-source data collection by on-vehicle sensors, infrastructure sensing devices, and intelligent mobile terminals, comprehensive, accurate, and real-time road condition information is obtained, providing more sufficient data support for subsequent accurate road condition assessment, traffic scheduling, and other operations, thereby improving the accuracy and effectiveness of traffic scheduling.

[0040] Furthermore, step S1 includes:

[0041] Step S11: Transmit the data collection results through the data transmission mode to obtain transmitted data.

[0042] Step S12: Perform data denoising, feature extraction, and compressed transmission on the transmitted data at the edge end to obtain preprocessed data.

[0043] Step S13: Input the preprocessed data into a road condition evaluation model obtained through BP neural network learning in the cloud, and analyze and obtain the global road condition evaluation.

[0044] Specifically, after data collection, the data collection results are transmitted to target nodes (such as other vehicles, roadside units, or the cloud) using data transmission modes such as vehicle-to-vehicle, vehicle-to-infrastructure, and vehicle-to-cloud. For example, a vehicle sends data such as its own vehicle speed and driving direction to surrounding vehicles through vehicle-to-vehicle communication, and at the same time sends data to a roadside base station through vehicle-to-infrastructure communication, and then uploads a large amount of data to the cloud through vehicle-to-cloud communication. Through multiple data transmission modes, it is ensured that data can be effectively transmitted between vehicles, infrastructure, and the cloud, realizing data sharing and circulation, and providing a data basis for subsequent processing.

[0045] After the target node obtains the transmitted data, it is first processed at the edge (such as in-vehicle devices or local edge computing servers close to the vehicle). Interference information in the data is removed through data denoising algorithms (such as mean filtering, median filtering, etc.) to improve the quality of the data. Then, the principal component analysis (PCA) algorithm is used for feature extraction to mine valuable features for road condition evaluation from the original data, such as extracting features of peak and trough periods from a series of vehicle speed and traffic flow data. Feature extraction can simplify data representation, highlight key information, and reduce the amount of calculation. Then, compression algorithms such as JPEG and H.264 are used for compressed transmission to compress the processed data to reduce the amount of data transmitted and improve transmission efficiency.

[0046] After the preprocessed data is transmitted to the cloud, it is used as input and fed into a road condition evaluation model obtained through BP neural network learning. This road condition evaluation model is pre-trained with a large amount of road condition data and can perform calculations based on the input preprocessed data (such as features like traffic flow, vehicle speed, and road facility status). Inside the model, the preprocessed data undergoes calculations and processing by multiple layers of neurons in the neural network to generate a global road condition evaluation result containing information such as congested areas, accident locations, and road construction.

[0047] The specific process of training a road condition assessment model based on a BP neural network is as follows: First, historical data collected by multi-source data collection methods such as in-vehicle sensors, infrastructure perception devices, and intelligent mobile terminals is obtained, including historical road information, historical road facility information, historical vehicle driving states, and historical external environment data. The collected data is processed such as denoising, feature extraction, and compressed transmission. The preprocessed historical data is divided into a training set and a test set. The training set is used to train the neural network, and the test set is used to evaluate the performance of the model. The BP neural network architecture is set using a deep learning framework. The BP neural network includes an input layer, a hidden layer, and an output layer. The backpropagation algorithm is used for training, that is, according to the error of the output layer, the error is backpropagated to the hidden layer and the input layer through the chain rule, and the weights and thresholds of each layer are adjusted to minimize the sum of squared errors of the network. The trained model is evaluated using the test set, and evaluation metrics such as mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) are used to evaluate the model performance. According to the evaluation results, the model is adjusted, such as adjusting the network structure, optimizing algorithm parameters, increasing training data, etc., until the model performance meets the requirements, and a road condition assessment model is obtained. The road condition assessment model based on the BP neural network can effectively evaluate road conditions and predict traffic flow through reasonable data preprocessing, network structure design, and training process. This model has strong nonlinear processing ability and self-adaptability, and can provide a reliable basis for subsequent traffic management and decision-making.

[0048] Further, as Figure 2 shown, step S2 includes:

[0049] Step S21: Conduct visual analysis on the global road condition assessment to identify traffic anomalies, where the traffic anomalies include traffic accidents, traffic congestion, road surface damage, and bad weather.

[0050] Step S22: Use the traffic accidents, traffic congestion, road surface damage, and bad weather to perform spatio-temporal data modeling to obtain the predicted traffic flow trend.

[0051] Step S23: Use the isolated forest to analyze the predicted traffic flow trend to identify the abnormal traffic behaviors.

[0052] Specifically, through image processing and computer vision technologies, visual information is extracted from the global road condition assessment to identify traffic anomalies such as traffic accidents, traffic congestion, road surface damage, and bad weather. Exemplarily, real-time image data is obtained from the global road condition assessment, and the image is preprocessed, such as grayscale conversion, binarization, denoising, etc., to improve the image quality. Edge detection (such as the Canny algorithm) is used to extract edge information in the image, and object detection algorithms (such as YOLO, SSD) are used to identify vehicles, pedestrians, obstacles, and weather conditions such as rain, snow, and fog in the image. Potholes, cracks, etc. are discovered from the road inspection images. Through visual analysis, traffic anomalies can be comprehensively and intuitively discovered from the global road condition assessment, providing basic information for subsequent traffic management and decision-making.

[0053] After identifying traffic anomalies such as traffic accidents, traffic congestion, road surface damage, and bad weather, the data of these traffic anomalies are modeled in the time and space dimensions to analyze the changing trend of traffic flow. Taking traffic congestion as an example, the area where congestion occurs is determined in space (such as the intersection of several streets), and information such as the start time and duration of congestion is recorded in time to construct a spatio-temporal data model. Through this spatio-temporal data model, the trend of traffic flow can be predicted, such as predicting how long the congestion will spread to the surrounding area, or under what circumstances the congestion will ease, etc. Exemplarily, ARIMA (Autoregressive Integrated Moving Average Model), LSTM (Long Short-Term Memory Network), and Graph Neural Network (GNN) can be used for modeling. Spatio-temporal data modeling can fully consider the spatio-temporal characteristics of traffic data and more accurately predict the traffic flow trend, which helps to formulate traffic management strategies in advance.

[0054] Analyze the predicted traffic flow trend using the Isolation Forest algorithm: Input the data points in the predicted traffic flow trend into the Isolation Forest algorithm. Since abnormal traffic behaviors have significant differences in data characteristics from normal traffic flow, the Isolation Forest can quickly isolate these abnormal points, thereby identifying abnormal traffic behaviors. For example, if the traffic volume in a certain area in the predicted traffic flow trend suddenly shows extremely low or extremely high conditions that do not match the surrounding areas and historical data, the Isolation Forest algorithm will identify this situation as an abnormal traffic behavior, improving the accuracy and efficiency of identifying abnormal traffic behaviors and providing a basis for taking timely countermeasures.

[0055] Furthermore, based on the vehicle driving trajectory, vehicle violation behaviors are judged and added to the abnormal traffic behaviors, where the vehicle violation behaviors include speeding, sudden braking, and occupying the emergency lane.

[0056] Specifically, in addition to abnormal traffic behaviors caused by traffic anomalies, it is also necessary to detect and identify violation behaviors such as speeding, sudden braking, and occupying the emergency lane, and incorporate them into the analysis of abnormal traffic behaviors to enhance the monitoring and management capabilities of road traffic safety.

[0057] First, a series of position points passed by the vehicle during driving are obtained through an in-vehicle GPS device or Beidou positioning system, and then these position points are connected to obtain the vehicle driving trajectory. Then, vehicle violation behaviors are judged based on the vehicle driving trajectory. For speeding determination, first, the speed information of the vehicle is calculated according to the change of the vehicle position over time. Then, the vehicle speed is compared with the speed limit stipulated for the current road. If the vehicle speed is greater than the stipulated speed limit, it is determined that the vehicle has a speeding behavior and is added to the abnormal traffic behaviors. The sudden braking determination is also based on the speed data in the vehicle driving trajectory. The speed change value is obtained by subtracting the speeds at adjacent time points, and then divided by the time interval to obtain the speed change rate. If the speed drops significantly within a short time interval, such as the speed change rate exceeds a certain threshold (e.g., the speed drops by more than 10 km / h per second), it can be determined as a sudden braking behavior and added to the abnormal traffic behaviors. For the determination of occupying the emergency lane, it is necessary to judge whether the vehicle is within the range of the emergency lane according to the position information in the vehicle driving trajectory. If the position points in the vehicle driving trajectory fall within the area of the emergency lane and the vehicle is not an emergency rescue vehicle, it is determined that the emergency lane is occupied and this behavior is added to the abnormal traffic behaviors.

[0058] Through driving trajectory analysis, violation behaviors can be identified in real time. Combining machine learning and sensor fusion technologies can reduce the false alarm rate, enhance detection reliability, and provide accurate violation data for intelligent traffic scheduling to facilitate the optimization of traffic scheduling strategies.

[0059] Further, step S3 includes:

[0060] Step S31: Generate a dynamic road condition map through the abnormal traffic behaviors, where the dynamic road condition map includes visual road condition information.

[0061] Step S32: Perform reinforcement learning on the dynamic road condition map to obtain optimized signal light timing and adjusted traffic flow directions.

[0062] Step S33: Perform path optimization according to the optimized signal light timing and adjusted traffic flow directions to obtain a path optimization result.

[0063] Step S34: Use the autonomous driving system to execute the vehicle-road collaborative control of the path optimization result to obtain the intelligent traffic scheduling result.

[0064] Specifically, integrate the abnormal traffic behaviors (such as traffic congestion, traffic accidents, vehicle violations, etc.) identified in the aforementioned steps into the map data, and use geographic information system (GIS) technology and visualization technology to present the road condition information in an intuitive way, generating a dynamic road condition map. This dynamic road condition map can reflect the real-time traffic situation. On this map, different traffic conditions are presented in a visual way. For example, different colors are used to represent different degrees of congestion (red represents severe congestion, yellow represents mild congestion, and green represents smooth traffic), and special icons are used to mark the locations of traffic accidents or road damages, etc.

[0065] Take the dynamic road condition map as the environment of reinforcement learning, and the traffic dispatching system as the agent, and try different signal timings and traffic flow adjustment strategies. For example, increase the green light duration of the traffic lights on the congested roads to guide vehicles to divert to the roads with less traffic flow. Use traffic efficiency indicators (such as increased average vehicle speed, reduced congested roads, etc.) as reward feedback, continuously optimize the strategies, and finally obtain the optimized signal timings and adjusted traffic flows.

[0066] Based on the optimized signal timings and traffic flow adjustments, optimize the vehicle routes. Use route planning algorithms (such as Dijkstra algorithm, A-star algorithm, etc.), consider factors such as road congestion and signal waiting time, calculate the best driving route from the vehicle's starting point to the end point, and obtain the route optimization result.

[0067] Transmit the route optimization result to the vehicle's autonomous driving system. The autonomous driving system, according to the route optimization result, conducts collaborative control with the road infrastructure. For example, the vehicle adjusts its speed according to the signal status to avoid unnecessary stops; the road infrastructure (such as intelligent traffic lights) further fine-tunes the signal timings according to the approaching speed and quantity of vehicles. Through this vehicle-road collaborative control, finally obtain the intelligent traffic dispatching result and achieve the efficient operation of traffic.

[0068] The above steps, through the generation of dynamic road condition maps, reinforcement learning optimization, route optimization, and vehicle-road collaborative control, realize the complete process from abnormal traffic behavior detection to intelligent traffic dispatching, not only improving the intelligent level of traffic dispatching, but also significantly enhancing traffic efficiency and safety through real-time optimization and collaborative control.

[0069] Furthermore, optimize the road condition assessment through a feedback mechanism, where the feedback mechanism includes navigation recommendations based on user behavior analysis and a closed-loop optimization mechanism for signal control.

[0070] Specifically, in the process of intelligent traffic scheduling, relying solely on one-time data analysis and optimization has limitations. Therefore, a feedback mechanism is needed to continuously adjust and optimize the road condition assessment to improve the accuracy of traffic scheduling. This feedback mechanism mainly includes navigation recommendations based on user behavior analysis and a closed-loop optimization mechanism for signal light control.

[0071] Feedback based on navigation recommendations from user behavior analysis first collects user behavior data from the navigation system, including information such as the user's departure location, destination, selected route, travel time, speed on different sections of the road, etc. Data mining and machine learning techniques are used to analyze the user behavior data to discover potential road condition problems. For example, users with similar behaviors are grouped through clustering analysis, or decision trees are used to analyze the relationship between user behavior and road conditions. If a large number of users avoid a certain recommended route, it is very likely that the actual road conditions of this route are worse than the assessment (such as undetected congestion or road construction); if the driving speed of users on a certain route is significantly lower than expected, it may also indicate an error in the road condition assessment. Based on the results of user behavior analysis, the road condition assessment is adjusted. If it is found that the actual road conditions of a certain route do not match the assessment, the road condition assessment parameters of this route (such as congestion level, estimated travel time, etc.) are corrected.

[0072] Feedback based on signal light control and traffic condition monitoring first performs signal light control, adjusting the signal timing according to the current traffic flow and road condition information. At the same time, traffic sensors (such as loop detectors, video detectors, etc.) are used to closely monitor changes in traffic conditions, including changes in traffic volume, vehicle speed, and congested sections. Special traffic data analysis software, such as TransCAD, is used to analyze the feedback information of traffic conditions after signal light control. If the traffic volume in a certain direction increases but the vehicle speed decreases after adjusting the signal timing, it means that the signal timing is still unreasonable and needs further adjustment. Or if the congestion situation in a certain congested section is alleviated after signal light adjustment, it indicates that the current signal light control strategy is effective. Based on the feedback analysis results of signal light control, the road condition assessment is optimized. If the traffic conditions in a certain area are improved due to signal light control, then the congestion level, traffic fluency, etc. indicators of this area are adjusted accordingly in the road condition assessment. On the contrary, if the traffic conditions deteriorate, the road condition factors in this area are re-evaluated.

[0073] Through the navigation recommendations from user behavior analysis and the closed-loop optimization mechanism for signal light control, deviations in road condition assessment can be detected in a timely manner and corrected, making the road condition assessment more in line with the actual traffic situation, and providing a more reliable basis for traffic management decisions, navigation recommendations, etc.

[0074] In summary, the collaborative intelligent road condition monitoring method for the vehicle network provided by the embodiments of this application has the following technical effects:

[0075] In the embodiments of the present application, through the fusion of multi-modal data collection and transmission, vehicle-to-vehicle, vehicle-to-infrastructure, and vehicle-to-cloud communications are utilized to enrich the data sources and improve the real-time and comprehensiveness of the monitoring data. Data preprocessing, feature extraction, and compression are performed at the edge to reduce transmission latency and improve data utilization efficiency. The cloud uses a BP neural network model for deep learning to improve the accuracy of the global road condition assessment. Visual analysis and spatio-temporal data modeling are performed on the global road condition assessment, and the Isolation Forest is used to identify abnormal traffic behaviors, realizing the precise monitoring and early warning of traffic abnormal events. A dynamic road condition map is generated based on the abnormal behaviors, and the signal timing and traffic flow are optimized by combining reinforcement learning. A dynamic road condition map is generated and the path is optimized, and intelligent scheduling is achieved through vehicle-road collaborative control. In addition, user behavior analysis and closed-loop optimization of signal control are introduced as feedback mechanisms to continuously optimize the road condition assessment and scheduling strategies. Overall, the embodiments of the present application improve the accuracy of road condition monitoring, increase the response speed of traffic scheduling, and achieve more intelligent and efficient traffic management.

[0076] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A collaborative intelligent road condition monitoring method for Internet of Vehicles, characterized in that: include: The data collection results are transmitted and integrated through the data transmission mode to obtain a global road condition assessment; Performing abnormal identification on the global road condition assessment to obtain abnormal traffic behavior; Performing road condition assessment and intelligent dispatching according to the abnormal traffic behavior to obtain intelligent traffic dispatching results; A lane dispatch report is generated using the intelligent traffic dispatch result.

2. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: The data transmission mode includes a first transmission mode, a second transmission mode and a third transmission mode, wherein the first transmission mode is a transmission mode for vehicle-to-vehicle communication, the second transmission mode is a transmission mode for vehicle-to-infrastructure communication, and the third transmission mode is a transmission mode for vehicle-to-cloud communication.

3. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: Real-time road information is collected based on vehicle-mounted sensors, real-time road facility perception is performed through infrastructure sensing equipment, and vehicle driving status and external environment data are collected using smart mobile terminals. The real-time road information, real-time road facilities, vehicle driving status and external environment data are added to the data collection results.

4. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: The data collection results are transmitted and integrated through the data transmission mode to obtain a global road condition assessment, including: Performing data transmission on the data collection result through the data transmission mode to obtain transmission data; Performing data denoising, feature extraction and compression transmission on the transmission data at the edge to obtain pre-processed data; The pre-processed data is input into a road condition assessment model obtained by learning a BP neural network in the cloud, and the global road condition assessment is obtained by analysis.

5. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: Performing abnormal identification on the global road condition assessment to obtain abnormal traffic behavior includes: Performing visual analysis on the global road condition assessment to identify traffic anomalies, wherein the traffic anomalies include traffic accidents, traffic congestion, road damage and bad weather; Using the traffic accidents, traffic congestion, road damage and bad weather to perform spatiotemporal data modeling, a predicted traffic flow trend is obtained; The predicted traffic flow trend is analyzed using isolation forest to identify the abnormal traffic behavior.

6. The method for collaborative intelligent road condition monitoring for Internet of Vehicles according to claim 1, characterized in that: The vehicle violation behavior is determined based on the vehicle driving trajectory and added to the abnormal traffic behavior, wherein the vehicle violation behavior includes speeding, sudden braking and occupying the emergency lane.

7. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: According to the abnormal traffic behavior, road condition assessment and intelligent dispatch are performed to obtain intelligent traffic dispatch results, including: Generating a dynamic traffic map through the abnormal traffic behavior, wherein the dynamic traffic map includes visualized traffic information; Reinforce learning is performed on the dynamic traffic map to optimize the timing of traffic lights and adjust the traffic flow direction; Optimize the route according to the optimized signal light timing and adjusted traffic flow direction to obtain a route optimization result; The automatic driving system is used to execute the vehicle-road cooperative control of the path optimization result to obtain the intelligent traffic scheduling result.

8. The collaborative intelligent road condition monitoring method for Internet of Vehicles according to claim 1, characterized in that: The road condition assessment is optimized through a feedback mechanism, wherein the feedback mechanism includes a closed-loop optimization mechanism of navigation recommendation based on user behavior analysis and traffic light regulation.

Images