Method and device for determining traffic flow, and nonvolatile storage medium

Through multi-source data fusion and machine learning technology, the accuracy and real-time problems of vehicle positioning monitoring systems in complex environments and extreme weather are solved, accurate traffic flow monitoring and management decisions are achieved, and traffic management efficiency is improved.

CN119007443APending Publication Date: 2024-11-22CHINA TELECOM CORP LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411148139.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing vehicle positioning monitoring system is insufficient in positioning accuracy and real-time under complex urban environments and extreme weather conditions, and cannot effectively process massive traffic data, affecting the efficiency of traffic management decision-making.

Method used

By obtaining vehicle position data, video data and social platform road conditions information, using multiple sensors and computer vision technologies to integrate data, generate accurate three-dimensional models and driving trajectories, combine machine learning and data mining technology to analyze traffic flow, and optimize traffic light control and traffic flow guidance strategy.

Benefits of technology

It improves the accuracy and real-timeness of vehicle positioning, accurately determines traffic flow, improves the response speed and decision-making efficiency of the traffic management system, and adapts to complex environments and extreme weather changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007443A_ABST
    Figure CN119007443A_ABST
Patent Text Reader

Abstract

The invention discloses a traffic flow determination method and device, and a nonvolatile storage medium. The method comprises the following steps: acquiring original data related to a vehicle; identifying a vehicle position in the video data to obtain third position data; according to different preset weights, performing fusion processing on the first position data, the second position data and the third position data to obtain positioning data; acquiring a preset three-dimensional model, and updating the preset three-dimensional model by using the video data and the positioning data of the road where the vehicle is located to obtain a target three-dimensional model; and analyzing the positioning data by using the target three-dimensional model, generating a driving track of the vehicle, and determining the traffic flow of the target area where the road is located according to the road condition information of the road and the driving track of the vehicle. The technical problem that the real-time condition of the traffic flow cannot be accurately determined due to the fact that the accuracy and the real-time performance of vehicle positioning are insufficient in the prior art is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of intelligent transportation systems, and in particular, to a method and device for determining traffic flow, and a non-volatile storage medium. Background Art

[0002] In today's society, with the acceleration of urbanization and the sharp increase in the number of vehicles, effective vehicle positioning and monitoring has become particularly critical. Internet of Things (IoT) technology, with its highly interconnected devices and advanced data processing capabilities, provides a revolutionary solution for vehicle monitoring, which can collect and analyze key information such as vehicle location and speed in real time, thereby optimizing traffic flow management, improving road safety and improving driving experience.

[0003] However, vehicle positioning and monitoring systems face multiple challenges in practical applications. First, insufficient positioning accuracy and real-time performance are a major problem, especially in complex urban environments, where high-rise buildings and dense urban structures often cause global positioning system (GPS) signal loss or interference, affecting positioning accuracy. Secondly, the limited integration and processing capabilities of traffic data make it impossible to efficiently process massive amounts of data from various resources, which in turn affects the response speed and decision-making efficiency of the traffic monitoring system. Thirdly, the poor performance of existing systems under extreme weather conditions and the lack of effective user participation mechanisms have limited the comprehensive application of vehicle monitoring systems.

[0004] To address these issues, the existing technology (Chinese invention patent: CN116978229A) mainly relies on traditional GPS positioning technology and attempts to improve performance by increasing ground monitoring points and improving server processing capabilities. However, these measures are still insufficient in solving positioning accuracy, real-time performance, and data processing capabilities. In particular, these existing technologies fail to provide fundamental solutions to address positioning challenges in complex urban environments, real-time processing of large-scale traffic data, and adaptation to extreme weather changes.

[0005] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiments of the present application provide a method and device for determining traffic flow, and a non-volatile storage medium, so as to at least solve the technical problem that the real-time status of traffic flow cannot be accurately determined due to the insufficient accuracy and real-time performance of vehicle positioning in related technologies.

[0007] According to one aspect of an embodiment of the present application, a method for determining traffic flow is provided, comprising: obtaining original data related to a vehicle, wherein the original data comprises: the vehicle's position data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the vehicle's position data comprises: first position data uploaded by a global positioning system of a vehicle-mounted terminal and second position data uploaded by a geomagnetic sensor; identifying the vehicle's position in the video data to obtain third position data; fusing the first position data, the second position data, and the third position data according to different preset weights to obtain positioning data; obtaining a preset three-dimensional model, and using the video data of the road where the vehicle is located and the positioning data to update the preset three-dimensional model to obtain a target three-dimensional model, wherein the preset three-dimensional model comprises: the vehicle and the road where the vehicle is located; using the target three-dimensional model to analyze the positioning data, generate a driving trajectory of the vehicle, and determine the traffic flow in a target area where the road is located according to the road condition information and the driving trajectory of the vehicle.

[0008] Optionally, obtaining raw data related to the vehicle includes: obtaining first location data from a first preset interface of the vehicle-mounted terminal based on a first predefined interface protocol; obtaining second location data from a second preset interface of the vehicle-mounted terminal based on a second predefined interface protocol; obtaining preset keywords, searching the social platform for target social platform content whose publishing location is in a target area and whose social platform content includes preset keywords, wherein the target area is an area where roads are located; if the target social platform content is text content, using natural language processing technology to identify road condition information in the text content; if the target social platform content contains picture content and video content associated with the text content, using computer vision technology to identify the content in the picture. The method further comprises: fusing the vehicle's location data, video data, and road condition information; using machine learning and data mining techniques to identify at least one of the following from the fused data: traffic flow rate information, congestion pattern information, and accident location information to obtain an identification result; sending the identification result to the first information publishing platform, and if the traffic flow rate in the identification result is greater than a first preset threshold and / or the number of accidents at the same location is greater than a second preset threshold, sending the identification result to the second information publishing platform.

[0009] Optionally, after obtaining the original data related to the vehicle, the method also includes: preprocessing the original data by the following method: removing invalid data and erroneous data in the original data, and normalizing the original data; classifying the normalized original data according to the data source and data type of the original data to obtain a classification result; after preprocessing the original data, performing data analysis on the original data, and when it is analyzed that the vehicle's position data, video data, and road condition information all include the same target location where the accident occurred, determining the accident information of the target location, and generating alarm information including the accident information, wherein the accident information includes at least one of the following: the location where the accident occurred, and the time when the accident occurred; after performing data analysis on the original data, determining the label information of the original data according to the classification result, and storing the original data in a unified data processing platform based on the label information of the original data.

[0010] Optionally, identifying the vehicle position in the video data to obtain third position data includes: acquiring a first image including the vehicle in the video data, and performing image enhancement processing on the first image to obtain a target image when the brightness of the first image is less than a first preset brightness and / or the contrast of the first image is less than a first preset contrast; and identifying the vehicle position in the target image through an image recognition model to obtain third position data.

[0011] Optionally, according to different preset weights, the first position data, the second position data and the third position data are fused and processed to obtain positioning data, including: respectively converting the first position data, the second position data and the third position data into vector form to obtain a first vector corresponding to the first position data, a second vector corresponding to the second position data and a third vector corresponding to the third position data; when there is missing data in the first position data, the second vector is determined as the system state vector, the first vector and the third vector are determined as observation vectors, and according to different preset weights, the system state vector and the observation vector are Kalman filtered to obtain positioning data; when there is missing data in the second position data, the first vector is determined as the system state vector, the second vector and the third vector are determined as observation vectors, and according to different preset weights, the system state vector and the observation vector are Kalman filtered to obtain positioning data; when there is missing data in the third position data, the first vector is determined as the system state vector, the second vector and the third vector are determined as observation vectors, and according to different preset weights, the system state vector and the observation vector are Kalman filtered to obtain positioning data.

[0012] Optionally, the preset three-dimensional model is updated using the video data and positioning data of the road where the vehicle is located to obtain a target three-dimensional model, including: integrating the positioning data into the preset three-dimensional model to obtain a first three-dimensional model; performing feature extraction on the content in the video frame of the video data to obtain a first target feature, wherein the first target feature includes at least one of the following: vehicle position, vehicle speed, vehicle driving direction, pedestrian position, road signs, traffic signals; using a deep learning network to determine whether the first target feature has changed compared to the feature in the first three-dimensional model, and when it is determined that the first target feature has changed compared to the feature in the first three-dimensional model, updating the changed content in the first three-dimensional model to obtain a target three-dimensional model.

[0013] Optionally, the first position data, the second position data and the third position data are fused according to different preset weights to obtain positioning data, including: obtaining weather data, performing feature extraction on the weather data to obtain a second target feature, wherein the second target feature includes at least one of the following: lighting time, temperature, humidity, rainfall, wind speed; determining the correlation between the weather data and the vehicle's positioning accuracy index according to the second target feature; adjusting the different preset weights of the first position data, the second position data and the video data in the fusion process according to the correlation, and fusing the first position data, the second position data and the video data of the road where the vehicle is located according to the adjusted weights to obtain positioning data.

[0014] Optionally, according to the association relationship, different preset weights of the first position data, the second position data and the video data in the fusion process are adjusted, including: when the association relationship indicates that the rainfall is negatively correlated with the positioning accuracy index, the weight of the video data in the fusion process is reduced, and the weight of the first position data and the second position data in the fusion process is increased; when the association relationship indicates that the wind speed is negatively correlated with the positioning accuracy index, the weight of the second position data in the fusion process is reduced, and the weight of the first position data and the third position data in the fusion process is increased; when the association relationship indicates that the temperature is negatively correlated with the positioning accuracy index, the weight of the first position data in the fusion process is reduced, and the weight of the second position data and the third position data in the fusion process is increased.

[0015] Optionally, after determining the correlation between the weather data and the vehicle's positioning accuracy index based on the second target feature, the method further includes: sending a first instruction to the Internet of Things system where the image acquisition device is located to adjust the exposure time of the image acquisition device based on the correlation.

[0016] Optionally, after analyzing the positioning data using the target three-dimensional model to generate the vehicle's driving trajectory, the method also includes: inputting the vehicle's driving trajectory into a machine learning model to obtain an output result of the machine learning model, wherein the output result includes at least: the vehicle's speed, the road's congestion index, and the road's congestion location; sending the road's congestion location to an information release platform, and sending a second instruction to a control platform for controlling traffic lights based on the vehicle's speed and the road's congestion index to control the traffic lights; generating a traffic management optimization suggestion report based on the output results, wherein the traffic management optimization suggestion report includes at least one of the following: bus route optimization suggestions, road construction optimization suggestions, traffic light control timing optimization suggestions, information on vehicles with speeds greater than a preset speed, and information on roads where the number of accidents is greater than a preset number.

[0017] Optionally, the traffic flow in the target area where the road is located is determined based on the road condition information and the vehicle's driving trajectory, including: inputting the vehicle's driving trajectory data and the road condition information into a time series analysis model, and obtaining the traffic flow information and road congestion information of the target area output by the time series analysis model, wherein the vehicle's driving trajectory data includes at least one of the following: position, speed and driving direction, and the road condition information includes at least one of the following: traffic congestion information, accident information and road closure information, and the time series analysis model is obtained by training with historical driving trajectory data and historical road condition information; based on the traffic flow information and road congestion information of the target area, a traffic light control strategy and a traffic diversion strategy are generated, wherein the traffic light control strategy at least includes: adjusting the operating time of the traffic light and changing the operating mode of the traffic light.

[0018] According to another aspect of the embodiment of the present application, a vehicle positioning device is also provided, including: an acquisition module, used to acquire original data related to the vehicle, wherein the original data includes: the vehicle's position data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the vehicle's position data includes: first position data uploaded by the global positioning system of the vehicle-mounted terminal and second position data uploaded by the geomagnetic sensor; an identification module, used to identify the vehicle position in the video data to obtain third position data; a fusion module, used to fuse the first position data, the second position data and the third position data according to different preset weights to obtain positioning data; an update module, used to acquire a preset three-dimensional model, and use the video data and positioning data of the road where the vehicle is located to update the preset three-dimensional model to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; a determination module, used to analyze the positioning data using the target three-dimensional model, generate the vehicle's driving trajectory, and determine the traffic flow in the target area where the road is located according to the road condition information and the vehicle's driving trajectory.

[0019] According to another aspect of the embodiment of the present application, a non-volatile storage medium is provided, the storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the above method for determining traffic flow.

[0020] According to another aspect of the embodiments of the present application, there is also provided an electronic device, including: a memory and a processor, the processor being used to run a program stored in the memory, wherein the above method for determining traffic flow is executed when the program is run.

[0021] According to yet another aspect of the embodiments of the present application, a computer program is further provided, wherein when the computer program is executed by a processor, the above method for determining traffic flow is implemented.

[0022] According to another aspect of the embodiment of the present application, a computer program product is also provided, the computer program product comprising a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above method for determining traffic flow is implemented.

[0023] In an embodiment of the present application, the original data related to the vehicle is obtained, wherein the original data includes: the position data of the vehicle, the video data collected from the image acquisition device set on the road where the vehicle is located, and the road condition information of the road obtained from the social platform, and the position data of the vehicle includes: the first position data uploaded by the global positioning system of the vehicle-mounted terminal and the second position data uploaded by the geomagnetic sensor; the vehicle position in the video data is identified to obtain the third position data; according to different preset weights, the first position data, the second position data and the third position data are fused and processed to obtain the positioning data; a preset three-dimensional model is obtained, and the preset three-dimensional model is updated using the video data and the positioning data of the road where the vehicle is located to obtain the target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; the positioning data is analyzed using the target three-dimensional model to generate the vehicle's driving trajectory, and the traffic flow in the target area of ​​the road is determined according to the road condition information and the vehicle's driving trajectory, so as to achieve the purpose of improving the accuracy and real-time performance of vehicle positioning, thereby achieving the technical effect of accurately determining the real-time status of traffic flow, thereby solving the technical problem that the real-time status of traffic flow cannot be accurately determined due to the insufficient accuracy and real-time performance of vehicle positioning in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 is a flow chart of a method for determining traffic flow according to an embodiment of the present application;

[0026] Figure 2 is a flow chart of another method for determining traffic flow according to an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a system for determining traffic flow according to an embodiment of the present application;

[0028] Figure 4 is an interaction flow chart of various components in a traffic flow determination system according to an embodiment of the present application;

[0029] Figure 5 is an interaction flow chart of various components in another traffic flow determination system according to an embodiment of the present application;

[0030] Figure 6 is a structural diagram of a traffic flow determination device according to an embodiment of the present application;

[0031] Figure 7It is a hardware structure block diagram of a computer terminal according to a method for determining traffic flow in an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] According to an embodiment of the present application, a method embodiment of a method for determining traffic flow is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0035] Figure 1 is a flow chart of a method for determining traffic flow according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0036] Step S102, obtaining original data related to the vehicle, wherein the original data includes: the vehicle's location data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform. The vehicle's location data includes: first location data uploaded by the global positioning system of the vehicle-mounted terminal and second location data uploaded by the geomagnetic sensor.

[0037] Among them, image acquisition devices include cameras, and social platforms include social media applications.

[0038] The location data uploaded by the geomagnetic sensor is the geographic positioning information obtained by capturing the changes in the earth's magnetic field and processing it through algorithms. As a device that can sense changes in the earth's magnetic field, geomagnetic sensors have been widely used in many fields, from simple electronic compasses to complex indoor positioning and navigation systems. This sensor measures the components of the earth's magnetic field in three different directions, which can not only provide direction information, but also combine other algorithms and technologies to achieve seamless positioning indoors and outdoors.

[0039] Step S104, identifying the vehicle position in the video data to obtain third position data.

[0040] For example, the following method is used to identify the position of a vehicle in video data: 1. Video input: First, a video source is required, that is, the video captured by the camera. 2. Video preprocessing: Extract individual frames from the video. Use filters (such as Gaussian blur, median filter, etc.) to reduce image noise. Converting images to different color spaces (such as HSV or LAB) can more effectively handle color-related tasks. 3. Vehicle detection: Use target detection algorithms to identify vehicles in the video. Use image feature extraction methods to help identify vehicles. 4. Vehicle tracking: Once a vehicle is detected, a tracking algorithm (such as Kalman filtering, optical flow, or other machine learning-based methods) can be used to track the movement of the vehicle in consecutive frames. Ensure that the same vehicle detected in consecutive frames is correctly associated. 5. Position positioning: The vehicle detection algorithm returns a bounding box that indicates the position of the vehicle in the image. Based on the coordinates of the bounding box, calculate the actual position of the vehicle in the video frame. 6. Result output: Output the detected vehicle position in an appropriate format, such as drawing a bounding box or reporting the coordinates in text form.

[0041] Step S106, fusing the first position data, the second position data and the third position data according to different preset weights to obtain positioning data.

[0042] Step S108, obtaining a preset three-dimensional model, using the video data and positioning data of the road where the vehicle is located, updating the preset three-dimensional model, and obtaining a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located.

[0043] Step S110, using the target three-dimensional model to analyze the positioning data, generate the vehicle's driving trajectory, and determine the traffic flow in the target area where the road is located based on the road condition information and the vehicle's driving trajectory.

[0044] According to the above steps, the original data related to the vehicle is obtained, wherein the original data includes: the vehicle's position data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the vehicle's position data includes: first position data uploaded by the global positioning system of the vehicle-mounted terminal and second position data uploaded by the geomagnetic sensor; the vehicle position in the video data is identified to obtain third position data; according to different preset weights, the first position data, the second position data and the third position data are fused and processed to obtain positioning data; a preset three-dimensional model is obtained, and the preset three-dimensional model is updated using the video data and positioning data of the road where the vehicle is located to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; the positioning data is analyzed using the target three-dimensional model to generate the vehicle's driving trajectory, and the traffic flow in the target area of ​​the road is determined according to the road condition information and the vehicle's driving trajectory, thereby achieving the purpose of improving the accuracy and real-time performance of vehicle positioning, thereby achieving the technical effect of accurately determining the real-time status of traffic flow.

[0045] The following Figure 1 The steps shown are exemplary and explanatory.

[0046] According to some optional embodiments of the present application, raw data related to the vehicle is obtained by the following method: based on a first predefined interface protocol, first position data is obtained from a first preset interface of the vehicle terminal; based on a second predefined interface protocol, second position data is obtained from a second preset interface of the vehicle terminal; preset keywords are obtained, and target social platform content whose publishing location is in a target area and whose social platform content includes preset keywords is searched in the social platform, wherein the target area is the area where the road is located; if the target social platform content is text content, natural language processing technology is used to identify road condition information of the road in the text content; if the target social platform content contains picture content and video content associated with the text content, computer vision technology is used to identify road condition information of the road in the picture content and video content.

[0047] Furthermore, after searching the social platform for target social platform content whose publishing location is in the target area and whose social platform content includes preset keywords, the following steps can also be performed: fusing the vehicle's location data, video data, and road condition information; using machine learning and data mining techniques to identify at least one of the following from the fused data: traffic flow change rate information, congestion pattern information, and accident location information, to obtain an identification result; sending the identification result to the first information publishing platform, and if the traffic flow change rate in the identification result is greater than a first preset threshold and / or the number of accidents occurring at the same location is greater than a second preset threshold, sending the identification result to the second information publishing platform.

[0048] It is understandable that the network connectivity capabilities of IoT devices (such as vehicle-mounted terminals) and other data sources (such as video surveillance systems and social media platforms) require the automatic collection and transmission of data through predefined interface protocols. Collect raw data from the above data sources through pre-configured interfaces. The vehicle-mounted terminal provides information such as the location and speed of the vehicle; the video surveillance system provides real-time images and videos of road conditions; and the social media platform provides user-generated traffic feedback and reports. Among them, the principle of collection using vehicle-mounted terminals is GPS technology and other sensor technologies, which are used to capture the vehicle's geographic location, speed and other information in real time. Video surveillance data collection is based on computer vision technology, which is used to obtain images and videos of road conditions. Social media data collection obtains user reports and feedback on traffic conditions through the application programming API interface.

[0049] For example, a vehicle’s GPS system sends location information every few seconds, including latitude and longitude, timestamp, and speed; at the same time, cameras at the city’s traffic monitoring center continuously monitor traffic flow at intersections, capturing images and video data; and the traffic management department’s social media accounts collect and publish road condition information, such as accident reports or construction notices.

[0050] Specifically, the traffic information of roads obtained from social platforms is obtained through the following methods: 1. Data collection and filtering: First, by setting specific keywords and topics (such as "traffic jam", "accident", "road closure", etc.), relevant user-posted content is captured in real time from social media platforms. These contents may exist in the form of text, pictures or videos. The geo-tagging function of the social media platform is used to filter out traffic feedback information in the specified area. This helps to limit the analysis focus to a specific geographical range and improve the relevance and accuracy of the data. 2. Data processing: For text content, natural language processing techniques such as sentiment analysis, topic identification, entity recognition, etc. are applied to extract valuable traffic information from the text posted by users. For example, sentiment analysis can be used to determine whether the traffic conditions posted by users are positive (such as unobstructed) or negative (such as severe congestion). For picture and video content, computer vision technology is used to identify information such as traffic signs, traffic flow, accident scenes, etc. in the image. This information can be used as supplementary data to help verify the results of text analysis. 3. Data fusion and analysis: Social media data is fused with other data sources (such as vehicle terminal data and video surveillance data) to comprehensively evaluate traffic conditions. Data fusion can help address the possible biases and limitations of a single data source. Using machine learning and data mining techniques, key information such as traffic flow trends, congestion patterns, and accident-prone locations can be identified from the fused data to provide decision support for traffic management and planning. 4. Warning and information release: Based on the above analysis results, real-time traffic warning information is generated to promptly notify traffic management departments and the public, especially when serious traffic accidents are detected or extreme weather affects traffic. Traffic information and warnings are fed back to the public through social media and other information release platforms to help drivers make timely travel decisions and improve road utilization efficiency and traffic safety.

[0051] Through the above steps, the processing and utilization of social media traffic feedback data not only enriches the data sources for traffic monitoring and analysis, but also improves the system's perception and response speed to traffic conditions, thereby effectively supporting the decision-making process of traffic management and planning.

[0052] Furthermore, after obtaining the original data related to the vehicle, the following steps may be performed:

[0053] Preprocess the original data: remove invalid data and erroneous data from the original data, and normalize the original data; classify the normalized original data according to the data source and data type of the original data to obtain the classification results.

[0054] The collected raw data is preprocessed, including data cleaning (removing invalid and erroneous data), data formatting (unifying the data format for processing), and data classification (classifying according to the type and source of the data), to obtain cleaned and formatted vehicle information data, video surveillance data, and social media traffic feedback data.

[0055] It should be noted that the noise (invalid, erroneous data) in the collected data is removed by algorithms, and the data is converted into a unified format for subsequent processing. This includes data cleaning and formatting, which is a key step before data analysis, aiming to improve the accuracy and consistency of the data.

[0056] After preprocessing the raw data, data analysis is performed on the raw data. When it is analyzed that the vehicle's position data, video data, and road condition information all include the same target location where the accident occurred, the accident information of the target location is determined, and alarm information including the accident information is generated, wherein the accident information includes at least one of the following: the location where the accident occurred and the time when the accident occurred.

[0057] Perform preliminary analysis on the preprocessed data to quickly identify obvious patterns or anomalies, such as identifying high-frequency traffic congestion points or identifying reports of emergencies from social media data. Use simple data analysis methods such as statistical analysis and pattern recognition to quickly identify key information and obvious anomalies or patterns in the data.

[0058] For example, at a busy intersection, the video surveillance system continuously collects image data of traffic flow, while the in-vehicle terminal provides speed and location information of passing vehicles. At the same time, on social media platforms, passing drivers and pedestrians may post real-time information about traffic accidents. The video surveillance data is analyzed by computer vision algorithms, which are trained to recognize specific image features of traffic accidents, such as sudden vehicle stops, changes in the shape of colliding vehicles, etc. When the algorithm identifies a possible accident scene, it immediately marks this information as a critical event. At the same time, data from the in-vehicle terminal shows that near the accident site, the speed of multiple vehicles suddenly dropped to zero, which is significantly different from the normal traffic flow pattern. By analyzing this location and speed information, the system can quickly locate the exact location and time of the accident. Preliminary data analysis also includes the retrieval and analysis of social media data. Through keyword searches (such as "traffic accident", "traffic jam", etc.) and sentiment analysis, the system can identify information posted by users related to the accident. This information can not only confirm the occurrence of the accident, but also provide a detailed description of the accident, the scope of impact, and immediate response. Based on the above analysis results, if the analysis results from the three data sources consistently point to a traffic accident at the same location, the system can quickly send an alert to the traffic management department and provide the specific location, time and possible impact range of the accident. The traffic management department can then quickly deploy emergency response measures, such as sending rescue vehicles, adjusting surrounding traffic lights, and issuing traffic guidance information to the public through social media and other communication channels.

[0059] Through the above examples, the above data analysis plays a key role in real-time monitoring systems, especially in quickly responding to traffic accidents or other emergencies. This not only improves the efficiency and response speed of traffic monitoring systems, but also enhances public safety and the effectiveness of traffic management.

[0060] After data analysis and processing of the original data, the label information of the original data is determined according to the classification results, and based on the label information of the original data, the original data is stored in a unified data processing platform.

[0061] That is to say, the data after preprocessing and preliminary analysis is stored in the database of the unified data processing platform, and indexes are created for these data for efficient retrieval and further in-depth analysis. Through the database management system, the processed data is persistently stored and indexed to speed up data retrieval. This ensures that the data can be effectively managed and quickly accessed, supporting subsequent data analysis and application needs.

[0062] Through the above steps, not only does it provide an accurate, consistent and timely data foundation for the vehicle positioning monitoring system, but it also improves the performance of the entire system by optimizing the quality and access efficiency of the data. This emphasis on data integration and preliminary processing ensures the effectiveness and reliability of the entire vehicle monitoring system, providing support for efficient traffic management and planning.

[0063] According to other optional embodiments of the present application, identifying the vehicle position in video data and obtaining third position data can be achieved by the following method: acquiring a first image including a vehicle in the video data, and performing image enhancement processing on the first image to obtain a target image when the brightness of the first image is less than a first preset brightness and / or the contrast of the first image is less than a first preset contrast; identifying the vehicle position in the target image through an image recognition model to obtain third position data.

[0064] As some optional embodiments of the present application, the preset three-dimensional model can be updated using the video data and positioning data of the road where the vehicle is located to obtain a target three-dimensional model by the following method: integrating the positioning data into the preset three-dimensional model to obtain a first three-dimensional model; performing feature extraction on the content in the video frame of the video data to obtain a first target feature, wherein the first target feature includes at least one of the following: vehicle position, vehicle speed, vehicle driving direction, pedestrian position, road signs, and traffic signals; using a deep learning network to determine whether the first target feature has changed compared to the feature in the first three-dimensional model, and when it is determined that the first target feature has changed compared to the feature in the first three-dimensional model, updating the changed content in the first three-dimensional model to obtain a target three-dimensional model.

[0065] Specifically, the following method is used to extract features from the content in the video frame of the video data: A convolutional neural network (CNN) is used to extract road features in the video data, such as lane lines, signs, and obstacles.

[0066] Convolution operation: y i ,j,k=∑ m,n,l x i+m,n,l ·w m,n,l,k +b k , where y i , j, k is the value of the kth channel at position (i, j) in the output feature map; x i+m,n,l is the value of the lth channel at position (i+m, j+n) in the input image; w m,n,l,k is the weight of the lth channel and the kth output channel of the convolution kernel at position (m, n); b k is the bias of the kth output channel.

[0067] Preferably, time series data (such as vehicle trajectories) can also be trained and predicted by a recurrent neural network (RNN) to identify dynamic changes and trends.

[0068] Specifically, we use a recurrent neural network (RNN) to process time series data: Hidden state update: h t =σ(W hh h t-1 +W xh x t +b h ), where h t is the hidden state at time t; σ is the activation function (such as tanh or ReLU); W hh is the weight matrix from hidden state to hidden state; h t-1 is the hidden state at time t-1; W xh is the weight matrix input to the hidden state; x t is the input at time t; b h Is the bias of the hidden state. Output calculation: y t =W hy h t +b y , where y t is the output at time t; W hy is the weight matrix from hidden state to output; b y is the output bias.

[0069] Furthermore, the output results of CNN and RNN are fused to form a comprehensive 3D model update framework. Based on the fused data, a dynamically adaptive 3D model update mechanism can be used to build and update the 3D model of roads and vehicles. The 3D model is dynamically adjusted and updated according to real-time data input to ensure that the model is consistent with the current actual situation. When the environment changes (such as road construction, traffic accidents), the 3D model is corrected and adjusted using real-time data and machine learning algorithms.

[0070] In some optional embodiments of the present application, the first position data, the second position data and the third position data are fused according to different preset weights to obtain positioning data, which can be achieved by the following method: the first position data, the second position data and the third position data are respectively converted into vector forms to obtain a first vector corresponding to the first position data, a second vector corresponding to the second position data and a third vector corresponding to the third position data; when there is missing data in the first position data, the second vector is determined as a system state vector, the first vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are Kalman filtered according to different preset weights to obtain positioning data; when there is missing data in the second position data, the first vector is determined as a system state vector, the second vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are Kalman filtered according to different preset weights to obtain positioning data; when there is missing data in the third position data, the first vector is determined as a system state vector, the second vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are Kalman filtered according to different preset weights to obtain positioning data.

[0071] It should be noted that the Kalman filter is an effective recursive filter used to estimate the state of a dynamic system from a series of observations. The core of the Kalman filter lies in two main steps: prediction (or time update) and update (or measurement update). In the prediction step, the state vector of the system is predicted based on the previous state estimate and the dynamic model of the system. This process involves the system state transfer matrix and process noise, which is usually assumed to be Gaussian white noise.

[0072] The Kalman filter is updated as follows. 1. Prediction step. State prediction: in, is the predicted state at time k; F is the state transfer matrix; is the estimated state at time k-1; B is the control input matrix; u k is the control input at time k.

[0073] Prediction covariance: P k|k-1 =FP k-1|k-1 F T +Q. Among them, P k|k-1 is the predicted covariance at time k; P k-1|k-1 is the estimated covariance at time k-1; Q is the process noise covariance.

[0074] 2. Update step: Kalman gain: K k =P k|k-1 H T (HPk|k-1 H T +R) -1 Among them, K k is the Kalman gain at time k; H is the observation matrix; R is the observation noise covariance.

[0075] Status Update: in, is the updated state at time k; z k is the observed value at time k.

[0076] Update covariance: P k|k =(IK k H)P k|k-1 Among them, P k|k is the updated covariance at time k.

[0077] For example, when a car is driving in a city with tall buildings and the GPS signal is interfered with, geomagnetic positioning provides auxiliary information to help correct the position deviation. At the same time, the camera in front of the vehicle captures images of road signs and surrounding buildings, and visual positioning technology further optimizes the positioning results by identifying these features. Through the Kalman filter algorithm, the system integrates the information provided by these three technologies to achieve accurate positioning in complex urban environments.

[0078] Kalman filtering is an optimal recursive filtering algorithm for linear dynamic systems. In this embodiment, new observation data is continuously integrated through two steps of prediction and update to estimate the current state of the system. In vehicle positioning, Kalman filtering can be used to integrate continuous positioning data from GPS and geomagnetic sensors to optimize the vehicle's position and speed estimation. It can effectively process the noise in the observation data and provide accurate state estimation.

[0079] By fusing the first location data, the second location data, and the third location data according to different preset weights, and then obtaining positioning data, the continuity and accuracy of vehicle positioning can be effectively guaranteed in a variety of complex environments, such as indoor parking lots, urban high-rise areas, tunnels, and other scenes where GPS signals are weak or unavailable. In addition, through accurate vehicle positioning, the system can more effectively track vehicles and analyze traffic flow, providing reliable data support for traffic management and planning.

[0080] On the other hand, the first position data, the second position data and the third position data are fused according to different preset weights to obtain positioning data, which can also be achieved by the following method: obtaining weather data, extracting features from the weather data, and obtaining second target features, wherein the second target features include at least one of the following: lighting time, temperature, humidity, rainfall, and wind speed; determining the correlation between the weather data and the vehicle's positioning accuracy index based on the second target feature; adjusting the different preset weights of the first position data, the second position data and the video data in the fusion process based on the correlation, and fusing the first position data, the second position data and the video data of the road where the vehicle is located based on the adjusted weights to obtain positioning data.

[0081] Preferably, after determining the correlation between the weather data and the vehicle's positioning accuracy index based on the second target feature, the following steps can also be performed: based on the correlation, a first instruction is sent to the Internet of Things system where the image acquisition device is located to adjust the exposure time of the image acquisition device.

[0082] The above-mentioned adjustment of different preset weights of the first position data, the second position data and the video data in the fusion process according to the association relationship can be achieved by the following method: when the association relationship indicates that the rainfall is negatively correlated with the positioning accuracy index, the weight of the video data in the fusion process is reduced, and the weight of the first position data and the second position data in the fusion process is increased; when the association relationship indicates that the wind speed is negatively correlated with the positioning accuracy index, the weight of the second position data in the fusion process is reduced, and the weight of the first position data and the third position data in the fusion process is increased; when the association relationship indicates that the temperature is negatively correlated with the positioning accuracy index, the weight of the first position data in the fusion process is reduced, and the weight of the second position data and the third position data in the fusion process is increased.

[0083] For example, weather data can be obtained through the following methods: 1. Obtain real-time weather data from meteorological sensors, including temperature, humidity, wind speed, rainfall, etc. 2. Access external weather data provided by meteorological services through API interfaces, including forecasts and current weather conditions. In addition, historical weather data can also be extracted from the database for model training and trend analysis.

[0084] Furthermore, real-time meteorological data and external weather data can be used to analyze current weather conditions. In weather data analysis, natural language processing (NLP) is used to parse text content: convert text into word vectors: Among them, v is a word vector, which represents a word in the text; d is the dimension of the word vector. It can be understood that natural language processing (NLP) is used to parse text data in social media and extract key traffic-related information. Then, compare the current weather data with historical weather data to identify abnormal weather patterns and trends. Evaluate the impact of different weather conditions on data collection and positioning accuracy, and generate a weather impact report. Finally, based on the results of the weather impact assessment, adjust the acquisition parameters of the sensor, such as the exposure time of the camera, the scanning frequency of the radar, etc. Adjust the parameters of the positioning algorithm, such as the gain coefficient of the filter, to cope with noise and interference under different weather conditions. Modify the data preprocessing and fusion strategies to ensure data quality under severe weather conditions.

[0085] Through the above methods, the camera exposure time and radar scanning frequency can be increased in heavy rain weather to enhance the ability to filter raindrops. In haze weather, the contrast and brightness of image processing can be increased to ensure that the camera can capture clear images. In hot weather, the sensor's operating frequency and sampling rate can be adjusted to prevent data distortion caused by sensor overheating.

[0086] As other optional embodiments of the present application, determining the traffic flow in the target area of ​​the road based on the road condition information and the vehicle's driving trajectory can be achieved by the following method: inputting the vehicle's driving trajectory data and the road condition information into a time series analysis model to obtain the traffic flow information and road congestion information of the target area output by the time series analysis model, wherein the vehicle's driving trajectory data includes at least one of the following: position, speed and driving direction, and the road condition information includes at least one of the following: traffic congestion information, accident information and road closure information, and the time series analysis model is obtained by training with historical driving trajectory data and historical road condition information; based on the traffic flow information and road congestion information of the target area, generating a traffic light control strategy and a traffic diversion strategy, wherein the traffic light control strategy at least includes: adjusting the operating time of the traffic light and changing the operating mode of the traffic light.

[0087] For example, the time series analysis model is ARIMA (Autoregressive Integrated Moving Average Model) or LSTM network. The CIFAS system conducts in-depth analysis of comprehensive traffic data sets to identify the real-time status and development trend of current traffic flow. These models can process and analyze time series data and effectively predict changes in traffic flow.

[0088] For example, traffic light control strategies include the following. Based on the traffic flow changes predicted by the model, the traffic light control system can adjust the green light duration in each direction to optimize the traffic flow. For example, if the model predicts that a certain intersection will have a large flow of traffic during the morning rush hour, increasing the green light duration in that direction in advance can effectively reduce congestion. The traffic management center integrates with the traffic light system, inputs the model's prediction data in real time, and adjusts the traffic light plan based on this data. This includes adjusting the green light duration, changing the operation mode of the traffic light (such as switching from a fixed duration mode to a traffic flow sensing mode), and even temporarily modifying the traffic rules at the intersection under certain circumstances. When it is predicted that a certain important traffic artery will face severe congestion during the afternoon rush hour, the traffic management center can decide to increase the green light duration of the artery in the direction of the city 30 minutes in advance, and reduce the green light duration in the direction of the city accordingly, in order to balance the traffic flow and reduce congestion.

[0089] Traffic diversion strategies include the following. By analyzing the traffic flow and congestion points predicted by the model, the traffic management department can design more effective traffic diversion plans, such as setting up temporary traffic signs to guide vehicles to detour, issuing traffic control notices, etc. Once potential congestion points or high-risk areas are identified, the traffic management center can release diversion information through various channels (such as traffic broadcasts, social media, and electronic display screens) to guide drivers to choose alternative routes in advance. At the same time, the direction of travel of some roads can be temporarily adjusted or emergency lanes can be opened to relieve traffic pressure. When the model predicts that a city center will face severe congestion during the weekend shopping peak, the traffic management department can issue early warning information on major traffic information platforms a few days in advance, advising citizens to avoid driving or choose public transportation as much as possible, and implement temporary traffic control measures on some roads during critical periods.

[0090] Through the above strategies, traffic management departments can use model prediction results to proactively respond to upcoming traffic conditions, improve road utilization efficiency, reduce traffic congestion, and enhance citizens' travel experience. This data- and model-based traffic management strategy demonstrates the great potential and value of intelligent transportation systems in modern urban management.

[0091] Figure 2 is a flow chart of another method for determining traffic flow according to an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0092] Step S201, collect raw data from vehicle terminals, video surveillance systems, social media and other sources through a unified data processing platform (UDPP).

[0093] Step S202, applying the multi-dimensional data fusion positioning system (MDFS) in UDPP, combining GPS, geomagnetic and visual positioning technology to perform data fusion processing to improve positioning accuracy.

[0094] Step S203, dynamically updating the three-dimensional models of roads and vehicles using a dynamically adaptive three-dimensional model update mechanism (DAMUM) to improve system adaptability.

[0095] Step S204: Adjust data collection parameters and processing strategies through the Intelligent Weather Adaptation System (IWAS) to optimize the data collection and positioning process under extreme weather conditions.

[0096] Step S205, using the updated three-dimensional model data and high-precision vehicle positioning data, the vehicle driving trajectory is tracked and analyzed in UDPP.

[0097] Step S206: The crowd intelligent participation traffic flow analysis system (CIFAS) combines vehicle driving trajectory data and social media traffic feedback data to conduct real-time monitoring and analysis of traffic flow.

[0098] Step S207, based on the traffic flow analysis report, the performance of each component of the system is evaluated and optimized to achieve continuous improvement of the system.

[0099] Figure 3 is a schematic diagram of a traffic flow determination system according to an embodiment of the present application, Figure 4 is an interaction flow chart of various components in a traffic flow determination system according to an embodiment of the present application, Figure 5 FIG. 1 is an interactive flow chart of various components in another traffic flow determination system according to an embodiment of the present application. Figures 2 to 5 The method for determining the traffic volume is explained in detail.

[0100] According to some optional embodiments of the present application, step S201 is specifically implemented by the following method.

[0101] Step S2011, multi-source data collection.

[0102] Step S20111, vehicle terminal data collection: collect real-time vehicle information data from the vehicle terminal, including GPS location, speed, direction, etc.

[0103] Step S20112, video surveillance data collection: obtaining real-time video streams of road and traffic conditions through a video surveillance system.

[0104] Step S20113, social media data collection: collect traffic feedback data from social media platforms through API interfaces, including user-generated content such as text, pictures and videos.

[0105] Step S2012: data preprocessing.

[0106] Step S2011, vehicle terminal data preprocessing: parsing GPS data, extracting vehicle location information, and performing denoising and format conversion.

[0107] Step S2012, video surveillance data preprocessing: use image processing technology (such as edge detection, feature extraction) to process the video data and extract key road and traffic features.

[0108] Step S2013, social media data preprocessing: using natural language processing (NLP) technology to parse text content, extract relevant traffic events and feedback information, and perform preliminary processing on image and video data.

[0109] Step S2013, data standardization.

[0110] Step S20131, vehicle terminal data standardization: converting the vehicle terminal data into a unified format and structure, such as a standardized timestamp and coordinate system.

[0111] Step S20132, video surveillance data standardization: converting the video data into a unified format to ensure the consistency of image and video data.

[0112] Step S20133, social media data standardization: converting social media data into a unified format, such as standardized text, image, and video data.

[0113] Step S2014, data filtering.

[0114] Step S20141, vehicle terminal data filtering: filter out invalid or duplicate GPS data points to ensure data quality.

[0115] Step S20142, video surveillance data filtering: remove irrelevant images and video frames through filtering algorithms to ensure the accuracy of video data.

[0116] Step S20143, social media data filtering: filter out irrelevant or duplicate social media posts and extract valuable traffic feedback information.

[0117] Step S2015, data storage.

[0118] Step S20151, storing the processed vehicle terminal data, video surveillance data and social media data in a distributed database and managing them using an efficient indexing mechanism.

[0119] Step S20152, data identification: add a timestamp and source identification to each piece of data to ensure the traceability and relevance of the data.

[0120] It should be noted that in the existing comparative scheme, the main function of the Internet of Things management platform is to receive vehicle information uploaded from the vehicle terminal. This means that the source of data is relatively single, mainly focusing on the data provided by the vehicle itself, such as location information, speed, etc. Through the unified data processing platform (UDPP) in this embodiment, not only the data of the vehicle terminal is used, but also multiple data sources such as video surveillance systems and social media are extended. This design enables UDPP to integrate various data sources and provide richer and multi-dimensional data integration, thereby enhancing the depth and breadth of data analysis. For example, social media data can provide real-time feedback on traffic accidents or changes in road conditions, while video surveillance systems can provide visual information about actual traffic flows and environmental conditions. The data integration strategy of this embodiment provides a comprehensive perspective to analyze and respond to traffic conditions. Compared with the existing scheme, it makes data processing more comprehensive and can support more complex analysis and decision-making processes. By integrating multiple data sources, this embodiment can more effectively identify and respond to complex traffic scenarios, improving the speed and accuracy of emergency response.

[0121] According to some other optional embodiments of the present application, step S202 is specifically implemented by the following method.

[0122] Step S2021, multi-source data collection.

[0123] Step S20211, GPS data collection: collect GPS positioning data from the vehicle-mounted terminal to provide basic location information.

[0124] Step S20212, geomagnetic data collection: geomagnetic data is collected through geomagnetic sensors for auxiliary positioning, especially in environments where GPS signals are limited (such as tunnels or urban areas with tall buildings).

[0125] Step S20213, use the vehicle-mounted camera or other visual sensors to collect video or image data, and obtain location information through computer vision technology (such as feature point matching, SLAM, etc.).

[0126] Step S2022, data preprocessing.

[0127] Step S20221, GPS data preprocessing: denoising and format conversion of GPS data to ensure data consistency and accuracy.

[0128] Step S20222, filtering and correcting the geomagnetic data to remove noise and interference.

[0129] Step S20223, visual data preprocessing: preprocess the visual data through image processing algorithms (such as edge detection, feature extraction) to extract key position information.

[0130] Step S2023: data fusion processing.

[0131] Step S20231, preliminary fusion: input the preprocessed GPS, geomagnetic and visual data into MDFS for preliminary fusion.

[0132] Step S20232, Kalman filtering: Apply the Kalman filtering algorithm to fuse the global positioning of GPS with the local position change of the geomagnetic sensor to optimize the position estimation.

[0133] Specifically, the update formula of the Kalman filter is as follows:

[0134] (1) Prediction step: State prediction: in, is the predicted state at time k; F is the state transfer matrix; is the estimated state at time k-1; B is the control input matrix; u k is the control input at time k; predicted covariance: P k|k-1 =FP k-1|k-1 F T +Q, where P k|k-1 is the predicted covariance at time k; P k-1|k-1 is the estimated covariance at time k-1; Q is the process noise covariance.

[0135] (2) Update step: Kalman gain: K k =P k|k-1 H T (HP k|k-1 H T +R) -1 , where K k is the Kalman gain at time k; H is the observation matrix; R is the observation noise covariance. State update: in, is the updated state at time k; z k is the observed value at time k. Update covariance: P k|k =(IK k H)P k|k-1 , where P k|k is the updated covariance at time k.

[0136] Step S20233, particle filtering: Based on the preliminary fusion, a particle filtering algorithm is used to process nonlinear and non-Gaussian distributed data to further improve positioning accuracy.

[0137] Step S2024, high-precision positioning results are generated.

[0138] Step S20241, real-time calculation: use the fused data to perform real-time calculation to generate a high-precision vehicle position.

[0139] Step S20242, positioning result correction: based on the fused high-precision data, the vehicle position is corrected in real time to ensure the accuracy and stability of positioning.

[0140] Step S2025, data output and application.

[0141] Step S20251, positioning data output: output high-precision vehicle positioning data to the unified data processing platform (UDPP) for subsequent analysis and application.

[0142] Step S20252, application example: In a complex urban environment, high-precision positioning is achieved through MDFS. For example, in a city center with tall buildings, accurate location updates are provided by combining geomagnetic and visual data; in an underground parking lot, geomagnetic and visual positioning are used to ensure accurate positioning of vehicles in an environment without GPS signals.

[0143] It should be noted that the lack of specific reference to multidimensional data fusion technology in the prior art may mean that there are limitations in processing and integrating positioning data from different sources, especially in scenarios where high-precision positioning information is required to support complex decisions. This embodiment introduces a multidimensional data fusion positioning system (MDFS), which is an advanced data processing technology that combines GPS, geomagnetic and visual positioning technologies. By fusing data from different sensors, MDFS can improve the accuracy and reliability of position data. For example, in an urban canyon environment where GPS signals are blocked, combining geomagnetic and visual data can significantly improve positioning accuracy. The multidimensional data fusion achieved by MDFS in this embodiment significantly improves positioning accuracy, which is particularly important when dealing with complex positioning situations in urban environments. The advanced fusion algorithm enables the present invention to have higher performance than existing solutions in identifying positions and tracking moving objects, especially in environments with visual occlusion or weak GPS signals, and can provide more stable and accurate positioning data. This capability is particularly critical in highly dynamic traffic environments and can support more accurate navigation, accident prevention and traffic management measures.

[0144] In some optional embodiments of the present application, step S203 is specifically implemented by the following method.

[0145] Step S2031, real-time data collection.

[0146] Step S20311, video surveillance data collection: obtain real-time video stream from the video surveillance system to capture road and traffic conditions.

[0147] Step S20312, positioning data collection: obtaining high-precision vehicle positioning data from step 202, including the real-time position and trajectory information of the vehicle.

[0148] Step S20313, sensor data collection: integrating data from other sensors (such as LIDAR, infrared sensors) to provide more environmental information.

[0149] Step S2032: data preprocessing.

[0150] Step S20321, video data preprocessing: use image processing technology (such as edge detection, feature extraction) to process the video surveillance data and extract key road and traffic features.

[0151] Step S20322, positioning data preprocessing: denoising and correcting the high-precision positioning data to ensure the accuracy of the data.

[0152] Step S20323, sensor data preprocessing: standardize the data from different sensors to ensure data consistency and availability.

[0153] Step S2033, application of machine learning algorithm.

[0154] Step S20331, data input: input the preprocessed video data, positioning data and sensor data into the machine learning model.

[0155] Step S20332, feature extraction: Use a convolutional neural network (CNN) to extract road features in the video data, such as lane lines, landmarks, and obstacles.

[0156] Convolutional neural network (CNN) extracts features: Among them, the convolution layer calculation includes: Convolution operation: y i ,j,k=∑ m,n, l x i+m,n,l ·w m,n,l,k +b k .

[0157] Among them, y i , j, k is the value of the kth channel at position (i, j) in the output feature map; x i+m,n,l is the value of the lth channel at position (i+m, j+n) in the input image; w m,n,l,k is the weight of the lth channel and the kth output channel of the convolution kernel at position (m, n); b k is the bias of the kth output channel.

[0158] Convolutional Neural Networks (CNNs) are used to extract features from video data, such as lane lines, landmarks, and obstacles.

[0159] Step S20333, model training and prediction: Use a recurrent neural network (RNN) to train and predict time series data (such as vehicle trajectories) to identify dynamic changes and trends.

[0160] Recurrent Neural Network (RNN) processing time series data: Hidden state update: h t =σ(W hh h t-1 +W xh x t +b h ), where h t is the hidden state at time t; σ is the activation function (such as tanh or ReLU); W hh is the weight matrix from hidden state to hidden state; h t-1 is the hidden state at time t-1; W xh is the weight matrix input to the hidden state; x t is the input at time t; b h is the bias of the hidden state.

[0161] Output calculation: y t =W hy h t +b y , where y t is the output at time t; W hy is the weight matrix from hidden state to output; b y is the bias of the output. Recurrent Neural Networks (RNNs) are used to process time series data, such as vehicle trajectories, and identify dynamic changes and trends.

[0162] Step S20334, model fusion: fuse the output results of CNN and RNN to form a comprehensive 3D model update framework.

[0163] Step S2034, dynamic model update.

[0164] Step S20341, model construction: based on the fused data, use the dynamically adaptive three-dimensional model update mechanism (DAMUM) to build and update the three-dimensional model of the road and the vehicle.

[0165] Step S20342, real-time update: dynamically adjust and update the three-dimensional model based on real-time data input to ensure that the model is consistent with the current actual situation.

[0166] Step S20343, model correction: when the environment changes (such as road construction, traffic accidents), the three-dimensional model is corrected and adjusted using real-time data and machine learning algorithms.

[0167] Step S20335, data output and application.

[0168] Step S203351, model data output: output the updated road and vehicle three-dimensional model data to the unified data processing platform (UDPP) for subsequent analysis and application.

[0169] Step S203352, application example: In actual traffic management, accurate road and traffic condition modeling is achieved through DAMUM. For example, during road construction, the 3D model is updated through real-time video and positioning data to provide accurate road information and detour suggestions.

[0170] It should be noted that in the prior art, the CIM platform receives three-dimensional models of roads and vehicles and builds a matching system to obtain the road conditions on which the vehicle is traveling. This method relies on pre-established models, which are usually created based on static data and are not updated frequently, which may lead to information lag in practical applications due to environmental changes (such as new buildings or road reconstruction) not being reflected in the model in a timely manner. In contrast, the dynamically adaptive three-dimensional model update mechanism (DAMUM) introduced in this embodiment is an innovative technology that uses real-time data to dynamically update the three-dimensional models of roads and vehicles through machine learning algorithms. This method allows the system to continuously learn and optimize the model based on the latest data collected (for example, through a video surveillance system), so that it can adapt to immediate changes in the environment. The real-time dynamic update capability of this embodiment significantly improves the accuracy and practicality of the three-dimensional model, enabling the system to effectively respond to environmental changes, such as the opening of new roads, traffic accidents, or temporary road closures. This flexibility is lacking in existing solutions, giving this embodiment significant advantages in real-time and adaptability.

[0171] As some optional embodiments of the present application, step S204 is specifically implemented by the following method.

[0172] Step S2041, weather data collection.

[0173] Step S20411, meteorological sensor data collection: obtain real-time weather data from meteorological sensors, including temperature, humidity, wind speed, rainfall, etc.

[0174] Step S20412, external weather data collection: Access external weather data provided by meteorological services through the API interface, including forecasts and current weather conditions.

[0175] Step S20413, historical weather data collection: extract historical weather data from the database for model training and trend analysis.

[0176] Step S2042: weather data analysis.

[0177] Step S20421, real-time weather analysis: Analyze current weather conditions using real-time meteorological data and external weather data. Specifically, in weather data analysis, natural language processing (NLP) is applied to parse text content: Convert text into word vectors: Where v is a word vector, which represents a word in the text; d is the dimension of the word vector. Natural language processing (NLP) is used to parse text data in social media and extract key information related to traffic.

[0178] Step S20422, historical data comparison and analysis: compare current weather data with historical weather data to identify abnormal weather patterns and trends.

[0179] Step S20423, weather impact assessment: assess the impact of different weather conditions on data collection and positioning accuracy, and generate a weather impact report.

[0180] Step S2043: formulating parameter adjustment strategy.

[0181] Step S20431, data acquisition parameter adjustment: according to the weather impact assessment results, adjust the sensor acquisition parameters, such as the camera exposure time, radar scanning frequency, etc.

[0182] Step S20432, positioning algorithm parameter adjustment: adjusting the parameters of the positioning algorithm, such as the gain coefficient of the filter, to cope with noise and interference under different weather conditions.

[0183] Step S20433, data processing strategy adjustment: modify data preprocessing and fusion strategies to ensure data quality under severe weather conditions.

[0184] Step S2044: dynamic adaptation mechanism is implemented.

[0185] Step S20441, real-time monitoring: The IWAS system continuously monitors weather changes and adjusts data collection and processing strategies in real time.

[0186] Step S20442, dynamic parameter adjustment: automatically update and adjust sensor and algorithm parameters according to real-time weather analysis results.

[0187] Step S20443, adaptive feedback: Through the feedback mechanism, the parameter adjustment strategy is continuously optimized to improve the system's adaptability under extreme weather conditions.

[0188] Step S2045, specifically adjust the instance.

[0189] Step S20451, heavy rain weather adjustment: In heavy rain weather, the IWAS system increases the exposure time of the camera and the scanning frequency of the radar to enhance the ability to filter raindrops.

[0190] Step S20452, haze weather adjustment: In haze weather, the system increases the contrast and brightness of image processing to ensure that the camera can capture clear images.

[0191] Step S20453, high temperature weather adjustment: In hot weather, the IWAS system adjusts the operating frequency and sampling rate of the sensor to prevent data distortion caused by sensor overheating.

[0192] It should be noted that the existing solutions do not explicitly mention how to deal with data collection and analysis issues under extreme weather conditions. This may mean that under extreme weather conditions, such as heavy fog, heavy rain or snow, the performance of the system and the accuracy of the data may be affected. The intelligent weather adaptation system (IWAS) in this embodiment is specially designed to analyze the current weather conditions and adjust the data collection parameters and processing strategies accordingly. IWAS can adjust the sensitivity of the sensor in real time, select a suitable data processing algorithm, or temporarily rely on a more robust data source to optimize the system performance under adverse weather conditions. This embodiment provides a dynamic adjustment strategy to adapt to different environments and weather conditions through the introduction of IWAS, which not only enhances the robustness of the system, but also ensures the accuracy and reliability of the data. This is not covered by the prior art, making the application of this embodiment under extreme conditions more extensive and effective. Through the above improvements, this embodiment not only improves the adaptability and accuracy of the monitoring system, but also ensures efficient operation under various environments and weather conditions.

[0193] In some optional embodiments of the present application, step S205 is specifically implemented by the following method.

[0194] Step S2051, data input.

[0195] Step S20511, input of road and vehicle three-dimensional model data: obtaining updated road and vehicle three-dimensional model data from step S203.

[0196] Step S20512, high-precision vehicle positioning data input: obtain the fused high-precision vehicle positioning data from step S202.

[0197] Step S2052: trajectory data preprocessing.

[0198] Step S20521, data cleaning: clean the input data to remove noise and invalid data to ensure the accuracy and completeness of the data.

[0199] Step S20522, data format conversion: converting data in different formats into a unified format to facilitate subsequent processing and analysis.

[0200] Step S2053, trajectory tracking algorithm application.

[0201] Step S20531, trajectory matching: matching the high-precision vehicle positioning data with the road data in the three-dimensional model to generate a preliminary vehicle driving trajectory.

[0202] Step S20532, trajectory smoothing: Apply a trajectory smoothing algorithm (such as Kalman filtering) to optimize the preliminary trajectory, reduce errors and fluctuations, and generate a smooth driving trajectory.

[0203] Step S20533, anomaly detection: use machine learning algorithms (such as SVM, LSTM) to detect abnormal points in the trajectory, such as sharp turns, sudden brakes, etc., and mark abnormal behaviors.

[0204] Step S2054: trajectory data analysis.

[0205] Step S20541, driving pattern analysis: Analyze the vehicle's driving pattern through a clustering algorithm (such as K-means, DBSCAN) to identify common driving paths and behavior patterns.

[0206] Step S20542, traffic incident detection: using trajectory data to identify and detect traffic incidents, such as traffic congestion, traffic accidents, etc., to provide decision support for traffic management.

[0207] Step S20543, statistical analysis: perform statistical analysis on the trajectory data, calculate key indicators such as average speed, driving time, driving distance, and generate an analysis report.

[0208] Step S2055, data output and application.

[0209] Step S20551, trajectory data storage: the analyzed vehicle driving trajectory data is stored in the database of the unified data processing platform (UDPP) to ensure the persistence and accessibility of the data.

[0210] Step S20552, application example: applying the trajectory analysis results in a traffic management system, for example, optimizing the timing of traffic lights based on the driving trajectory and formulating traffic diversion strategies; in an intelligent transportation system, real-time monitoring of vehicle trajectories, providing traffic warnings and navigation suggestions.

[0211] It should be noted that in the prior art, vehicle information and road conditions are mainly used for real-time monitoring and trajectory tracking. This usually relies on static data and traditional data processing methods, which may not be sufficient to handle complex or rapidly changing traffic conditions, especially in large-scale or changeable urban environments. This embodiment tracks vehicle driving trajectories by utilizing dynamically updated three-dimensional models and precise positioning data, which provides more accurate and real-time trajectory information. In addition, this embodiment combines social media data, which is an innovation because social media can provide instant traffic updates and public feedback, thereby enhancing the accuracy and timeliness of traffic flow monitoring. This embodiment provides a deep and multi-dimensional traffic analysis framework by combining advanced data fusion technology and social media analysis. This approach allows the system to more effectively predict and manage traffic flow, especially when responding to emergencies, and can provide faster and more accurate responses than existing solutions.

[0212] As other optional embodiments of the present application, step S206 is specifically implemented by the following method.

[0213] Step S2061, data input.

[0214] Step S20611, vehicle driving trajectory data input: obtain the vehicle driving trajectory data from step S205, including the vehicle's real-time position, speed and direction information.

[0215] Step S20612, social media traffic feedback data input: Obtain the traffic feedback data posted by users on the social media platform from step S201, including text, pictures and videos.

[0216] Step S2062: data preprocessing.

[0217] Step S20621, trajectory data preprocessing: denoising, format conversion and standardization of vehicle driving trajectory data to ensure data consistency and accuracy.

[0218] Step S20622, social media data preprocessing: use natural language processing (NLP) technology to parse text content, extract key traffic-related information, and perform preliminary processing on image and video data.

[0219] Step S2063: data fusion and analysis.

[0220] Step S20631, data fusion: fuse the preprocessed trajectory data and social media data, and identify potential traffic events and patterns through association analysis.

[0221] Step S20632, real-time monitoring: Use machine learning algorithms (such as LSTM, time series analysis) to monitor the fused data in real time to identify traffic flow changes and anomalies.

[0222] Step S20633, traffic event detection: Detect and annotate traffic events such as traffic accidents and road congestion by analyzing social media data and trajectory data.

[0223] Step S2064, traffic flow analysis.

[0224] Step S20641, traffic flow calculation: based on the vehicle driving trajectory data, calculate the traffic flow of each road segment, including the number of vehicles, traffic density, average speed, etc.

[0225] Step S20642, pattern recognition: using clustering algorithms (such as K-means, DBSCAN) to identify traffic flow patterns and determine the characteristics of traffic flow during peak and trough periods.

[0226] Step S20643, predictive analysis: predict traffic flow through machine learning algorithms to identify possible future traffic congestion and abnormal situations.

[0227] Step S2065, data output and application.

[0228] Step S20651, traffic analysis report generation: Generate a detailed traffic flow analysis report based on the analysis results, including the real-time traffic conditions and prediction results of each road segment.

[0229] Step S20652, data output: output the traffic analysis results to the traffic management system and other related platforms for use by the traffic management department.

[0230] Step S20653, application example: applying the traffic analysis results in traffic management, for example, adjusting the timing of traffic lights based on real-time traffic data to optimize traffic flow; in intelligent transportation systems, providing real-time traffic warnings and navigation suggestions.

[0231] It should be noted that the optimization and feedback mechanism of the system is not described in detail in the prior art. This may mean that the adjustment and optimization of the system after deployment is limited, or the optimization process is not systematic enough, which may affect the long-term performance of the system and its ability to adapt to new challenges. In contrast, the present embodiment includes a clear optimization cycle and continuous improvement mechanism in design. System performance evaluation reports are generated regularly, which record the performance of each component and potential areas for improvement in detail. Based on these reports, system components such as UDPP, MDFS, DAMUM, IWAS and CIFAS will be optimized in a targeted manner to ensure that the system not only maintains its current performance, but also adapts to future needs and challenges. The optimization strategy of this embodiment provides a framework for continuous self-evaluation and improvement, which enables the system to continuously adapt to changing environments and technological advances. This clear optimization cycle and feedback mechanism is unclear in existing solutions and is a significant advantage of this embodiment, which helps to ensure long-term reliability and high performance.

[0232] In some optional embodiments, step S207 is specifically implemented by the following method.

[0233] Step S2071, data collection and integration,

[0234] Step S20711, analysis report collection: collect the traffic flow analysis report from step 206, including real-time traffic conditions and prediction results.

[0235] Step S20712, system operation data collection: collect system operation data from UDPP, MDFS, DAMUM, IWAS and CIFAS, including processing time, accuracy, resource usage, etc.

[0236] Step S2072, performance evaluation.

[0237] Step S20721, data comparison and analysis: compare the data in the traffic flow analysis report with the system operation data to identify system performance bottlenecks and deficiencies.

[0238] Step S20722, accuracy assessment: assess positioning accuracy, model update accuracy, and traffic analysis accuracy to identify sources of deviation and error.

[0239] Step S2073, resource utilization evaluation: analyzing the resource usage of each component of the system, and evaluating the utilization efficiency of the CPU, memory, storage, and network bandwidth.

[0240] Step S2073, optimization strategy formulation.

[0241] Step S20731, parameter optimization: according to the performance evaluation results, adjust the parameter settings of each component, such as data collection frequency, algorithm threshold and filter parameters.

[0242] Step S20732, algorithm optimization: improve data processing and analysis algorithms to enhance processing efficiency and accuracy.

[0243] Step S20733, resource configuration optimization: optimize system resource allocation to ensure efficient operation of each component and reduce resource waste.

[0244] Step S2074, system optimization implementation.

[0245] Step S20741, parameter adjustment: implement the optimized parameter settings and adjust the operating parameters of each component.

[0246] Step S20742, algorithm upgrade: deploy the optimized algorithm and update the data processing and analysis algorithms in the system.

[0247] Step S20743, resource reallocation: reallocate system resources according to the optimization strategy to ensure efficient operation of each component.

[0248] Step S2075, optimization effect evaluation.

[0249] Step S20751, re-evaluation: perform performance evaluation on the optimized system to verify the optimization effect.

[0250] Step S20752, continuous improvement: Based on the results of the re-evaluation, further adjust and optimize the system to form a closed-loop mechanism for continuous improvement.

[0251] Figure 6 is a structural diagram of a traffic flow determination device according to an embodiment of the present application, such as Figure 6 As shown, the device comprises:

[0252] The acquisition module 60 is used to acquire raw data related to the vehicle, wherein the raw data includes: the location data of the vehicle, video data collected from an image acquisition device set on the road where the vehicle is located, and road condition information of the road obtained from a social platform. The location data of the vehicle includes: first location data uploaded by a global positioning system of the vehicle terminal and second location data uploaded by a geomagnetic sensor;

[0253] An identification module 62, used to identify the vehicle position in the video data and obtain third position data;

[0254] A fusion module 64, configured to fuse the first position data, the second position data and the third position data according to different preset weights to obtain positioning data;

[0255] An updating module 66 is used to obtain a preset three-dimensional model, and update the preset three-dimensional model using the video data and positioning data of the road where the vehicle is located, to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located;

[0256] The determination module 68 is used to analyze the positioning data using the target three-dimensional model to generate the vehicle's driving trajectory, and determine the traffic flow in the target area where the road is located based on the road condition information and the vehicle's driving trajectory.

[0257] Optionally, the acquisition module 60 is also used to perform the following steps: based on a first predefined interface protocol, obtaining first location data from a first preset interface of the vehicle terminal; based on a second predefined interface protocol, obtaining second location data from a second preset interface of the vehicle terminal; obtaining preset keywords, and searching the social platform for target social platform content whose publishing location is in a target area and whose social platform content includes preset keywords, wherein the target area is the area where the road is located; if the target social platform content is text content, using natural language processing technology to identify road condition information in the text content; if the target social platform content contains picture content and video content associated with the text content, using computer vision technology to identify road condition information in the picture content and video content.

[0258] The traffic flow determination device also includes: a first processing module, which is used to search the social platform for the target social platform content whose publishing location is in the target area and whose social platform content includes preset keywords, and then execute the following steps: fusing the vehicle's location data, video data, and road condition information; using machine learning and data mining technology to identify at least one of the following from the fused data: traffic flow change rate information, congestion mode information, and accident location information to obtain an identification result; sending the identification result to the first information publishing platform, and if the traffic flow change rate in the identification result is greater than the first preset threshold and / or the number of accidents occurring at the same location is greater than the second preset threshold, sending the identification result to the second information publishing platform.

[0259] Optionally, the traffic flow determination device also includes: a second processing module, which is used to execute the following steps after obtaining raw data related to the vehicle: pre-processing the raw data by the following method: removing invalid data and erroneous data in the raw data, and normalizing the raw data; classifying the normalized raw data according to the data source and data type of the raw data to obtain a classification result; after pre-processing the raw data, performing data analysis on the raw data, and when it is analyzed that the vehicle's position data, video data and road condition information all include the same target location where the accident occurred, determining the accident information of the target location, and generating alarm information including the accident information, wherein the accident information includes at least one of the following: the location where the accident occurred, the time when the accident occurred; after performing data analysis on the raw data, determining the label information of the raw data according to the classification result, and storing the raw data in a unified data processing platform based on the label information of the raw data.

[0260] Optionally, the recognition module 62 is also used to execute the following steps: acquiring a first image including a vehicle in the video data, and performing image enhancement processing on the first image to obtain a target image when the brightness of the first image is less than a first preset brightness and / or the contrast of the first image is less than a first preset contrast; and identifying the position of the vehicle in the target image through an image recognition model to obtain third position data.

[0261] Optionally, the fusion module 64 is also used to perform the following steps: respectively convert the first position data, the second position data and the third position data into vector form to obtain a first vector corresponding to the first position data, a second vector corresponding to the second position data and a third vector corresponding to the third position data; when there is missing data in the first position data, the second vector is determined as the system state vector, the first vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are subjected to Kalman filtering according to different preset weights to obtain positioning data; when there is missing data in the second position data, the first vector is determined as the system state vector, the second vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are subjected to Kalman filtering according to different preset weights to obtain positioning data; when there is missing data in the third position data, the first vector is determined as the system state vector, the second vector and the third vector are determined as observation vectors, and the system state vector and the observation vector are subjected to Kalman filtering according to different preset weights to obtain positioning data.

[0262] Optionally, the update module 66 is also used to perform the following steps: integrating the positioning data into a preset three-dimensional model to obtain a first three-dimensional model; performing feature extraction on the content in the video frame of the video data to obtain a first target feature, wherein the first target feature includes at least one of the following: vehicle position, vehicle speed, vehicle driving direction, pedestrian position, road signs, and traffic signals; using a deep learning network to determine whether the first target feature has changed compared to the feature in the first three-dimensional model, and when it is determined that the first target feature has changed compared to the feature in the first three-dimensional model, updating the changed content in the first three-dimensional model to obtain a target three-dimensional model.

[0263] Optionally, the fusion module 64 is also used to execute the steps of: acquiring weather data, performing feature extraction on the weather data, and obtaining a second target feature, wherein the second target feature includes at least one of the following: lighting time, temperature, humidity, rainfall, and wind speed; determining a correlation between the weather data and a positioning accuracy index of the vehicle based on the second target feature; adjusting different preset weights of the first position data, the second position data, and the video data in the fusion process based on the correlation, and performing a fusion process on the first position data, the second position data, and the video data of the road where the vehicle is located based on the adjusted weights to obtain positioning data.

[0264] Optionally, the fusion module 64 is also used to execute the following steps: when the correlation relationship indicates that the rainfall is negatively correlated with the positioning accuracy index, the weight of the video data in the fusion process is reduced, and the weight of the first position data and the second position data in the fusion process is increased; when the correlation relationship indicates that the wind speed is negatively correlated with the positioning accuracy index, the weight of the second position data in the fusion process is reduced, and the weight of the first position data and the third position data in the fusion process is increased; when the correlation relationship indicates that the temperature is negatively correlated with the positioning accuracy index, the weight of the first position data in the fusion process is reduced, and the weight of the second position data and the third position data in the fusion process is increased.

[0265] Optionally, the traffic flow determination device also includes: an adjustment module, which is used to perform the following steps after determining the correlation between weather data and the vehicle positioning accuracy index based on the second target feature: based on the correlation, sending a first instruction to the Internet of Things system where the image acquisition device is located to adjust the exposure time of the image acquisition device.

[0266] Optionally, the traffic flow determination device also includes: a sending module, which is used to analyze the positioning data using the target three-dimensional model to generate the vehicle's driving trajectory, and then execute the following steps: input the vehicle's driving trajectory into the machine learning model to obtain the output result of the machine learning model, wherein the output result at least includes: the vehicle's speed, the road's congestion index, and the road's congestion location; send the road's congestion location to the information release platform, and send a second instruction to the control platform for controlling the traffic lights based on the vehicle's speed and the road's congestion index to control the traffic lights; generate a traffic management optimization suggestion report based on the output results, wherein the traffic management optimization suggestion report includes at least one of the following: bus route optimization suggestions, road construction optimization suggestions, traffic light control timing optimization suggestions, information on vehicles with speeds greater than a preset speed, and information on roads where the number of accidents is greater than a preset number.

[0267] Optionally, the determination module 68 is also used to perform the following steps: input the vehicle's driving trajectory data and the road condition information of the road into the time series analysis model, and obtain the traffic flow information and road congestion information of the target area output by the time series analysis model, wherein the vehicle's driving trajectory data includes at least one of the following: position, speed and driving direction, and the road condition information includes at least one of the following: traffic congestion information, accident information and road closure information, and the time series analysis model is obtained by training with historical driving trajectory data and historical road condition information; based on the traffic flow information and road congestion information of the target area, generate a traffic light control strategy and a traffic diversion strategy, wherein the traffic light control strategy at least includes: adjusting the operating time of the traffic light and changing the operating mode of the traffic light.

[0268] It should be noted that the above Figure 6 The modules in the embodiment may be program modules (e.g., a set of program instructions for implementing a specific function) or hardware modules. For the latter, they may be expressed in the following forms, but are not limited thereto: the expression form of the above modules is a processor, or the functions of the above modules are implemented by a processor.

[0269] It should be noted that Figure 6 The preferred implementation of the illustrated embodiment can be found in Figure 1 The relevant description of the illustrated embodiment will not be repeated here.

[0270] Figure 7 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a method for determining traffic flow. Figure 7As shown, the computer terminal 70 may include one or more (shown as 702a, 702b, ..., 702n in the figure) processors 702 (the processor 702 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 704 for storing data, and a transmission module 706 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 7 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 7 More or fewer components as shown, or with Figure 7 Different configurations shown.

[0271] It should be noted that the one or more processors 702 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 70. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0272] The memory 704 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the method for determining traffic flow in the embodiment of the present application. The processor 702 executes various functional applications and data processing by running the software programs and modules stored in the memory 704, that is, the method for determining traffic flow described above is realized. The memory 704 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 704 may further include a memory remotely arranged relative to the processor 702, and these remote memories may be connected to the computer terminal 70 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0273] The transmission module 706 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 70. In one example, the transmission module 706 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 706 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0274] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 70 .

[0275] It should be noted that, in some optional embodiments, the above Figure 7 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 7 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0276] It should be noted that Figure 7 The computer terminal shown is used to execute Figure 1 The method for determining the traffic flow shown in the figure, therefore the relevant explanations and instructions in the execution method of the above command are also applicable to the electronic device and will not be repeated here.

[0277] An embodiment of the present application further provides a non-volatile storage medium, which includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the above method for determining traffic flow.

[0278] A program for a non-volatile storage medium to perform the following functions: obtaining raw data related to a vehicle, wherein the raw data includes: the vehicle's position data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the vehicle's position data includes: first position data uploaded by a global positioning system of an on-board terminal and second position data uploaded by a geomagnetic sensor; identifying the vehicle's position in the video data to obtain third position data; fusing the first position data, the second position data, and the third position data according to different preset weights to obtain positioning data; obtaining a preset three-dimensional model, and using the video data and positioning data of the road where the vehicle is located to update the preset three-dimensional model to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; using the target three-dimensional model to analyze the positioning data, generate a driving trajectory of the vehicle, and determine the traffic flow in a target area where the road is located based on the road condition information and the driving trajectory of the vehicle.

[0279] An embodiment of the present application further provides an electronic device, including: a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the above method for determining traffic flow is executed when the program is running.

[0280] The processor is used to run a program that performs the following functions: obtaining original data related to the vehicle, wherein the original data includes: the vehicle's position data, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the vehicle's position data includes: first position data uploaded by a global positioning system of a vehicle-mounted terminal and second position data uploaded by a geomagnetic sensor; identifying the vehicle's position in the video data to obtain third position data; fusing the first position data, the second position data, and the third position data according to different preset weights to obtain positioning data; obtaining a preset three-dimensional model, using the video data and positioning data of the road where the vehicle is located, to update the preset three-dimensional model to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; using the target three-dimensional model to analyze the positioning data, generate a driving trajectory of the vehicle, and determine the traffic flow in a target area where the road is located based on the road condition information and the driving trajectory of the vehicle.

[0281] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0282] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0283] In the above-mentioned embodiments of the present application, the collected information is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary protection measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0284] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0285] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0286] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0287] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the relevant technology or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or CD-ROM and other media that can store program codes.

[0288] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for determining traffic flow, characterized in that: include: Acquire raw data related to the vehicle, wherein the raw data includes: location data of the vehicle, video data collected from an image acquisition device set up on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the location data of the vehicle includes: first location data uploaded by a global positioning system of an on-board terminal and second location data uploaded by a geomagnetic sensor; Identifying a vehicle position in the video data to obtain third position data; According to different preset weights, the first position data, the second position data and the third position data are fused to obtain positioning data; Acquire a preset three-dimensional model, and update the preset three-dimensional model using the video data of the road where the vehicle is located and the positioning data to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; The positioning data is analyzed using the target three-dimensional model to generate the driving trajectory of the vehicle, and the traffic flow in the target area where the road is located is determined based on the road condition information and the driving trajectory of the vehicle.

2. The method according to claim 1, characterized in that Get raw data related to the vehicle, including: Based on a first predefined interface protocol, the first location data is obtained from a first preset interface of the vehicle terminal; based on a second predefined interface protocol, the second location data is obtained from a second preset interface of the vehicle terminal; a preset keyword is obtained, and a target social platform content whose publishing location is in a target area and whose social platform content includes the preset keyword is searched in the social platform, wherein the target area is the area where the road is located; If the target social platform content is text content, natural language processing technology is used to identify the road condition information of the road in the text content; if the target social platform content contains picture content and video content associated with the text content, computer vision technology is used to identify the road condition information of the road in the picture content and the video content; After searching the social platform for target social platform content whose publishing location is in the target area and whose social platform content includes the preset keyword, the method further includes: Performing fusion processing on the vehicle's location data, the video data, and the road condition information; Using machine learning and data mining technology, identify at least one of the following from the fused data: traffic flow rate of change information, congestion mode information, accident location information, and obtain an identification result; The recognition result is sent to the first information publishing platform. If the rate of change of traffic flow in the recognition result is greater than a first preset threshold and / or the number of accidents occurring at the same location is greater than a second preset threshold, the recognition result is sent to the second information publishing platform.

3. The method according to any one of claims 1 to 2, characterized in that: After acquiring the raw data related to the vehicle, the method further includes: The raw data is preprocessed by the following method: invalid data and erroneous data in the raw data are removed, and the raw data is normalized; the normalized raw data is classified according to the data source and data type of the raw data to obtain a classification result; After preprocessing the raw data, data analysis processing is performed on the raw data, and when it is analyzed that the vehicle position data, the video data, and the road condition information all include the same target location where the accident occurred, accident information of the target location is determined, and alarm information including the accident information is generated, wherein the accident information includes at least one of the following: the location where the accident occurred, and the time when the accident occurred; After data analysis and processing is performed on the original data, label information of the original data is determined according to the classification result, and based on the label information of the original data, the original data is stored in a unified data processing platform.

4. The method according to claim 1, characterized in that: Identifying the vehicle position in the video data to obtain third position data includes: Acquire a first image including the vehicle in the video data, and when the brightness of the first image is less than a first preset brightness and / or the contrast of the first image is less than a first preset contrast, perform image enhancement processing on the first image to obtain a target image; The vehicle position in the target image is identified through an image recognition model to obtain third position data.

5. The method according to claim 1, characterized in that According to different preset weights, the first position data, the second position data, and the third position data are fused to obtain positioning data, including: Respectively converting the first position data, the second position data, and the third position data into vector forms to obtain a first vector corresponding to the first position data, a second vector corresponding to the second position data, and a third vector corresponding to the third position data; In the case where there is missing data in the first position data, the second vector is determined as a system state vector, the first vector and the third vector are determined as observation vectors, and Kalman filtering is performed on the system state vector and the observation vector according to different preset weights to obtain the positioning data; In the case where there is missing data in the second position data, the first vector is determined as a system state vector, the second vector and the third vector are determined as observation vectors, and Kalman filtering is performed on the system state vector and the observation vector according to different preset weights to obtain the positioning data; When there is missing data in the third position data, the first vector is determined as the system state vector, the second vector and the third vector are determined as observation vectors, and Kalman filtering is performed on the system state vector and the observation vector according to different preset weights to obtain the positioning data.

6. The method according to claim 1, characterized in that Using the video data of the road where the vehicle is located and the positioning data, updating the preset three-dimensional model to obtain a target three-dimensional model includes: Integrating the positioning data into the preset three-dimensional model to obtain a first three-dimensional model; Extracting features from the content in the video frame of the video data to obtain a first target feature, wherein the first target feature includes at least one of the following: vehicle position, vehicle speed, vehicle driving direction, pedestrian position, road sign, and traffic signal; A deep learning network is used to determine whether the first target feature has changed compared to the feature in the first three-dimensional model. When it is determined that the first target feature has changed compared to the feature in the first three-dimensional model, the changed content is updated in the first three-dimensional model to obtain the target three-dimensional model.

7. The method according to claim 1, characterized in that According to different preset weights, the first position data, the second position data, and the third position data are fused to obtain positioning data, including: Acquire weather data, perform feature extraction on the weather data, and obtain a second target feature, wherein the second target feature includes at least one of the following: sunshine time, temperature, humidity, rainfall, and wind speed; Determining, according to the second target feature, a correlation relationship between the weather data and a positioning accuracy index of the vehicle; According to the association relationship, the different preset weights of the first position data, the second position data and the video data in the fusion process are adjusted, and according to the adjusted weights, the first position data, the second position data and the video data of the road where the vehicle is located are fused to obtain the positioning data.

8. The method according to claim 7, characterized in that According to the association relationship, adjusting different preset weights of the first position data, the second position data, and the video data in a fusion process includes: When the association relationship indicates that the rainfall is negatively correlated with the positioning accuracy index, the weight of the video data in the fusion process is reduced, and the weight of the first position data and the second position data in the fusion process is increased; When the association relationship indicates that the wind speed is negatively correlated with the positioning accuracy index, reducing the weight of the second position data in the fusion process, and increasing the weights of the first position data and the third position data in the fusion process; When the association relationship indicates that the temperature is negatively correlated with the positioning accuracy index, the weight of the first position data in the fusion process is reduced, and the weights of the second position data and the third position data in the fusion process are increased.

9. The method according to claim 7, characterized in that: After determining the correlation between the weather data and the positioning accuracy index of the vehicle according to the second target feature, the method further includes: According to the association relationship, a first instruction is sent to the Internet of Things system where the image acquisition device is located to adjust the exposure time of the image acquisition device.

10. The method according to claim 1, characterized in that After analyzing the positioning data using the target three-dimensional model to generate the driving trajectory of the vehicle, the method further includes: Inputting the driving trajectory of the vehicle into a machine learning model to obtain an output result of the machine learning model, wherein the output result at least includes: the speed of the vehicle, the congestion index of the road, and the congestion location of the road; The congestion position of the road is sent to an information publishing platform, and according to the speed of the vehicle and the congestion index of the road, a second instruction is sent to a control platform for controlling a traffic light to control the traffic light; A traffic management optimization suggestion report is generated based on the output results, wherein the traffic management optimization suggestion report includes at least one of the following: bus route optimization suggestions, road construction optimization suggestions, traffic light control timing optimization suggestions, information on vehicles with speeds greater than a preset speed, and information on roads where the number of accidents is greater than a preset number.

11. The method according to claim 1, characterized in that: Determining the traffic flow in the target area where the road is located according to the road condition information of the road and the driving trajectory of the vehicle, including: Inputting the driving trajectory data of the vehicle and the road condition information of the road into a time series analysis model, obtaining the traffic flow information and road congestion information of the target area output by the time series analysis model, wherein the driving trajectory data of the vehicle includes at least one of the following: position, speed and driving direction, and the road condition information of the road includes at least one of the following: traffic congestion information, accident information and road closure information, and the time series analysis model is obtained by training using historical driving trajectory data and historical road condition information; A traffic light control strategy and a traffic flow diversion strategy are generated according to the traffic flow information and the road congestion information of the target area, wherein the traffic light control strategy at least includes: adjusting the operation time of the traffic light and changing the operation mode of the traffic light.

12. A device for determining traffic flow, characterized in that: include: An acquisition module, used to acquire raw data related to the vehicle, wherein the raw data includes: the location data of the vehicle, video data collected from an image acquisition device set on the road where the vehicle is located, and road condition information of the road obtained from a social platform, and the location data of the vehicle includes: first location data uploaded by a global positioning system of a vehicle-mounted terminal and second location data uploaded by a geomagnetic sensor; an identification module, used to identify the vehicle position in the video data to obtain third position data; a fusion module, configured to fuse the first position data, the second position data and the third position data according to different preset weights to obtain positioning data; An updating module is used to obtain a preset three-dimensional model, and update the preset three-dimensional model using the video data of the road where the vehicle is located and the positioning data to obtain a target three-dimensional model, wherein the preset three-dimensional model includes: the vehicle and the road where the vehicle is located; A determination module is used to analyze the positioning data using the target three-dimensional model to generate the driving trajectory of the vehicle, and determine the traffic flow in the target area where the road is located based on the road condition information and the driving trajectory of the vehicle.

13. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the method for determining traffic flow according to any one of claims 1 to 11.

14. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program, when running, executes the method for determining traffic flow as described in any one of claims 1 to 11.

15. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for determining traffic flow according to any one of claims 1 to 11 is implemented.

Citation Information

Cited By

  • Traffic flow monitoring method and system for traffic control point

    CN120148261A

  • Method and device for determining traffic flow and electronic equipment

    CN120496332A