A method, system, electronic device and medium for dynamic and static traffic management

Abstract

The present invention relates to the field of intelligent transportation technology, and in particular to a dynamic and static traffic management method, system, electronic equipment and medium. Multi-source data is collected through video monitoring, sensor networks, and GPS positioning, and data preprocessing and fusion technology is adopted. Dynamic and static traffic feature extraction is combined, and Transformer, YOLO and graph convolutional neural network models are used to perform deep learning and cross-domain learning on traffic flow and road features. Through real-time traffic prediction and intelligent decision-making, the optimal traffic signal control strategy can be generated according to actual traffic conditions, and better traffic management solutions can be provided. The method is suitable for a variety of traffic scenarios, has the characteristics of high adaptability, flexibility and intelligence, and effectively improves the efficiency and accuracy of traffic management.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a dynamic and static traffic management method, system, electronic equipment and medium. Background Art

[0002] The existing urban traffic management system has many shortcomings, such as single data source, limited dynamic response capability, insufficient use of multimodal data and lack of cross-domain learning ability, which greatly reduces the system's real-time management and optimization capabilities in complex traffic environments.

[0003] First, existing methods or systems mostly rely on a single data source, such as video surveillance and traffic flow sensors. Although these data can provide local information, they have obvious limitations. Video surveillance is limited by the camera's field of view and it is difficult to fully cover urban traffic; traffic sensors are mostly limited to fixed sections and cannot dynamically capture global changes. In addition, the lack of fusion of multimodal data has led to the fragmentation of video, sensor, and vehicle GPS data, making it impossible to form collaborative decisions, affecting the integrity and accuracy of information.

[0004] Secondly, the dynamic response capabilities of existing methods or systems are insufficient, especially in traffic flow prediction and emergency handling. Most systems rely on simple historical data regression and are unable to timely predict and respond to real-time traffic changes. During peak hours or when emergencies (such as traffic accidents) occur, it is difficult for existing systems to provide effective control measures in a timely manner. Abnormal event detection is also mainly based on a single data source, especially video detection, which has reduced recognition accuracy under complex conditions such as low light and occlusion, resulting in delayed emergency response.

[0005] In addition, existing methods or systems lack cross-domain learning capabilities, and the traffic characteristics of different cities vary significantly, making it difficult to migrate the system between different regions. The model needs to be retrained every time it is expanded to a new area, which consumes a lot of time and manpower, limiting the promotion of the system. Moreover, the insufficient utilization of multimodal data fusion is also a shortcoming of existing methods or systems. Traffic management involves multimodal data such as video, sensors, and geographic information, but due to different structures and sources, there is a lack of effective fusion mechanism, which makes it difficult for the system to fully explore the correlation between information, thus affecting the accuracy of global decision-making.

[0006] Therefore, the existing urban traffic management methods or systems are insufficient in data processing, dynamic response and cross-domain application, and a new type of dynamic and static traffic management method and system is urgently needed to overcome these challenges. Summary of the invention

[0007] To achieve the above-mentioned purpose, the present invention proposes a dynamic and static traffic management method, system, electronic device and medium, which realizes the comprehensive collection and analysis of traffic information by integrating multi-source data such as video surveillance, sensor network, GPS positioning, etc., and adopts data preprocessing and fusion technology, combines dynamic and static traffic feature extraction, and uses Transformer, YOLO and graph convolutional neural network models to perform deep learning and cross-domain learning on traffic flow and road characteristics. Through real-time traffic prediction and intelligent decision-making, it can generate the optimal traffic signal control strategy according to the actual traffic conditions, provide better traffic management solutions, and is suitable for a variety of traffic scenarios. It has the characteristics of high adaptability, flexibility and intelligence, and effectively improves the efficiency and accuracy of traffic management.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] In a first aspect, the present invention provides a method for dynamic and static traffic management, comprising the following steps:

[0010] S1. Multimodal traffic data collection: Through video surveillance systems, sensor networks, and GPS positioning systems, data sources from different domains are collected to obtain real-time traffic information of vehicles and pedestrians, that is, multimodal traffic data;

[0011] S2. Data preprocessing and fusion: preprocess the collected multimodal traffic data, eliminate noise and redundant information, use interpolation methods to fill missing values, identify outliers through box plots, and use statistics to correct outliers;

[0012] S3, Dynamic and static traffic feature extraction: Based on Transformer, spatial feature extraction is performed on video data and sensor data, and the YOLO algorithm is applied to vehicle detection and density analysis in image data to extract dynamic traffic change features; at the same time, GPS data is combined with other geographic location information to extract urban static features through graph convolutional neural network;

[0013] S4. Dynamic traffic management decision-making: Use the MLP model to generate real-time traffic prediction results, generate corresponding traffic signal control strategies and vehicle scheduling plans, and dynamically adjust traffic light time and lane allocation traffic management measures.

[0014] As a further technical solution of the present invention, the specific process of step S2 is:

[0015] S21. Combine the vehicle speed and the crowd flow speed to obtain processed speed data ,in , The sensor network is The vehicle speed and crowd flow speed collected at all times, , , is the number of intersections;

[0016] S22. Process missing values ​​and outliers. First, use the box plot technique to effectively identify outliers. The box plot uses the interquartile range. To determine outliers, the interquartile range is the third quartile and the first quartile The distance between them, that is: , then the set of outliers for: , using non-outliers to calculate the statistical characteristics of spatiotemporal data, namely the mean and variance Correction for outliers and missing values.

[0017] As a further technical solution of the present invention, the specific process of step S3 is:

[0018] S31, using YOLO algorithm to monitor the frequency of the system The image obtained at the moment Perform target recognition and obtain Traffic volume , traffic data , then the data dimensions identified every day are: , ,in , is the height and width of the image; and the traffic volume , Traffic data The speed data processed in step S21 Splice to get data : ;

[0019] S32, use transformer to extract spatiotemporal features, first input the data of transformer Through a linear transformation Map to high-dimensional feature space to obtain high-dimensional features ,in, is the feature dimension represented in Transformer, and position encoding is used to add position information to the input data. For each time step t, the position encoding is calculated by the following formula:

[0020] ,

[0021] ;

[0022] Then use a multi-head attention mechanism to capture cross-temporal dependencies: ,in, , , ,and It is a learnable weight matrix, and a feedforward neural network is connected after the self-attention mechanism to further perform nonlinear transformation on the spatiotemporal features: ,in is a learnable parameter. After each self-attention mechanism and feedforward neural network, residual connections and layer normalization are used. The formula is as follows: After being processed by multiple Transformer layers, the extracted dynamic spatiotemporal features are obtained. , which contains the spatiotemporal characteristics of vehicle flow, pedestrian flow and spatial location at all intersections;

[0023] S33, extracting urban static features through graph convolutional networks, the spatial relationship between urban intersections is represented by a graph data structure Given, where It is a collection of subway stations. is the set of edges between subway stations, and the spatial correlation between points is represented by the adjacency matrix To measure, when and hour, Not 0; use the diffusion map convolution layer to extract spatial information. The expression of the diffusion map convolution layer is: ,in, Represents the transfer matrix The power series of, in the case of an undirected graph, ; In the case of a directed graph, the diffusion process has two directions, forward and reverse, and the forward transfer matrix is , the reverse transfer matrix is ; is the learnable parameter matrix of graph convolution.

[0024] As a further technical solution of the present invention, the specific process of step S4 is: using an MLP model to predict the traffic volume, pedestrian volume, traffic speed and crowd flow speed of each intersection in the city within the next hour : ,in is a learnable parameter to calculate the intersection The number of seconds of green lights on motorway in the future Green light seconds for non-motor vehicle lanes :

[0025] ,

[0026] ,

[0027] in Separate intersection The saturated flow of motor vehicles and non-motor vehicles.

[0028] In a second aspect, the present invention provides a dynamic and static traffic management system, comprising the following four modules:

[0029] The multimodal traffic data collection module is used to collect data sources from different domains through video surveillance systems, sensor networks, and GPS positioning devices to obtain traffic information of traffic participants in real time;

[0030] Data preprocessing and fusion module, used to preprocess the collected multimodal traffic data, remove noise and eliminate redundant information;

[0031] The dynamic and static traffic feature extraction module is used to extract traffic features from different data types. It uses Transformer to extract spatial features from video data and sensor data, and applies the YOLO algorithm to perform vehicle detection and traffic flow density analysis to identify dynamic traffic changes. At the same time, GPS data is combined with other geographic location information, and a graph convolutional neural network (GCN) is used to extract static urban features.

[0032] The dynamic traffic management decision-making module is used to generate intelligent traffic signal control strategies and vehicle scheduling solutions based on the real-time traffic prediction results of the MLP model.

[0033] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0034] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention obtains more comprehensive traffic information by integrating multimodal data such as video surveillance, sensor networks, GPS, etc., and combines dynamic traffic flow with static road structure characteristics, uses deep learning models for comprehensive analysis, and provides accurate traffic management solutions. The method and system have cross-domain learning capabilities, can adapt to different cities and traffic conditions, and have a wide range of applications. It can automatically generate traffic signal control based on real-time traffic forecasts, optimize traffic light times, and significantly improve traffic management efficiency. At the same time, through efficient data preprocessing and outlier correction, data quality is ensured, providing a reliable basis for intelligent decision-making, and overall improving the accuracy and adaptability of traffic management. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present disclosure and do not constitute improper limitations on the present invention.

[0038] Figure 1 A schematic flow chart of a dynamic and static traffic management method provided by the present invention.

[0039] Figure 2 A schematic diagram of the framework of a dynamic and static traffic management system provided by the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0041] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0042] It should be noted that the terms used herein are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "include" and "have" 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.

[0043] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0044] Embodiment 1:

[0045] like Figure 1As shown, this embodiment provides a dynamic and static traffic management method, including the following steps:

[0046] S1. Multimodal traffic data collection:

[0047] Traffic data is collected by video surveillance systems and sensor networks installed at various intersections in the city. Assuming that there are intersection, record the image obtained by the video surveillance system at time , the sensor network at time Collect vehicle speed , crowd flow speed , collecting data at 1-minute intervals, the dimensions of the data collected every day are: , .

[0048] S2. Data preprocessing and fusion:

[0049] First, the vehicle speed and crowd flow speed are spliced ​​to obtain the processed speed data: ;

[0050] Then, missing values ​​and outliers are processed. First, the box plot technology is used to effectively identify outliers. The box plot uses the interquartile range. To determine outliers, the interquartile range is defined as the third quartile First quartile The distance between them, that is: , then the set of outliers is defined as: ;

[0051] Then use the non-outliers to calculate the statistical characteristics of the spatiotemporal data, that is, the mean and variance Correction for outliers and missing values.

[0052] First use the YOLO algorithm to analyze the image Perform target recognition and obtain Traffic volume , traffic data , then the data dimensions identified every day are: , ;

[0053] Then combine the flow data with the speed data: ;

[0054] Then use transformer to extract spatiotemporal features. First, input data Mapping to a high-dimensional feature space, this embodiment maps the input features to a higher-dimensional space through a linear transformation (fully connected layer): ,in, is the feature dimension represented in the Transformer model. Since there is no natural time order information in the Transformer model, position encoding is used to add position information to the input data. For each time step t, the position encoding is calculated by the following formula:

[0055] ,

[0056] ;

[0057] Then a multi-head attention mechanism is used to capture cross-temporal dependencies: ,in, , , ,and is a learnable weight matrix. In the Transformer layer, the self-attention mechanism is followed by a feedforward neural network to further perform nonlinear transformations on spatiotemporal features: ,in are learnable parameters, obtained after each module (self-attention mechanism and feedforward neural network) using residual connections and layer normalization: After being processed by multiple Transformer layers, the extracted spatiotemporal features are obtained , which contains the spatiotemporal characteristics of vehicle flow, pedestrian flow and spatial location at all intersections;

[0058] Finally, the static features of the city are extracted through the graph convolutional network. The spatial relationship between the city intersections is given by a graph data structure. Specifically, the graph is ,in It is a collection of subway stations. is the set of edges between subway stations, and the spatial correlation between points is represented by the adjacency matrix To measure, when and hour, is not 0, and the spatial information is extracted. This embodiment uses a diffusion map convolution layer, and the expression of the diffusion map convolution layer is: ,in, Represents the transfer matrix The power series of, in the case of an undirected graph, ; In the case of a directed graph, the diffusion process has two directions, forward and reverse, and the forward transfer matrix is , the reverse transfer matrix is ;in is the learnable parameter matrix of graph convolution.

[0059] S4. Dynamic traffic management decision-making:

[0060] Through an MLP layer, the traffic volume, pedestrian volume, traffic speed and crowd flow speed of each intersection in the city within the next hour are predicted : ,in is a learnable parameter, and finally calculates the intersection The number of seconds of green lights on motorway in the future Green light seconds for non-motor vehicle lanes :

[0061] ,

[0062] ,

[0063] in Separate intersection The saturated flow of motor vehicles and non-motor vehicles.

[0064] Embodiment 2:

[0065] like Figure 2 As shown, this embodiment provides a dynamic and static traffic management system, including:

[0066] The multimodal traffic data acquisition module is used to collect data sources from different domains through video surveillance systems, sensor networks, and GPS positioning devices to obtain traffic information of traffic participants in real time. Through the collaborative work of multi-source devices, it can comprehensively collect multimodal data reflecting traffic conditions, such as video data, sensor data, GPS trajectory data, etc., to provide a basis for subsequent data fusion and analysis, and ensure the comprehensiveness and real-time nature of the data.

[0067] Data preprocessing and fusion module, used to preprocess the collected multimodal traffic data, remove noise and eliminate redundant information;

[0068] The dynamic and static traffic feature extraction module is used to extract traffic features from different data types. It uses Transformer to extract spatial features from video data and sensor data, and applies the YOLO algorithm to perform vehicle detection and traffic flow density analysis to identify dynamic traffic changes. At the same time, GPS data is combined with other geographic location information, and a graph convolutional neural network (GCN) is used to extract urban static features, such as road structure and lane information, so as to fully reflect the dynamic and static characteristics of the traffic system.

[0069] The dynamic traffic management decision-making module is used to generate intelligent traffic signal control strategies and vehicle scheduling plans based on the real-time traffic prediction results of the MLP model. This module can automatically adjust decisions such as traffic light duration and lane allocation based on the continuously acquired traffic environment feedback, thereby improving traffic flow efficiency and reducing congestion time, thereby realizing intelligent and dynamic traffic management.

[0070] It should be noted that the above modules correspond to steps S1 to S4 in Example 1, and the examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above Example 1. It should be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer executable instructions.

[0071] In further embodiments, there is also provided:

[0072] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, it will not be described in detail here.

[0073] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0074] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0075] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in Example 1 is completed.

[0076] The method in Example 1 can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it is not described in detail here.

[0077] Those skilled in the art will appreciate that the units, i.e., algorithm steps, of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0078] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0079] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for dynamic and static traffic management, characterized in that: The following steps are involved: S1. Multimodal traffic data collection: Through video surveillance systems, sensor networks, and GPS positioning systems, data sources from different domains are collected to obtain real-time traffic information of vehicles and pedestrians, that is, multimodal traffic data; S2. Data preprocessing and fusion: preprocess the collected multimodal traffic data, eliminate noise and redundant information, use interpolation methods to fill missing values, identify outliers through box plots, and use statistics to correct outliers; S3, Dynamic and static traffic feature extraction: Based on Transformer, spatial feature extraction is performed on video data and sensor data, and the YOLO algorithm is applied to vehicle detection and density analysis in image data to extract dynamic traffic change features; at the same time, GPS data is combined with other geographic location information to extract urban static features through graph convolutional neural network; The specific process is: S31, using YOLO algorithm to monitor the frequency of the system The image obtained at the moment Target recognition is performed and the Traffic flow at the time , traffic data , then the data dimensions identified every day are: ,in , For images height and width; and , Traffic data The speed data processed in step S21 Splice to get data : ； S32, use transformer to extract spatiotemporal features, first input the data of transformer Through a linear transformation Map to high-dimensional feature space to obtain high-dimensional features ,in, is the feature dimension represented in Transformer, and position encoding is used to add position information to the input data. For each time step t, the position encoding is calculated by the following formula: ， ； Then use a multi-head attention mechanism to capture cross-temporal dependencies: ,in, ,and It is a learnable weight matrix, and a feedforward neural network is connected after the self-attention mechanism to further perform nonlinear transformation on the spatiotemporal features: ,in is a learnable parameter. After each self-attention mechanism and feedforward neural network, residual connections and layer normalization are used. The formula is as follows: ， After being processed by multiple Transformer layers, the extracted dynamic spatiotemporal features are obtained , which contains the spatiotemporal characteristics of vehicle flow, pedestrian flow and spatial location at all intersections; S33, extracting urban static features through graph convolutional networks, the spatial relationship between urban intersections is represented by a graph data structure Given, where It is a collection of subway stations. is the set of edges between subway stations, and the spatial correlation between points is represented by the adjacency matrix To measure, when and hour, Not 0; use the diffusion map convolution layer to extract spatial information. The expression of the diffusion map convolution layer is: ， in, Represents the transfer matrix The power series of, in the case of an undirected graph, ; In the case of a directed graph, the diffusion process has two directions, forward and reverse, and the forward transfer matrix is , the reverse transfer matrix is ; is the learnable parameter matrix of graph convolution; S4. Dynamic traffic management decision-making: Use the MLP model to generate real-time traffic prediction results, generate corresponding traffic signal control strategies and vehicle scheduling plans, and dynamically adjust traffic light time and lane allocation traffic management measures.

2. The method for dynamic and static traffic management according to claim 1, characterized in that: The specific process of step S2 is: S21. Combine the vehicle speed and the crowd flow speed to obtain processed speed data ,in , The sensor network is The vehicle speed and crowd flow speed collected at all times, , , is the number of intersections; S22. Process missing values ​​and outliers. First, use the box plot technique to effectively identify outliers. The box plot uses the interquartile range. To determine outliers, the interquartile range is the third quartile and the first quartile The distance between them, that is: , then the set of outliers for: , using non-outliers to calculate the statistical characteristics of spatiotemporal data, namely the mean and variance Correction for outliers and missing values.

3. The method for dynamic and static traffic management according to claim 2 is characterized in that: The specific process of step S4 is: predict the traffic volume, pedestrian volume, traffic speed and crowd flow speed Y at each intersection in the city within the next hour through an MLP model : ,in is a learnable parameter to calculate the intersection The number of seconds of green lights on motorway in the future Green light seconds for non-motor vehicle lanes : ， ， in The intersections The saturated flow of motor vehicles and non-motor vehicles.

4. A dynamic and static traffic management system, characterized in that: Implementing the method according to any one of claims 1 to 3, comprising the following four modules: The multimodal traffic data collection module is used to collect data sources from different domains through video surveillance systems, sensor networks, and GPS positioning devices to obtain traffic information of traffic participants in real time; Data preprocessing and fusion module, used to preprocess the collected multimodal traffic data, remove noise and eliminate redundant information; The dynamic and static traffic feature extraction module is used to extract traffic features from different data types. It uses Transformer to extract spatial features from video data and sensor data, and applies the YOLO algorithm to perform vehicle detection and traffic flow density analysis to identify dynamic traffic changes. At the same time, GPS data is combined with other geographic location information, and a graph convolutional neural network is used to extract static features of the city. The dynamic traffic management decision-making module is used to generate intelligent traffic signal control strategies and vehicle scheduling solutions based on the real-time traffic prediction results of the MLP model.

5. An electronic device, characterized in that: The invention comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 3 is completed.

6. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method as claimed in any one of claims 1 to 3.

Images