Traffic behavior prediction and risk assessment method based on deep neural network

Abstract

The invention discloses a traffic behavior prediction and risk assessment method based on a deep neural network, and the method comprises the steps: obtaining historical traffic behavior data, carrying out the preprocessing of the data, inputting the preprocessed data into a pre-trained traffic behavior prediction model, and generating a local space-time diagram; performing feature extraction on the local space-time diagram by using a convolutional neural network, and performing time sequence analysis on a feature extraction result to obtain an analysis prediction result of the traffic behavior; and setting a risk assessment mechanism to carry out quantitative assessment on the analysis and prediction result of the traffic behavior to obtain a risk assessment result of the corresponding behavior. According to the method provided by the invention, traffic behaviors can be accurately predicted, and potential traffic modes and trends can be identified, so that the traffic safety and smoothness are remarkably enhanced, traffic accidents are reduced, and the overall operation efficiency of an urban traffic system and the capability of dealing with emergencies are improved.

Description

Technical Field

[0001] The present invention relates to the field of machine learning prediction technology, and in particular to a traffic behavior prediction and risk assessment method based on a deep neural network. Background Art

[0002] With the acceleration of urbanization and the rapid growth of the number of motor vehicles, traffic congestion, frequent traffic accidents and other problems are becoming increasingly serious, which not only affects people's daily quality of life, but also brings huge economic losses to society. In order to improve this situation, intelligent transportation systems (ITS) came into being. By integrating multidisciplinary achievements such as advanced information technology, communication technology, sensor technology and control theory, it aims to improve traffic safety and efficiency. Among the many functions of intelligent transportation systems, traffic behavior prediction plays a vital role. Traditional traffic behavior analysis methods mainly rely on statistical models and expert rule bases. Although they can reflect the basic characteristics of traffic flow to a certain extent, they are unable to cope with complex dynamic changes. For example, when encountering emergencies or unstructured road environments, traditional methods are difficult to respond quickly and accurately.

[0003] The development of neural networks in recent years has provided new ideas for solving the above problems. Deep learning, as a machine learning algorithm, is good at automatically extracting deep features from data and can effectively capture the complex relationships between time series data. Compared with traditional methods, methods based on deep neural networks can more accurately understand the intentions of traffic participants and predict their future behavior trajectories, thereby identifying potential risk points in advance and taking preventive measures. Specifically, by using deep learning architectures suitable for processing time series data, such as long short-term memory networks and gated recurrent units, patterns in different traffic scenarios can be learned from massive historical traffic data, including but not limited to changes in vehicle speed, acceleration and deceleration behavior, and lane change frequency. In addition, combined with computer vision technology, the location information of vehicles, pedestrians, and other obstacles can be directly obtained from images or videos, further enhancing the ability to perceive the surrounding environment.

[0004] Therefore, there is an urgent need to study a traffic behavior prediction and risk assessment method based on deep neural networks to improve the intelligence level of the traffic management system, which can also play an important role in ensuring the safety of public travel. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a traffic behavior prediction and risk assessment method based on deep neural networks to solve the problem that traditional statistical models and expert rule bases are difficult to quickly and accurately predict traffic behaviors and identify potential risks when dealing with complex dynamic changes and unstructured scenarios.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a traffic behavior prediction and risk assessment method based on a deep neural network, including: obtaining historical traffic behavior data and preprocessing it, inputting the preprocessed data into a pre-trained traffic behavior prediction model to generate a local space-time graph; using a convolutional neural network to extract features from the local space-time graph, and performing time series analysis on the results of the feature extraction to obtain analysis and prediction results of the traffic behavior; setting a risk assessment mechanism to quantitatively evaluate the analysis and prediction results of the traffic behavior to obtain risk assessment results of the corresponding behavior.

[0008] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, the preprocessing process includes: Identify missing values ​​in the acquired historical traffic behavior data and use interpolation to complete them according to the characteristics of the time series, and standardize the completed data; Dividing the standardized data into multiple spatiotemporal units according to preset time and space granularity, and aggregating the traffic behavior data in each spatiotemporal unit; Construct a spatiotemporal grid, where each grid cell corresponds to a geographic location and time interval, and fill it with the corresponding aggregate data; The topological structure information of the traffic road network is mapped onto the spatiotemporal grid to reflect the road connectivity.

[0009] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, the generating of the local spatiotemporal graph includes: Arrange the traffic topologies at different times in the space-time grid in order, establish a sliding window on the traffic topology sequence, and slide backward with a preset step size; Connecting the road nodes in the traffic behavior data in the sliding window with themselves at the previous moment and the next moment, adding self-loops, and constructing an adjacency matrix of the local space-time graph; According to the constructed local space-time topology, a local space-time graph data matrix is ​​formed.

[0010] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, wherein: the feature extraction of the local spatiotemporal graph using convolutional neural network includes: Performing Chebyshev graph convolution on the local space-time graph data matrix and the adjacency matrix of the local space-time graph; The output after graph convolution is activated by using the ReLU function to obtain a feature matrix, and the feature matrix is ​​segmented, and the redundant information of the segmented data is eliminated by using the maximum pooling to obtain a local spatiotemporal feature matrix; If the split feature matrix meets the screening characteristics, the feature matrix information that meets the screening characteristics is retained and a new valid information set is imported; If the segmented feature matrix does not meet the screening characteristics, it is judged as redundant information and is removed, and the valid information set is not imported; After selectively screening the characteristic information, the effective information set consisting of the effective information is sorted to obtain a local spatiotemporal characteristic matrix.

[0011] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, the time series analysis of the result of the feature extraction includes: Inputting the local spatiotemporal feature matrix into a time series analysis unit of a traffic behavior prediction model, wherein the time series analysis unit includes three gating mechanisms, namely an input gate, a forget gate, and an output gate; The input gate is used to receive the local spatiotemporal feature matrix input at the current moment and the hidden state of the previous time step, and calculate the first threshold value as the activation value of the input gate through the activation function; The forget gate is used to calculate a second threshold according to the current input and the hidden state of the previous time step to determine the information that needs to be retained in the cell state of the previous time step; The output gate calculates a third threshold value through an activation function, and multiplies the third threshold value by the cell state adjusted by the hyperbolic tangent function to obtain a hidden state of the time step, and obtains an analysis and prediction result of the traffic behavior based on the hidden state.

[0012] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, the training and updating of the traffic behavior prediction model includes: Divide the preprocessed traffic behavior data into a training set and a validation set; Training the traffic behavior prediction model through the training set and its corresponding labels, setting the batch size and rounds, and updating the weights in each iteration to minimize the loss; Dynamically adjust the parameters of the traffic behavior prediction model, and use the validation set to evaluate the model performance.

[0013] As a preferred solution of the traffic behavior prediction and risk assessment method based on deep neural network described in the present invention, wherein: the setting of risk assessment mechanism to quantitatively evaluate the analysis and prediction results of the traffic behavior includes: Set a risk assessment benchmark range for traffic behavior and compare the analysis and prediction results with the benchmark range; The risk assessment benchmark range is expressed as: ,in, is the mean, is the coefficient, is the standard deviation, To analyze the prediction results; If the analysis and prediction results are within the risk assessment benchmark, the conventional monitoring process is executed to adjust the duration of traffic lights and optimize road use according to traffic flow; If the analysis prediction result is greater than , it indicates that the traffic flow has increased abnormally. When an abnormal increase in traffic is detected, real-time road condition information is obtained for secondary judgment. When the relative gap between traffic flow and road capacity exceeds a specified threshold, it indicates that the road system is in a clear congestion state. The intelligent transportation system will re-plan the traffic flow and guide vehicles to alternative routes. If there is no suitable alternative route, temporary traffic control measures will be implemented; If the analysis prediction result is less than When the traffic flow is lower than the minimum safe traffic flow demand of the road, the intelligent transportation system will be used to dispatch traffic flow of other sections. If the minimum safe traffic flow demand still cannot be met, the emergency response plan will be activated. If the analysis and prediction result is greater than the minimum safe traffic demand of the road but less than , uses intelligent transportation technology to optimize traffic flow distribution and scheduling based on real-time data analysis, and adjusts the access rights and priorities of different road sections in real time by dynamically monitoring the relationship between traffic flow and road capacity.

[0014] In a second aspect, the present invention provides a traffic behavior prediction and risk assessment system based on a deep neural network, comprising: A data processing module, used to obtain and pre-process historical traffic behavior data, input the pre-processed data into a pre-trained traffic behavior prediction model, and generate a local spatiotemporal graph; An analysis and prediction module, used to extract features from the local spatiotemporal graph using a convolutional neural network, and perform time series analysis on the results of the feature extraction to obtain analysis and prediction results of traffic behavior; The risk assessment module is used to set a risk assessment mechanism to quantitatively evaluate the analysis and prediction results of the traffic behavior to obtain the risk assessment results of the corresponding behavior.

[0015] In a third aspect, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the traffic behavior prediction and risk assessment method based on a deep neural network are implemented.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the traffic behavior prediction and risk assessment method based on a deep neural network.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a traffic behavior prediction and risk assessment method based on a deep neural network, which acquires and preprocesses historical traffic behavior data, generates a local spatiotemporal graph using a pre-trained traffic behavior prediction model, and uses a convolutional neural network for feature extraction and time series analysis. It can accurately predict traffic behavior and identify potential traffic patterns and trends, which not only helps to identify high-risk traffic conditions, but also provides traffic management departments with decision support for optimizing traffic flow, preventing accidents and improving emergency response efficiency, thereby significantly enhancing traffic safety and smoothness, reducing the occurrence of traffic accidents, and improving the overall operating efficiency of the urban transportation system and its ability to respond to emergencies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 A schematic diagram of the overall process logic of a traffic behavior prediction and risk assessment method based on a deep neural network according to an embodiment of the present invention; Figure 2 A block diagram of a time series analysis unit of a traffic behavior prediction and risk assessment method based on a deep neural network according to an embodiment of the present invention; Figure 3 This is a risk assessment flow chart of a traffic behavior prediction and risk assessment method based on a deep neural network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0021] Embodiment 1: Reference Figure 1-Figure 3 As an embodiment of the present invention, a traffic behavior prediction and risk assessment method based on a deep neural network is provided, such as Figure 1 The specific steps shown include: S100: Obtaining historical traffic behavior data and preprocessing it, inputting the preprocessed data into a pre-trained traffic behavior prediction model, and generating a local spatiotemporal graph; S200: extract features from the local spatiotemporal graph using a convolutional neural network, and perform time series analysis on the feature extraction results to obtain analysis and prediction results of traffic behavior; S300: Setting a risk assessment mechanism to quantitatively assess the analysis and prediction results of traffic behaviors, and obtaining risk assessment results of corresponding behaviors.

[0022] It should be noted that the present invention provides a traffic behavior prediction and risk assessment method based on a deep neural network. By acquiring and preprocessing historical traffic behavior data, using a pre-trained traffic behavior prediction model to generate a local spatiotemporal graph, and using a convolutional neural network for feature extraction and time series analysis, it can accurately predict traffic behavior and identify potential traffic patterns and trends. It not only helps to identify high-risk traffic conditions, but also provides traffic management departments with decision support for optimizing traffic flow, preventing accidents and improving emergency response efficiency, thereby significantly enhancing traffic safety and smoothness, reducing the occurrence of traffic accidents, and improving the overall operating efficiency of the urban transportation system and its ability to respond to emergencies.

[0023] In the embodiment of the present application, step S100 obtains historical traffic behavior data and preprocesses it, inputs the preprocessed data into a pre-trained traffic behavior prediction model, and generates a local spatiotemporal graph including: Specifically, historical traffic behavior data is obtained, including but not limited to: vehicle traffic volume, pedestrian flow, traffic speed data, traffic accident data, weather data, traffic lights and road infrastructure data.

[0024] Specifically, preprocessing the acquired historical traffic behavior data includes: Identify missing values ​​in the acquired historical traffic behavior data and use interpolation to complete them according to the characteristics of the time series, and standardize the completed data; The standardized data is divided into multiple spatiotemporal units according to the preset time and space granularity, and the traffic behavior data in each spatiotemporal unit is aggregated; Construct a spatiotemporal grid, where each grid cell corresponds to a geographic location and time interval, and fill it with the corresponding aggregate data; The topological structure information of the traffic road network is mapped onto the space-time grid to reflect the road connectivity.

[0025] Specifically, the preprocessed data is input into the pre-trained traffic behavior prediction model to generate a local spatiotemporal graph including: Arrange the traffic topology at different times in the space-time grid in order, establish a sliding window on the traffic topology sequence, and slide backward with a preset step size; connect the road nodes in the traffic behavior data in the sliding window with themselves at the previous moment and the next moment, add self-loops, and construct the adjacency matrix of the local space-time graph; based on the constructed local space-time topology, form a local space-time graph data matrix.

[0026] It should be noted that the traffic behavior prediction model in this embodiment is a type of traffic behavior prediction model based on deep learning with specified features, where the specified features are: for a serialized model input containing several steps, the model output is a sequence with the same number of steps, and the output of each step is only related to all inputs up to this step, and is independent of all inputs after this time step.

[0027] In an embodiment of the present application, the window slides backward on the time series composed of historical data with a preset step size, and before each slide, the road nodes on each time slice in the window are connected to themselves at the adjacent time, and self-loops are added to form a new topology, namely, a local space-time graph, and then the traffic data of the time slices in the window are combined to construct a data matrix corresponding to the local space-time graph.

[0028] It should be noted that the above step S100 provides high-quality input for the subsequent prediction model by acquiring and preprocessing historical traffic behavior data, ensuring that the model can generate a local spatiotemporal graph based on accurate and highly relevant data, thereby improving prediction accuracy and reliability, effectively filtering out noise and irrelevant information, enhancing data consistency and integrity, and making traffic behavior patterns easier to be captured and understood by the model.

[0029] In the embodiment of the present application, the above step S200 specifically includes the following sub-steps B1 to B3: In sub-step B1: using a convolutional neural network to extract features from the local spatiotemporal graph, including: Perform Chebyshev graph convolution on the local space-time graph data matrix and the adjacency matrix of the local space-time graph; The output after graph convolution is activated by ReLU function to obtain the feature matrix, which is then segmented. The redundant information of the segmented data is eliminated by maximum pooling to obtain the local spatiotemporal feature matrix. If the split feature matrix meets the screening characteristics, the feature matrix information that meets the screening characteristics is retained and imported into a new valid information set; if the split feature matrix does not meet the screening characteristics, it is judged as redundant information, removed, and not imported into the valid information set; After selectively screening the characteristic information, the effective information set consisting of the effective information is sorted to obtain a local spatiotemporal characteristic matrix.

[0030] In sub-step B2: performing time series analysis on the feature extraction results to obtain analysis and prediction results of traffic behavior, including: The local spatiotemporal feature matrix is ​​input into the time series analysis unit of the traffic behavior prediction model, where Figure 2 The timing analysis unit shown includes three gating mechanisms, namely, input gate, forget gate and output gate; The input gate is used to receive the local spatiotemporal feature matrix input at the current moment and the hidden state of the previous time step, and calculate the first threshold as the activation value of the input gate through the activation function; The forget gate is used to calculate the second threshold based on the current input and the hidden state of the previous time step to determine the information that needs to be retained in the cell state of the previous time step; The output gate calculates the third threshold through the activation function, and multiplies the third threshold with the cell state adjusted by the hyperbolic tangent function to obtain the hidden state of the time step, and obtains the analysis and prediction results of the traffic behavior based on the hidden state.

[0031] In an optional embodiment, the time series analysis unit is constructed by a stack of multiple LSTM blocks, in which the output of a specific network at state s−1 and the output of the previous network at state s are used as input data of each LSTM block at state s, and the formula is expressed as: ; ; ; ; in, represents the input gate activation value, that is, the first threshold, represents the forget gate activation value, i.e., the second threshold, Indicates the unit status, represents the output gate activation value, that is, the third threshold, and Respectively represent input and the hidden state at the previous time step The weight matrix of and Denote the weight matrices of the input and previous hidden state, respectively, and Denote the weight matrices of the input and previous hidden state, respectively, and Denote the weight matrices of the input and previous hidden state, respectively, represents the bias term of the input gate, represents the bias term of the forget gate, represents the bias term for updating the unit state, represents the bias term of the output gate, represents the input of the current time step, represents the hidden state of the previous time step; It should be noted that the input gate activation value is calculated Used to determine the current input information How much information should be stored in the cell state and calculate the activation value of the forget gate Used to determine how much information of the cell state in the previous time step needs to be retained or forgotten, and calculate the output gate activation value Used to decide how much information in the updated cell state needs to be output as the current hidden state.

[0032] In an optional embodiment, the equation representing the updating of the unit state includes: Forget Gate Determines whether to retain the cell state from the previous time step Part of the information, input gate Controls the addition of new information. The added information is passed through The activation function adjusts the range, where the hidden state is represented as: ; in, Indicates the updated unit status, represents the output gate activation value; Should be explained, hidden state Through the output gate Scaling unit status and through The activation function is adjusted and this hidden state is passed to the next LSTM unit or used for prediction.

[0033] In sub-step B3: training and updating the traffic behavior prediction model, including: Divide the preprocessed traffic behavior data into a training set and a validation set; Train the traffic behavior prediction model through the training set and its corresponding labels, set the batch size and rounds, and update the weights in each iteration to minimize the loss; The parameters of the traffic behavior prediction model are dynamically adjusted, and the model performance is evaluated using a validation set.

[0034] It should be noted that the above step S200 uses a convolutional neural network to extract features from the local spatiotemporal graph, which can automatically learn and capture the complex patterns and nonlinear relationships in traffic behaviors, and maintain the temporal dynamic characteristics of the data through time series analysis, thereby effectively improving the accuracy and robustness of traffic behavior prediction.

[0035] In the embodiment of the present application, the above step S300 sets a risk assessment mechanism to perform quantitative assessment on the analysis and prediction results of traffic behavior to obtain the risk assessment results of the corresponding behavior, such as Figure 3 Specifically shown include: Set the risk assessment benchmark range of traffic behavior and judge the analysis and prediction results against the benchmark range. The risk assessment benchmark range is expressed as: ; in, is the mean, is the coefficient, is the standard deviation, To analyze the prediction results; If the analysis and prediction results are within the risk assessment benchmark, the regular monitoring process will be implemented to adjust the duration of traffic lights and optimize road use according to traffic flow; If the analysis prediction result is greater than , it indicates that the traffic flow has increased abnormally. When an abnormal increase in traffic is detected, real-time road condition information is obtained for secondary judgment. When the relative gap between traffic flow and road capacity exceeds a specified threshold, it indicates that the road system is in a clear congestion state. The intelligent transportation system will re-plan the traffic flow and guide vehicles to alternative routes. If there is no suitable alternative route, temporary traffic control measures will be implemented; If the analysis prediction result is less than When the traffic flow is lower than the minimum safe traffic flow demand of the road, the intelligent transportation system will be used to dispatch traffic flow of other sections. If the minimum safe traffic flow demand still cannot be met, the emergency response plan will be activated. If the analysis prediction result is greater than the minimum safe traffic demand of the road but less than , uses intelligent transportation technology to optimize traffic flow distribution and scheduling based on real-time data analysis, and adjusts the access rights and priorities of different road sections in real time by dynamically monitoring the relationship between traffic flow and road capacity.

[0036] In an optional embodiment, when the relative gap between traffic flow and road capacity exceeds a defined threshold, the setting of the defined threshold should be based on historical data statistical analysis and traffic engineering theory. For example, by analyzing the traffic flow data that has caused congestion in the past, the critical point between traffic flow and road capacity in different time periods, different weather conditions, special events, etc. can be found, and this critical point can be set as the defined threshold. The design capacity of the road can also be referred to, that is, the maximum number of vehicles that the road can accommodate per hour under ideal conditions. The actual traffic flow is compared with the design capacity. When the actual flow reaches or exceeds a certain proportion of the design capacity (for example, 85%-90%), it can be considered to be close to or in a congested state.

[0037] In an optional embodiment, the minimum safe road flow requirement refers to the minimum vehicle traffic volume that must be maintained on the road to ensure the safety and basic fluidity of road traffic. Too low traffic flow may cause the traffic flow on certain sections to become unstable. For example, if there are few vehicles on the highway, it may cause drivers to relax their vigilance, increasing the possibility of dangerous driving behaviors such as speeding and changing lanes at will. Therefore, when the actual traffic flow is lower than this requirement, measures need to be taken to adjust the traffic flow to ensure the continued safety and efficient operation of the transportation system.

[0038] It should be noted that the above step S300, by setting up a risk assessment mechanism, can convert the predicted results of traffic behavior into quantitative risk situations, provide intuitive risk indications, help decision makers quickly identify potential problems and take preventive measures, and improve the efficiency and accuracy of responding to traffic risks.

[0039] Embodiment 2: This embodiment provides a traffic behavior prediction and risk assessment system based on a deep neural network, including: A data processing module is used to obtain historical traffic behavior data and perform preprocessing, input the preprocessed data into a pre-trained traffic behavior prediction model, and generate a local spatiotemporal graph; The analysis and prediction module is used to extract features from the local spatiotemporal graph using a convolutional neural network and perform time series analysis on the results of the feature extraction to obtain analysis and prediction results of traffic behavior; The risk assessment module is used to set up a risk assessment mechanism to conduct quantitative assessment of the analysis and prediction results of traffic behaviors and obtain risk assessment results of corresponding behaviors.

[0040] It should be noted that the technical solution of the system for traffic behavior prediction and risk assessment based on deep neural networks and the technical solution of the above-mentioned method for traffic behavior prediction and risk assessment based on deep neural networks belong to the same concept. For details not described in detail in the technical solution of the system for traffic behavior prediction and risk assessment based on deep neural networks in this embodiment, please refer to the description of the technical solution of the above-mentioned method for traffic behavior prediction and risk assessment based on deep neural networks.

[0041] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0042] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a traffic behavior prediction and risk assessment method based on a deep neural network is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse.

[0043] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method proposed in the above embodiment is implemented.

[0044] The storage medium proposed in this embodiment and the method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0045] Through the above description of the implementation mode, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer's floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform the method of the embodiment of the present invention.

[0046] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0047] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The scheme in the embodiments of the present application may be implemented in various computer languages.

[0048] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0049] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0050] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0051] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0052] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A traffic behavior prediction and risk assessment method based on deep neural network, characterized in that: include: Acquire historical traffic behavior data and perform preprocessing, input the preprocessed data into a pre-trained traffic behavior prediction model, and generate a local spatiotemporal graph; Using a convolutional neural network to extract features from the local spatiotemporal graph, and performing time series analysis on the results of the feature extraction to obtain analysis and prediction results of traffic behavior; A risk assessment mechanism is set up to quantitatively evaluate the analysis and prediction results of the traffic behavior to obtain the risk assessment results of the corresponding behavior.

2. The traffic behavior prediction and risk assessment method based on deep neural network according to claim 1, characterized in that: The pre-treatment process includes: Identify missing values ​​in the acquired historical traffic behavior data and use interpolation to complete them according to the characteristics of the time series, and standardize the completed data; Dividing the standardized data into multiple spatiotemporal units according to preset time and space granularity, and aggregating the traffic behavior data in each spatiotemporal unit; Construct a spatiotemporal grid, where each grid cell corresponds to a geographic location and time interval, and fill it with the corresponding aggregate data; The topological structure information of the traffic road network is mapped onto the spatiotemporal grid to reflect the road connectivity.

3. The traffic behavior prediction and risk assessment method based on deep neural network as claimed in claim 2, characterized in that: Generating a local spatiotemporal graph comprises: Arrange the traffic topologies at different times in the space-time grid in order, establish a sliding window on the traffic topology sequence, and slide backward with a preset step size; Connecting the road nodes in the traffic behavior data in the sliding window with themselves at the previous moment and the next moment, adding self-loops, and constructing an adjacency matrix of the local space-time graph; According to the constructed local space-time topology, a local space-time graph data matrix is ​​formed.

4. The traffic behavior prediction and risk assessment method based on deep neural network as claimed in claim 3 is characterized in that: The extracting features of the local spatiotemporal graph using a convolutional neural network includes: Performing Chebyshev graph convolution on the local space-time graph data matrix and the adjacency matrix of the local space-time graph; The output after graph convolution is activated by using the ReLU function to obtain a feature matrix, and the feature matrix is ​​segmented, and the redundant information of the segmented data is eliminated by using the maximum pooling to obtain a local spatiotemporal feature matrix; If the split feature matrix meets the screening characteristics, the feature matrix information that meets the screening characteristics is retained and a new valid information set is imported; If the segmented feature matrix does not meet the screening characteristics, it is judged as redundant information and is removed, and the valid information set is not imported; After selectively screening the characteristic information, the effective information set consisting of the effective information is sorted to obtain a local spatiotemporal characteristic matrix.

5. The traffic behavior prediction and risk assessment method based on deep neural network as claimed in claim 4, characterized in that: The performing time series analysis on the result of the feature extraction comprises: Inputting the local spatiotemporal feature matrix into a time series analysis unit of a traffic behavior prediction model, wherein the time series analysis unit includes three gating mechanisms, namely an input gate, a forget gate, and an output gate; The input gate is used to receive the local spatiotemporal feature matrix input at the current moment and the hidden state of the previous time step, and calculate the first threshold value as the activation value of the input gate through the activation function; The forget gate is used to calculate a second threshold according to the current input and the hidden state of the previous time step to determine the information that needs to be retained in the cell state of the previous time step; The output gate calculates a third threshold value through an activation function, and multiplies the third threshold value by the cell state adjusted by the hyperbolic tangent function to obtain a hidden state of the time step, and obtains an analysis and prediction result of the traffic behavior based on the hidden state.

6. The method for predicting and assessing traffic behavior based on a deep neural network as claimed in claim 5, characterized in that: The training and updating of the traffic behavior prediction model includes: Divide the preprocessed traffic behavior data into a training set and a validation set; Training the traffic behavior prediction model through the training set and its corresponding labels, setting the batch size and rounds, and updating the weights in each iteration to minimize the loss; Dynamically adjust the parameters of the traffic behavior prediction model, and use the validation set to evaluate the model performance.

7. The method for predicting and assessing traffic behavior based on a deep neural network as claimed in claim 6, characterized in that: The setting of the risk assessment mechanism to quantitatively assess the analysis and prediction results of the traffic behavior includes: Set a risk assessment benchmark range for traffic behavior and compare the analysis and prediction results with the benchmark range; The risk assessment benchmark range is expressed as: ,in, is the mean, is the coefficient, is the standard deviation, To analyze the prediction results; If the analysis and prediction results are within the risk assessment benchmark, the conventional monitoring process is executed to adjust the duration of traffic lights and optimize road use according to traffic flow; If the analysis prediction result is greater than , it indicates that the traffic flow has increased abnormally. When an abnormal increase in traffic is detected, real-time road condition information is obtained for secondary judgment. When the relative gap between traffic flow and road capacity exceeds a specified threshold, it indicates that the road system is in a clear congestion state. The intelligent transportation system will re-plan the traffic flow and guide vehicles to alternative routes. If there is no suitable alternative route, temporary traffic control measures will be implemented; If the analysis prediction result is less than When the traffic flow is lower than the minimum safe traffic flow demand of the road, the intelligent transportation system will be used to dispatch traffic flow of other sections. If the minimum safe traffic flow demand still cannot be met, the emergency response plan will be activated. If the analysis and prediction result is greater than the minimum safe traffic demand of the road but less than , uses intelligent transportation technology to optimize traffic flow distribution and scheduling based on real-time data analysis, and adjusts the access rights and priorities of different road sections in real time by dynamically monitoring the relationship between traffic flow and road capacity.

8. A system using the traffic behavior prediction and risk assessment method based on a deep neural network as described in any one of claims 1 to 7, characterized in that: include: A data processing module, used to obtain and pre-process historical traffic behavior data, input the pre-processed data into a pre-trained traffic behavior prediction model, and generate a local spatiotemporal graph; An analysis and prediction module, used to extract features from the local spatiotemporal graph using a convolutional neural network, and perform time series analysis on the results of the feature extraction to obtain analysis and prediction results of traffic behavior; The risk assessment module is used to set a risk assessment mechanism to quantitatively evaluate the analysis and prediction results of the traffic behavior to obtain the risk assessment results of the corresponding behavior.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method described in claims 1 to 7 are implemented.

10. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that: When the computer executable instructions are executed by a processor, the steps of the method described in claims 1 to 7 are implemented.

Images