Traffic accident prediction method and system based on lie group manifold and time sequence knowledge graph

Abstract

The application relates to a traffic accident prediction method and system based on a Lie group manifold and a time sequence knowledge graph. The method comprises the following steps: collecting historical traffic accident data and performing a data cleaning operation; using a Lie group manifold mapping method to perform feature extraction and sorting to obtain a feature set; establishing a time sequence knowledge graph, performing deep learning through a temporal graph attention mechanism (TGAT), and optimizing to obtain a knowledge graph prediction model, predicting the dependency relationship between a prediction target entity and each entity, predicting the risk of a traffic accident based on the dependency relationship, and obtaining a prediction result. The TGAT can highlight the potential time dependency and multi-element information between traffic accident related indicators, effectively mine the dependency relationship through sequential learning, make up for the low accuracy of traditional link prediction models, and improve the accuracy of link prediction combining the Lie group manifold analysis, TGAT processing and time sequence knowledge graph.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation, in particular to a traffic accident prediction method and system based on Lie group manifold and time sequence knowledge graph. BACKGROUND

[0002] In today's society, with the rapid development of economy and the intensification of urbanization process, traffic safety problems have gradually become the focus of attention, especially in urban transportation systems, traffic accidents not only threaten people's life and property safety, but also seriously affect traffic efficiency and the smooth operation of the city. Traffic congestion and frequent accidents have become a common problem faced by many cities, which not only increases the travel time of citizens, but also has a negative impact on the environment. Therefore, in-depth research and accurate prediction of traffic accident risk have important social value and practical significance for improving traffic management level, reducing accidents and ensuring the safety of citizens' travel. Through scientific accident prediction and management, cities can effectively reduce the frequency of accidents, improve road safety, and optimize traffic flow, thereby enhancing the overall operational efficiency of the city and the quality of life of residents.

[0003] With the development of big data technology and the continuous progress of artificial intelligence algorithms, using historical traffic accident data to explore the rules of accident occurrence and achieve accurate road-level traffic accident risk prediction has become an important direction of intelligent transportation system construction. Big data technology enables traffic management departments to collect and analyze data from various sensors, surveillance cameras and social media in real time, and artificial intelligence uses deep learning and machine learning algorithms to model and predict these massive data. Data-driven intelligent transportation systems not only can monitor and analyze traffic flow in real time, but also can predict future traffic accident risks based on historical data and real-time information. This prediction capability can help traffic management departments take proactive measures to optimize traffic signal control, adjust traffic flow, thereby reducing the probability of accidents and improving overall traffic safety. In addition, accurate prediction can also help traffic departments develop more reasonable traffic planning and emergency plans, further enhancing the resilience of the transportation system and the ability to respond to emergencies. And with the rapid development of emerging technologies such as big data and artificial intelligence in recent years, relevant traffic data resources have been integrated. Using advanced intelligent algorithms to analyze these data can improve the accuracy of traffic accident prediction. Currently, support vector machines (SVM), negative binomial regression and decision trees are representative technologies that have played an important role in traffic accident analysis.

[0004] However, in the representation learning of traffic accident risk prediction, the problem of vector representation is particularly critical because they need to accurately capture and reflect the dynamic characteristics and multi-dimensional factors of traffic accident risk. Existing methods face many difficulties in converting traffic accident-related entities and attributes into vector representations. For example, traditional vector representations are difficult to adapt to the rapid changes of traffic accident risk over time, and the effective integration of time information requires models to handle different time scales and time dependencies. In addition, high-dimensional traffic accident data may lose key information when converted into low-dimensional vectors, affecting the predictive ability of the model. Traditional machine learning algorithms, such as support vector machines (SVM), decision trees, etc., have achieved certain results in traffic accident prediction, but they usually have limitations in the following aspects. First, these algorithms usually assume that features are independent of each other. This assumption, while simplifying the complexity of the model in many cases, ignores the potential complex relationships between features. Static data prediction usually uses knowledge graph link prediction models, but most traffic data is dynamic and has a time sequence, making it difficult to accurately and efficiently predict.

[0005] Therefore, the traditional traffic accident prediction method often has the problem of low prediction accuracy. SUMMARY

[0006] Therefore, in order to solve the above technical problems, a traffic accident prediction method and system based on Lie group manifold and time sequence knowledge graph are provided, which can improve the accuracy of traffic accident prediction.

[0007] A traffic accident prediction method based on Lie group manifold and time sequence knowledge graph, the method comprising:

[0008] Collecting historical traffic accident data and performing data cleaning operations on the historical traffic accident data to obtain cleaned historical traffic accident data;

[0009] Using a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, determine the non-linear correlation between variables, and obtain a feature set;

[0010] Establishing a time sequence knowledge graph according to the feature set, and constructing a preliminary prediction model based on the time sequence knowledge graph, the preliminary prediction model is optimized by deep learning through TGAT to obtain a knowledge graph prediction model;

[0011] Determine the time sequence information of the prediction target and input it into the knowledge graph prediction model, predict the dependency relationship between the prediction target entity and each entity, and predict the risk of traffic accidents based on the dependency relationship to obtain a prediction result.

[0012] In one embodiment, the method further comprises:

[0013] obtaining a traffic accident occurrence result, and comparing the traffic accident occurrence result with the prediction result to obtain a comparison result;

[0014] determining a model evaluation index, and evaluating the knowledge graph prediction model according to the model evaluation index based on the comparison result to obtain an evaluation result;

[0015] optimizing and adjusting the knowledge graph prediction model according to the evaluation result.

[0016] In one of the embodiments, the method further comprises:

[0017] decomposing a fourth-order tensor in the time-series knowledge graph to obtain each factor tensor;

[0018] mapping each factor tensor to a Lie group space to obtain a rotation matrix;

[0019] converting the rotation matrix into a skew-symmetric matrix using a logarithmic mapping to obtain a mapped factor tensor;

[0020] optimizing each factor tensor based on the mapped factor tensor.

[0021] In one of the embodiments, the constructing a preliminary prediction model based on the time-series knowledge graph comprises:

[0022] mapping entities and relationships in the feature set mapped to the Lie group flow into a graph structure through a relationship data RGCN model of a graph convolution network;

[0023] learning relationships between each entity based on the graph structure using a graph convolution operation;

[0024] realizing vectorization of entities and relationships in the time-series knowledge graph according to each entity and the learned relationships between entities, and constructing a preliminary prediction model.

[0025] In one of the embodiments, the method further comprises:

[0026] performing a basis decomposition regularization operation on entities and relationships in the feature set during use of the graph convolution operation;

[0027] performing multi-layer graph convolution on data after the basis decomposition regularization operation through the graph structure, and using softmax for classification and result prediction in the last layer of graph convolution to complete optimization of the preliminary prediction model.

[0028] In one of the embodiments, the preliminary prediction model is optimized through deep learning by TGAT to obtain a knowledge graph prediction model, comprising:

[0029] introducing a TGAT in the RGCN model;

[0030] extracting time information mapped to a Lie group manifold, approximating the time information by Monte Carlo integration through the TGAT, and encoding the time information using a time encoding function;

[0031] obtaining a learning weight matrix and calculating a self-attention mechanism according to the learning weight matrix;

[0032] optimizing the preliminary prediction model based on the self-attention mechanism and the encoded information to obtain a knowledge graph prediction model.

[0033] In one embodiment, the historical traffic accident data includes accident time, location, weather condition, and road type.

[0034] In one embodiment, the historical traffic accident data is subjected to a data cleaning operation to obtain cleaned historical traffic accident data, including:

[0035] The historical traffic accident data is subjected to an operation of removing abnormal values, missing values, or duplicate values to obtain cleaned historical traffic accident data.

[0036] A traffic accident prediction system based on Lie group manifold and time sequence knowledge graph, the system comprising:

[0037] a data processing module for collecting historical traffic accident data and performing a data cleaning operation on the historical traffic accident data to obtain cleaned historical traffic accident data;

[0038] a Lie group manifold mapping module for performing feature extraction and sorting on the cleaned historical traffic accident data using a Lie group manifold mapping method, determining the non-linear correlation between variables, and obtaining a feature set;

[0039] a model establishment module for establishing a time sequence knowledge graph according to the feature set and constructing a preliminary prediction model based on the time sequence knowledge graph, the preliminary prediction model being subjected to deep learning through a TGAT and optimization to obtain a knowledge graph prediction model;

[0040] a prediction module for determining time sequence information of a prediction target and inputting the time sequence information into the knowledge graph prediction model, predicting the dependency relationship between the prediction target entity and each entity, and predicting the risk of traffic accidents based on the dependency relationship to obtain a prediction result.

[0041] The traffic accident prediction method and system based on the Lie group manifold and the time sequence knowledge graph can make more accurate prediction of the time sequence knowledge graph in the field of traffic accident prediction than the prior art by using the Lie group manifold mapping method for feature extraction and sorting, combining the time sequence knowledge graph to train a knowledge graph prediction model for traffic accident prediction; the TGAT can highlight the potential time dependence and multi-element information between the traffic accident related indicators, effectively mine the dependence relationship by using the sequence learning, make up for the limitation of low accuracy of the traditional link prediction model, and improve the accuracy of the link prediction combining the Lie group manifold analysis, TGAT processing and time sequence knowledge graph. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 An application environment diagram of the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph in an embodiment;

[0043] Figure 2 A flowchart of the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph in an embodiment;

[0044] Figure 3 A flowchart of data processing using the Lie group manifold mapping in an embodiment;

[0045] Figure 4 A flowchart of graph convolution operation in an embodiment;

[0046] Figure 5 A flowchart of sequence learning in an embodiment;

[0047] Figure 6 A structural block diagram of the traffic accident prediction system based on the Lie group manifold and the time sequence knowledge graph in an embodiment;

[0048] Figure 7 A structural block diagram of the traffic accident prediction system based on the Lie group manifold and the time sequence knowledge graph in another embodiment;

[0049] Figure 8 An internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0051] The traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph provided by the embodiments of the present application can be applied in the application environment as shown in Figure 1 The traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph provided by the embodiments of the present application can be applied in the application environment as shown inFigure 1 As shown, the application environment includes a computer device 110. The computer device 110 can collect historical traffic accident data and perform data cleaning operations on the historical traffic accident data to obtain cleaned historical traffic accident data; the computer device 110 can use a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, determine the nonlinear correlation between variables, and obtain a feature set; the computer device 110 can establish a time sequence knowledge graph according to the feature set, and construct a preliminary prediction model based on the time sequence knowledge graph, the preliminary prediction model is learned deeply by TGAT and optimized to obtain a knowledge graph prediction model; the computer device 110 can determine the time sequence information of the prediction target and input it into the knowledge graph prediction model, predict the dependency relationship between the prediction target entity and each entity, and predict the risk of traffic accidents based on the dependency relationship to obtain a prediction result. The computer device 110 can be, but is not limited to, various personal computers, notebook computers, smartphones, robots, unmanned aerial vehicles, and other devices.

[0052] In one embodiment, as Figure 2 shown, a traffic accident prediction method based on Lie group manifold and time sequence knowledge graph is provided, including the following steps:

[0053] Step 202, collecting historical traffic accident data and performing data cleaning operations on the historical traffic accident data to obtain cleaned historical traffic accident data.

[0054] The computer device can collect historical traffic accident data, wherein in one embodiment, the historical traffic accident data includes accident time, location, weather conditions, road type and other related data.

[0055] After collecting the historical traffic accident data, data cleaning operations can be performed on the historical traffic accident data to obtain cleaned historical traffic accident data. Specifically, in one embodiment, a traffic accident prediction method based on Lie group manifold and time sequence knowledge graph is also provided, which includes a data cleaning process, and the specific process includes: removing outliers, missing values or duplicate values from the historical traffic accident data to obtain cleaned historical traffic accident data.

[0056] Step 204, using a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, determine the nonlinear correlation between variables, and obtain a feature set.

[0057] In the Lie group manifold mapping process, the computer device can use the Lie group manifold mapping method to extract and sort features from each data to determine the nonlinear correlation between variables and generate a feature set.

[0058] In one embodiment, the traffic accident prediction method based on Lie group manifold and time sequence knowledge graph further includes a Lie group manifold mapping process, and the specific process includes: decomposing a fourth-order tensor in the time sequence knowledge graph to obtain each factor tensor; mapping each factor tensor to a Lie group space to obtain a rotation matrix; converting the rotation matrix into a skew-symmetric matrix using a logarithmic mapping to obtain a mapped factor tensor; and optimizing each factor tensor based on the mapped factor tensor.

[0059] The computer device can decompose a fourth-order tensor Y in the time sequence knowledge graph: where Y is a target tensor, u r ,v r ,w r ,t r is a factor tensor, λ r is a scaling factor; and each factor tensor e is mapped to an SO(2) Lie group space to obtain a rotation matrix R: Then, the computer device can convert the rotation matrix on SO(2) into a skew-symmetric matrix in so(2) using a logarithmic mapping: where θ(R) is a rotation angle of the matrix R, and is calculated as: trace(R) is the trace of R, that is, the sum of the diagonal elements.

[0060] Then, the difference between the original factor tensor and the mapped factor tensor can be calculated: Then, the optimized factor tensor is used for tensor decomposition, and a loss function is optimized: The factor tensor is further optimized using a full multi-class logarithmic softmax loss function and N3 regularization: where λμ is the weight of N3 regularization, is the inverse of the relationship, and φ is a score function, to map the factor tensor to the Lie group manifold to obtain a mapped feature matrix.

[0061] In one embodiment, as shown in Figure 3 , the traffic accident prediction method based on Lie group manifold and time sequence knowledge graph further includes a process of using a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, as shown in Figure 3 , and the specific process includes: selecting traffic accident data indicators to determine parameter sequences; mapping factors in the data into a form suitable for Lie group manifold analysis; performing Lie group manifold mapping; and then performing feature extraction and feature fusion.

[0062] At step 206, the time sequence knowledge graph is established according to the feature set, and a preliminary prediction model is constructed based on the time sequence knowledge graph. The preliminary prediction model is deep-learned by TGAT and optimized to obtain a knowledge graph prediction model.

[0063] In the process of establishing the time sequence knowledge graph prediction model, the knowledge graph model is trained based on the feature set of the Lie group manifold analysis to predict the relationship between the target entity and other entities, and the time sequence knowledge graph prediction model is generated. Then, deep learning can be performed by TGAT to optimize the preliminary prediction model.

[0064] In one embodiment, the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph can further include a process of constructing a preliminary prediction model, and the specific process includes: mapping the entities and relationships in the feature set mapped to the Lie group flow into a graph structure by using an RGCN model; learning the relationships between the entities based on the graph structure by using a graph convolution operation; and realizing the vectorization of the entities and relationships in the time sequence knowledge graph according to the entities and the learned relationships between the entities, and constructing a preliminary prediction model.

[0065] Specifically, the computer device can use the RGCN model to map the entities and relationships into a graph structure, and then use the graph convolution operation to learn the relationships between the entities, thereby realizing the vectorization of the entities and complex semantic relationships in the time sequence knowledge graph.

[0066] In one embodiment, the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph can further include a process of performing a graph convolution operation, as shown in Figure 4 The specific process includes: inputting data, a feature matrix transformed by the Lie group manifold; then performing feature embedding; then performing graph convolution operation; and finally outputting data by using an output layer.

[0067] In one embodiment, the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph can further include a process of model optimization, and the specific process includes: performing basis decomposition regularization operation on the entities and relationships in the feature set during the use of the graph convolution operation; performing multi-layer graph convolution on the data after the basis decomposition regularization operation in the graph structure, and using softmax for classification and result prediction in the last layer of graph convolution, thereby completing the optimization of the preliminary prediction model.

[0068] In this embodiment, the mapping to the Lie group manifold, the labeled multi-graph G with direction can be used for graph convolution operation to update the hidden state of the node. Where V is the node (entity) set, E is the labeled edge (relationship) set, and R is the set of relationship types. Given a node v i The hidden state of the first layer Its new state at the l+1 layer It can be calculated by the formula: Where, The neighbor node index set of node v i under the relationship r, c i,r is a problem-specific normalization constant, which can be fixed or learned, is the weight matrix for relationship r, is the weight matrix of self-connection, and σ is the activation function.

[0069] Then, the basis decomposition line regularization operation can be performed in the convolution process: After passing through multiple layers of graph convolution layers, and using softmax at the last layer for multi-class classification and result prediction, the loss function is cross-entropy loss: Where Y is the labeled node index set, is the kth entry of the network output, corresponding to the labeled node i, t ik is the corresponding real label, and the model is optimized accordingly.

[0070] In one embodiment, the traffic accident prediction method based on Lie group manifold and time sequence knowledge graph provided can further include the process of introducing a time sequence graph attention mechanism, and the specific process includes: introducing TGAT in the RGCN model; extracting time information mapped to the Lie group manifold, using TGAT to approximate by Monte Carlo integration, and using a time encoding function to encode the time information; obtaining a learning weight matrix, calculating a self-attention mechanism according to the learning weight matrix; and optimizing the preliminary prediction model based on the self-attention mechanism and the encoded information to obtain a knowledge graph prediction model.

[0071] Specifically, TGAT can be introduced in the RGCN model to add information in the time sequence knowledge graph for sequential learning. As shown in Figure 5 The process of sequential learning can include: determining the feature matrix after the Lie group manifold; then performing functional time encoding; calculating the self-attention mechanism through the time sequence graph attention layer; and then entering the convolution layer of the RGCN.

[0072] Specifically, the time information mapped to the Lie group manifold can be encoded through a time domain T to R drThe time kernel K can be expressed as: K(t1, t2) = <phi(t1), phi(t2)>; where phi is obtained by Fourier transform: K(t1, t2) = E w [cos(w(t1-t2))] ; where w is the frequency, i is the imaginary unit, and E represents the expected value; the time encoding function phi d (t) is expressed as: where d is the encoding dimension, w i is the frequency parameter of random sampling, and t is the timestamp; then, the combination of node and time encoding can be performed: where h i (t i ) is the hidden state of node v i at time t i , represents vector splicing.

[0073] where the calculation of the self-attention mechanism can include: self-attention query (Query) calculation: Q(t) = Z(t)W Q ; self-attention key (Key) calculation: K(t) = Z(t)W K ; self-attention value (Value) calculation: V(t) = Z(t)W V ; where W Q ,W K ,W V are learnable weight matrices. The attention weight is: where N(v0; t) is the set of neighbor nodes of the target node V0 at time t. Where the neighbor information aggregation can be expressed as: where V i is the value of node V V obtained by linear transformation W i ; V i = h i (t)W V ; finally, enter the feedforward neural network: FFN is a feedforward neural network; enter the convolutional layer of RGCN.

[0074] Step 208, determine the time sequence information of the prediction target and input it into the knowledge graph prediction model, predict the dependency relationship between the prediction target entity and each entity, and predict the risk of traffic accidents based on the dependency relationship, to obtain the prediction result.

[0075] Specifically, in the process of data prediction, computer equipment can use the trained knowledge graph prediction model according to the time series information of the prediction target, and perform link prediction on the time series knowledge graph in the field of traffic accidents by predicting the dependency relationship between the target entity and other entities, thereby obtaining the prediction result.

[0076] In this embodiment, data collection is first performed to collect traffic accident data, including accident time, location, weather conditions, road type and other relevant data; data cleaning is then performed to clean the collected data, including removing outliers, missing values ​​or duplicate values; Lie group manifold analysis is performed, and the Lie group manifold analysis method is used to extract and sort features of various environmental data to determine the nonlinear correlation between variables and generate a feature set; then a time series knowledge graph model is established, and a knowledge graph model is constructed and trained based on the feature set of Lie group manifold analysis to predict the relationship between the target entity and other entities and generate a prediction model; and data prediction is performed, according to the time series information of the prediction target, using the trained knowledge graph prediction model, by predicting the dependency relationship between the target entity and other entities, a link prediction of the risk of traffic accidents is performed; model evaluation is performed, and finally the prediction results are evaluated, including error analysis, accuracy analysis, etc., to determine the reliability of the prediction model. Due to the adoption of the TGAT temporal graph attention mechanism, the potential temporal dependencies and multivariate information between traffic accident-related indicators can be highlighted, and their dependencies can be effectively mined using serialized learning, which makes up for the limitations of the low accuracy of traditional link prediction models and improves the accuracy of link prediction that combines Lie group manifold analysis and temporal knowledge graphs.

[0077] It should be understood that, although the various steps in the above flow chart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the above flow chart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0078] In one embodiment, Figure 6 As shown, a traffic accident prediction system based on Lie group manifold and time series knowledge graph is provided, including: a data processing module 610, a Lie group manifold mapping module 620, a model building module 630 and a prediction module 640, wherein:

[0079] The data processing module 610 is configured to collect historical traffic accident data, and perform data cleaning on the historical traffic accident data to obtain cleaned historical traffic accident data.

[0080] The Lie group manifold mapping module 620 is configured to use a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, determine nonlinear correlations between variables, and obtain a feature set.

[0081] The model establishing module 630 is configured to establish a time sequence knowledge graph according to the feature set, and construct a preliminary prediction model based on the time sequence knowledge graph. The preliminary prediction model is subjected to deep learning by TGAT and is optimized to obtain a knowledge graph prediction model.

[0082] The prediction module 640 is configured to determine time sequence information of a prediction target and input the time sequence information into the knowledge graph prediction model, predict a dependency relationship between the prediction target entity and each entity, and predict a traffic accident occurrence risk based on the dependency relationship to obtain a prediction result.

[0083] As shown in FIG. 6, Figure 7 In one embodiment, the traffic accident prediction system based on the Lie group manifold and the time sequence knowledge graph further includes a model evaluation and adjustment module 650. The model evaluation and adjustment module 650 is configured to obtain a traffic accident occurrence result, compare the traffic accident occurrence result with the prediction result to obtain a comparison result, determine a model evaluation index, evaluate the knowledge graph prediction model based on the comparison result according to the model evaluation index to obtain an evaluation result, and optimize and adjust the knowledge graph prediction model according to the evaluation result.

[0084] In one embodiment, the Lie group manifold mapping module 620 is further configured to decompose a fourth-order tensor in the time sequence knowledge graph to obtain each factor tensor, map each factor tensor to a Lie group space to obtain a rotation matrix, convert the rotation matrix into a skew-symmetric matrix using a logarithmic mapping to obtain a mapped factor tensor, and optimize each factor tensor based on the mapped factor tensor.

[0085] In one embodiment, the model establishing module 630 is further configured to map entities and relationships in the feature set mapped to the Lie group into a graph structure by using an RGCN model, learn relationships between each entity based on the graph structure by using a graph convolution operation, and realize vectorization of the entities and relationships in the time sequence knowledge graph according to each entity and the learned relationships between the entities to construct the preliminary prediction model.

[0086] In one embodiment, the model establishing module 630 is further configured to perform a base decomposition regularization operation on the entities and the relationships in the feature set during the use of the graph convolution operation; perform multi-layer graph convolution on the data after the base decomposition regularization operation in the graph structure, and use softmax in the last layer of the graph convolution to perform classification and result prediction, thereby optimizing the preliminary prediction model.

[0087] In one embodiment, the model establishing module 630 is further configured to introduce a TGAT time series graph attention mechanism in the RGCN model; extract time information mapped to a Lie group manifold, use the TGAT to perform Monte Carlo integration approximation, and use a time encoding function to encode the time information; obtain a learning weight matrix, calculate a self-attention mechanism according to the learning weight matrix; and optimize the preliminary prediction model based on the self-attention mechanism and the encoded information, thereby obtaining the knowledge graph prediction model.

[0088] In one embodiment, the historical traffic accident data includes accident time, location, weather condition, and road type.

[0089] In one embodiment, the data processing module 610 is further configured to perform an operation of removing outliers, missing values, or duplicate values on the historical traffic accident data, thereby obtaining the cleaned historical traffic accident data.

[0090] In one embodiment, a computer device is provided, which can be a terminal, and an internal structure diagram of the computer device can be as shown in Figure 8 The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. The processor of the computer device is configured 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 operating system and the computer program in the non-volatile storage medium to run. The network interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a traffic accident prediction method based on a Lie group manifold and a time series knowledge graph. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, trackball, or touchpad arranged on the shell of the computer device. In addition, the input device can be an external keyboard, touchpad, or mouse, etc.

[0091] Those skilled in the art can understand that, Figure 8The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0092] In one embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph when executing the computer program.

[0093] In one embodiment, a computer readable storage medium is provided, storing a computer program, and the processor implements the steps of the traffic accident prediction method based on the Lie group manifold and the time sequence knowledge graph when executing the computer program.

[0094] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium and can include the processes of the above-mentioned embodiments when executed. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0095] Each technical feature of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of each technical feature in the above embodiments are not described, but as long as the combination of technical features does not exist, it should be considered as the scope of the present application.

[0096] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific and detailed manner, but should not be construed as limiting the scope of the patent. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these are all within the scope of the present application. Therefore, the scope of protection of the patent of the present application should be subject to the appended claims.

Claims

1. A traffic accident prediction method based on Lie group manifold and time-series knowledge graph, characterized in that, The method comprises: collecting historical traffic accident data, and performing data cleaning operation on the historical traffic accident data to obtain cleaned historical traffic accident data; extracting and sorting features from the cleaned historical traffic accident data using a Lie group manifold mapping method to determine the non-linear correlation between variables and obtain a feature set, including: decomposing a fourth-order tensor in the cleaned historical traffic accident data to obtain various factor tensors; mapping each of the factor tensors to a Lie group space to obtain a rotation matrix; converting the rotation matrix into a skew-symmetric matrix using a logarithmic mapping to obtain a mapped factor tensor; and optimizing each factor tensor based on the mapped factor tensor; establishing a time sequence knowledge graph according to the feature set and constructing a preliminary prediction model based on the time sequence knowledge graph, including: mapping entities and relationships in the feature set mapped to the Lie group flow into a graph structure through a relationship data RGCN model of a graph convolution network; learning the relationships between entities based on the graph structure using a graph convolution operation; and realizing vectorization of entities and relationships in the time sequence knowledge graph according to each of the entities and the learned relationships between entities to construct a preliminary prediction model; the preliminary prediction model is subjected to deep learning through TGAT and optimized to obtain a knowledge graph prediction model; determining time sequence information of a prediction target and inputting the time sequence information into the knowledge graph prediction model to predict the dependency relationship between the prediction target entity and each entity, and predicting the risk of traffic accidents based on the dependency relationship to obtain a prediction result.

2. The traffic accident prediction method based on Lie group manifold and timing knowledge graph according to claim 1, characterized in that, The method further comprises: obtaining a traffic accident occurrence result and comparing the traffic accident occurrence result with the prediction result to obtain a comparison result; determining a model evaluation index, evaluating the knowledge graph prediction model according to the model evaluation index based on the comparison result to obtain an evaluation result; optimizing and adjusting the knowledge graph prediction model according to the evaluation result. 3.The traffic accident prediction method based on Lie group manifold and time-series knowledge graph according to claim 1, characterized in that, The method further comprises: performing basis decomposition regularization operation on entities and relationships in the feature set during the use of the graph convolution operation; The data after the base decomposition regularization operation is subjected to multi-layer graph convolution in the graph structure, and the last layer graph convolution uses Classification and result prediction are performed, and optimization of the preliminary prediction model is completed. 4.The traffic accident prediction method based on Lie group manifold and time-series knowledge graph according to claim 1, characterized in that, the preliminary prediction model is subjected to deep learning through TGAT and optimized to obtain a knowledge graph prediction model, including: introducing TGAT into the RGCN model; extracting time information mapped to the Lie group manifold, approximating the time information using the TGAT through Monte Carlo integration, and encoding the time information using a time encoding function; obtaining a learning weight matrix, calculating a self-attention mechanism according to the learning weight matrix; optimizing the preliminary prediction model based on the self-attention mechanism and the encoded information to obtain a knowledge graph prediction model. 5.The traffic accident prediction method based on Lie group manifold and timing knowledge graph according to claim 1, characterized in that, The historical traffic accident data includes accident time, location, weather condition, and road type. 6.The traffic accident prediction method based on Lie group manifold and timing knowledge graph according to claim 1, characterized in that, The data cleaning operation on the historical traffic accident data to obtain cleaned historical traffic accident data includes: performing an operation of removing outliers, missing values, or duplicate values on the historical traffic accident data to obtain cleaned historical traffic accident data. 7.A traffic accident prediction system based on Lie group manifold and time-series knowledge graph, characterized in that, The system comprises: The data processing module is configured to collect historical traffic accident data, and perform data cleaning on the historical traffic accident data to obtain cleaned historical traffic accident data. The Lie group manifold mapping module is configured to use a Lie group manifold mapping method to extract and sort features from the cleaned historical traffic accident data, determine nonlinear correlations between variables, and obtain a feature set, including: decomposing a fourth-order tensor in the cleaned historical traffic accident data to obtain various factor tensors; mapping each factor tensor to a Lie group space to obtain a rotation matrix; converting the rotation matrix into a skew-symmetric matrix using logarithmic mapping to obtain a mapped factor tensor; and optimizing each factor tensor based on the mapped factor tensor. The model establishing module is configured to establish a time-series knowledge graph based on the feature set, and construct a preliminary prediction model based on the time-series knowledge graph, including: mapping entities and relationships in the feature set mapped to the Lie group to a graph structure through a relationship data RGCN model of a graph convolution network; learning relationships between entities based on the graph structure using graph convolution operations; and realizing vectorization of entities and relationships in the time-series knowledge graph according to each entity and the learned relationships between entities, and constructing a preliminary prediction model; the preliminary prediction model is subjected to deep learning through TGAT and is optimized to obtain a knowledge graph prediction model. The prediction module is configured to determine time-series information of a prediction target and input the time-series information into the knowledge graph prediction model, predict a dependency relationship between the prediction target entity and each entity, and predict a traffic accident occurrence risk based on the dependency relationship to obtain a prediction result. 8.The traffic accident prediction system based on Lie group manifold and time-sequenced knowledge graph according to claim 7, wherein, The system further includes a model evaluation and adjustment module configured to obtain a traffic accident occurrence result, compare the traffic accident occurrence result with the prediction result, and obtain a comparison result. A model evaluation index is determined, the knowledge graph prediction model is evaluated based on the model evaluation index according to the comparison result to obtain an evaluation result, and the knowledge graph prediction model is optimized and adjusted according to the evaluation result.

Images