Accident analysis and intelligent intervention method and system based on intelligent connected automobile

Abstract

The invention discloses an accident analysis and intelligent intervention method and system based on an intelligent networked automobile, and particularly relates to the technical field of accident analysis, which comprises the following steps: when the intelligent networked automobile approaches an accident-prone place, extracting important features from historical traffic data in the accident-prone place of the intelligent networked automobile by using a t-SNE dimension reduction method; acquiring real-time traffic data of the intelligent networked automobile, and determining similar information of the traffic data of the intelligent networked automobile; bayesian inference is carried out based on historical traffic data in an accident-prone place of the intelligent networked automobile and real-time traffic data of the intelligent networked automobile, and probability information of the traffic data of the intelligent networked automobile is determined by analyzing statistical data and constructing a machine learning mode; through comprehensive analysis of the similar information and the probability information of the traffic data, the risk that the intelligent networked automobile enters the accident-prone place is evaluated, the road traffic safety can be improved, and the accident occurrence rate is reduced.

Description

Technical Field

[0001] This invention relates to the field of accident analysis technology, and more specifically, to a method and system for accident analysis and intelligent intervention based on intelligent connected vehicles. Background Technology

[0002] With the rapid development of intelligent connected vehicle technology, more and more cars are equipped with sensors, vehicle communication systems and intelligent driving assistance systems. These technologies enable cars to acquire and transmit vehicle operating status, road information and surrounding environment in real time. At present, the prediction and prevention of accidents in intelligent transportation systems rely on the fusion analysis of historical data and real-time data. However, existing technologies mainly focus on single-dimensional monitoring, such as vehicle speed, distance, and vehicle density, without fully considering various dynamic factors in accident-prone areas. Furthermore, there are still significant shortcomings in the identification and early warning of high-risk vehicles, failing to achieve accurate early warning and dynamic intervention.

[0003] To address the aforementioned shortcomings, a technical solution is provided. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an accident analysis and intelligent intervention method and system based on intelligent connected vehicles to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An accident analysis and intelligent intervention method based on intelligent connected vehicles includes the following steps:

[0007] S1: Based on the sensors and vehicle communication system equipped in intelligent connected vehicles, whenever an intelligent connected vehicle approaches an accident-prone area, important features are extracted from the historical traffic data of the accident-prone area by using the t-SNE dimensionality reduction method, and real-time traffic data of the intelligent connected vehicle is obtained to determine the similarity information of the traffic data of the intelligent connected vehicle.

[0008] S2: Based on historical traffic data and real-time traffic data of intelligent connected vehicles in areas with frequent accidents, Bayesian inference is performed. By analyzing statistical data and constructing machine learning methods, the probability information of intelligent connected vehicle traffic data is determined.

[0009] S3: By comprehensively analyzing similar and probabilistic information from traffic data, construct a traffic accident assessment model for intelligent connected vehicles approaching accident-prone areas to assess the risk of intelligent connected vehicles entering accident-prone areas;

[0010] S4: Based on the location and driving information of intelligent connected vehicles when driving in accident-prone areas, and by collecting traffic accident assessment models of intelligent connected vehicles when driving in accident-prone areas, determine the traffic accident assessment information of intelligent connected vehicles when driving in accident-prone areas, and identify intelligent connected vehicles that are high-risk vehicles in accident-prone areas.

[0011] In a preferred embodiment, similar information regarding intelligent connected vehicle traffic data includes:

[0012] The similarity information of traffic data is represented by the dimensionality-reduced feature similarity coefficient;

[0013] The logic for obtaining the dimensionality reduction feature similarity coefficient is as follows: collect historical traffic data of accident-prone areas, extract features from the historical traffic data of accident-prone areas, including time features, traffic flow features, weather features and driving features, fuse the time features, traffic flow features, weather features and driving features, and obtain the historical comprehensive traffic feature vector of each accident in the accident-prone area through standardization processing, and obtain the real-time comprehensive traffic feature vector with the same dimension as the comprehensive traffic feature vector based on real-time traffic data.

[0014] The historical comprehensive traffic feature vector of each accident is input into the t-SNE algorithm for dimensionality reduction. The embedding coordinates of the historical comprehensive traffic feature vector of each accident in the low-dimensional space are represented as: X = [x1, x2, x3, ..., x...]. n ]; where n is the number of accidents in accident-prone areas;

[0015] By projecting the real-time integrated traffic feature vector onto the embedded coordinates of the historical integrated traffic feature vector in the low-dimensional space, the low-dimensional projection of the real-time integrated traffic feature vector is represented as: R, where the mapping function is learned by training the t-SNE algorithm, and the projection of the real-time integrated traffic feature vector is calculated using the trained mapping function. The mapping function is adjusted to the low-dimensional projection coordinates by the gradient descent algorithm to minimize the KL divergence between the real-time integrated traffic feature vector and the historical integrated traffic feature vector.

[0016] After dimensionality reduction using the t-SNE algorithm, the distance between the real-time integrated traffic feature vector and the historical integrated traffic feature vector is calculated using Euclidean distance. The calculation formula is as follows: Where i = 1, 2, 3, ..., n, x i This is the historical comprehensive traffic feature vector at the time of the i-th accident;

[0017] The similarity coefficient of the dimensionality-reduced features is calculated using the Gaussian kernel function, and the formula is as follows: Among them, TZ xs σ is the similarity coefficient for the dimensionality reduction features, and σ is the bandwidth parameter.

[0018] In a preferred embodiment, the probabilistic information of the intelligent connected vehicle traffic data includes:

[0019] The probabilistic information of traffic data is represented by post-accident probability coefficients;

[0020] The logic for obtaining the post-accident probability coefficient is as follows: Based on historical traffic data of accident-prone areas and current real-time traffic data of intelligent connected vehicles, determine the characteristics and accident labels of each accident in the historical traffic data, and determine the characteristics of the real-time traffic data. The prior probability P(A) is determined by the ratio of the number of accidents in the historical data to the total number of samples.

[0021] By training a logistic regression model, the predicted probability of an accident is calculated based on different combinations of features. The expression for logistic regression is: Where A is the accident label, A=1 indicates an accident has occurred, A=0 indicates no accident, and w1, w2, ..., w n are the weights of different features, and b is the bias term;

[0022] By training the weights by maximizing the log-likelihood function, and using the trained logistic regression model, the likelihood function at the time of the accident is obtained based on the predicted value of the accident probability of the logistic regression model, that is, the conditional probability of the accident under different features. Based on the features of real-time traffic data, the conditional probability of the accident between the features of real-time traffic data and the accident is determined, and the conditional probability of the accident between the features of real-time traffic data and the accident is denoted as: P(R|A).

[0023] Based on the current real-time traffic data of intelligent connected vehicles, the characteristics of the real-time traffic data are determined, and the evidence P(R) is determined by the probability of the real-time traffic data under the condition that the accident has occurred and has not occurred.

[0024] The posterior probability coefficient of the accident is calculated using the following formula: Among them, SG hy This represents the post-accident probability coefficient.

[0025] In a preferred embodiment, a traffic accident assessment model is constructed when an intelligent connected vehicle approaches an accident-prone area to assess the risk of the intelligent connected vehicle entering the accident-prone area, including:

[0026] The posterior probability coefficients of the accident are weighted and calculated. When the intelligent connected vehicle approaches an accident-prone area, a traffic accident assessment model for the intelligent connected vehicle is generated in real time, producing a traffic accident assessment coefficient. The formula for calculating the traffic accident assessment coefficient is: PG jt =α1TZ xs +α2SG hy Among them, PGjt Let α1 and α2 be the posterior probability coefficients of the accident and the proportional coefficients of the posterior probability coefficients of the accident, respectively, and both α1 and α2 are greater than 0.

[0027] Set a threshold for the traffic accident assessment coefficient, and mark the traffic accident assessment coefficient threshold as: PG yz When it is detected that the intelligent connected vehicle is heading to the next accident-prone area, the traffic accident assessment coefficient of the intelligent connected vehicle for the next accident-prone area is obtained in real time, and the traffic accident assessment coefficient of the intelligent connected vehicle is compared with the traffic accident assessment coefficient threshold.

[0028] If the traffic accident assessment coefficient is greater than the traffic accident assessment coefficient threshold, a warning signal will be generated; if the traffic accident assessment coefficient is less than the traffic accident assessment coefficient threshold, no warning signal will be generated.

[0029] In a preferred embodiment, the location information of the intelligent connected vehicle when driving in an accident-prone area includes:

[0030] The location information of intelligent connected vehicles when driving in accident-prone areas is represented by the relative distance coefficient.

[0031] The logic for obtaining the relative distance coefficient is as follows: After the intelligent connected vehicle enters an accident-prone area, the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is collected within the monitoring range, and the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is marked as: JL m,j Where m is the number of the other vehicles in the accident-prone area, m = 1, 2, 3, ..., M, M is a positive integer, j is the sampling number in the monitoring interval, j = 1, 2, 3, ..., J, J is a positive integer;

[0032] Calculate the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval, and label the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas as: JL avg and JL std ;in,

[0033] The coefficient of variation of the relative distance between the intelligent connected vehicle and other vehicles in accident-prone areas within the monitoring interval is calculated using the following formula: Among them, JL by The coefficient of variation of the relative distance between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval;

[0034] Calculate the relative relationship coefficient of vehicle distance. Among them, CJ xd This is the relative relationship coefficient between vehicle distances.

[0035] In a preferred embodiment, the driving information of the intelligent connected vehicle when driving in an accident-prone area includes:

[0036] The driving information of intelligent connected vehicles when driving in accident-prone areas is represented by the risk behavior divergence coefficient.

[0037] The logic for obtaining the divergence coefficient of the dangerous behavior is as follows: Based on historical traffic data from accident-prone areas, the dangerous driving behaviors of intelligent connected vehicles are determined. Based on historical data, the distribution of historical dangerous driving behaviors and the distribution of real-time dangerous driving behaviors are constructed using the kernel density estimation method. The distribution of historical dangerous driving behaviors is labeled as: P(h f The distribution of real-time dangerous driving behavior is labeled as: Q(h) f ), where f = 1, 2, 3, ..., F, F is a positive integer, f is the number of different dangerous driving behaviors, h f These are random variables representing different dangerous driving behaviors;

[0038] The formula for calculating the divergence coefficient of dangerous behavior is as follows: Among them, SD wx Y is the divergence coefficient of dangerous behavior. f Weights for different dangerous driving behaviors.

[0039] In a preferred embodiment, determining traffic accident assessment information when an intelligent connected vehicle is driving in an accident-prone area includes:

[0040] The traffic accident assessment information of intelligent connected vehicles driving in accident-prone areas will be represented by a traffic accident risk coefficient.

[0041] The logic for obtaining the traffic accident risk coefficient is as follows: Based on the real-time generation of a traffic accident assessment model for intelligent connected vehicles when they approach accident-prone areas, the traffic accident assessment model of the intelligent connected vehicle after entering the accident-prone area is obtained, resulting in traffic accident assessment coefficients at different times within the monitoring interval. These coefficients are then compared with a threshold value to obtain the traffic accident assessment coefficients within the monitoring interval that exceed the threshold. These traffic accident assessment coefficients are then marked as PG. k ;

[0042] The formula for calculating the traffic accident risk coefficient is as follows: Among them, FX pg This represents the risk coefficient for traffic accidents.

[0043] In a preferred embodiment, identifying intelligent connected vehicles that are considered high-risk vehicles in accident-prone areas includes:

[0044] Based on the comprehensive analysis of the location and driving information of intelligent connected vehicles entering accident-prone areas, and combined with traffic accident assessment information during the operation of intelligent connected vehicles in accident-prone areas, a risk assessment model is constructed by weighting the relative distance coefficient, the risk behavior divergence coefficient, and the traffic accident risk coefficient. This generates a risk assessment coefficient for intelligent connected vehicles, calculated using the formula: PG fx =β1CL xd -β2SD wx +β3FX pg Among them, PG fx β1, β2, and β3 are the relative relationship coefficients of vehicle distance, the divergence value coefficient of dangerous behavior, and the proportional coefficients of risk assessment coefficients, respectively, and β1, β2, and β3 are all greater than 0.

[0045] Set a risk assessment coefficient threshold, obtain the risk assessment coefficient of all intelligent connected vehicles in accident-prone areas, compare the risk assessment coefficient of each intelligent connected vehicle with the risk assessment coefficient threshold, if the risk assessment coefficient is greater than the risk assessment coefficient threshold, then mark the intelligent connected vehicle as a high-risk vehicle, if the risk assessment coefficient is less than the risk assessment coefficient threshold, then do not mark the intelligent connected vehicle.

[0046] In a preferred embodiment, an accident analysis and intelligent intervention system based on intelligent connected vehicles includes a data acquisition module, a traffic accident assessment module, and a high-risk vehicle identification module, with the modules connected by signals.

[0047] The data acquisition module is used to collect similarity and probability information of traffic data when intelligent connected vehicles approach accident-prone areas, and to collect location and driving information of intelligent connected vehicles when driving in accident-prone areas, and to determine traffic accident assessment information when intelligent connected vehicles are driving in accident-prone areas.

[0048] The traffic accident assessment module is used to generate a traffic accident assessment model for intelligent connected vehicles in real time by comprehensively analyzing similar and probabilistic information from traffic data, generating traffic accident assessment coefficients, and assessing the accident risk of intelligent connected vehicles when entering accident-prone areas.

[0049] The high-risk vehicle identification module is used to identify high-risk vehicles based on the location information, driving information, and traffic accident assessment model of intelligent connected vehicles when they approach accident-prone areas.

[0050] The technical effects and advantages of this invention are as follows:

[0051] This invention uses data acquisition, Bayesian inference, t-SNE dimensionality reduction, driving behavior analysis, and high-risk vehicle identification to assess the risk of intelligent connected vehicles entering accident-prone areas, and classifies the danger of different intelligent connected vehicles in accident-prone areas. This invention helps improve road traffic safety and reduce the accident rate. Attached Figure Description

[0052] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0053] Figure 1 This is a flowchart illustrating an accident analysis and intelligent intervention method based on intelligent connected vehicles according to the present invention.

[0054] Figure 2 This is a schematic diagram of the structure of an accident analysis and intelligent intervention system based on intelligent connected vehicles according to the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] Figure 1 This is a flowchart illustrating an accident analysis and intelligent intervention method based on intelligent connected vehicles according to the present invention, which specifically includes the following steps:

[0058] S1: Based on the sensors and vehicle communication system equipped in intelligent connected vehicles, whenever an intelligent connected vehicle approaches an accident-prone area, important features are extracted from the historical traffic data of the accident-prone area by using the t-SNE dimensionality reduction method, and real-time traffic data of the intelligent connected vehicle is obtained to determine the similarity information of the traffic data of the intelligent connected vehicle.

[0059] S2: Based on historical traffic data and real-time traffic data of intelligent connected vehicles in areas with frequent accidents, Bayesian inference is performed. By analyzing statistical data and constructing machine learning methods, the probability information of intelligent connected vehicle traffic data is determined.

[0060] S3: By comprehensively analyzing similar and probabilistic information from traffic data, construct a traffic accident assessment model for intelligent connected vehicles approaching accident-prone areas to assess the risk of intelligent connected vehicles entering accident-prone areas;

[0061] S4: Based on the location and driving information of intelligent connected vehicles when driving in accident-prone areas, and by collecting traffic accident assessment models of intelligent connected vehicles when driving in accident-prone areas, determine the traffic accident assessment information of intelligent connected vehicles when driving in accident-prone areas, and identify intelligent connected vehicles that are high-risk vehicles in accident-prone areas.

[0062] Intelligent connected vehicles are equipped with sensors and in-vehicle communication systems, which can collect information such as road conditions, vehicle status, and driver behavior. The in-vehicle communication system can exchange data with other vehicles and traffic infrastructure. By analyzing the frequency of accidents involving intelligent connected vehicles in different areas, accident heat maps can be generated to identify accident-prone areas of intelligent connected vehicles.

[0063] Whenever an intelligent connected vehicle approaches an accident-prone area, the t-SNE dimensionality reduction method is used to extract important features from the historical traffic data of the accident-prone area, visualize the degree of danger of the intelligent connected vehicle in the accident-prone area, obtain real-time traffic data of the intelligent connected vehicle based on the important features that affect the accident, and compare it with the historical traffic data of the accident-prone area to provide an instant assessment for the intelligent connected vehicle. The similarity information of the traffic data is represented by the dimensionality reduction feature similarity coefficient.

[0064] It should be noted that accidents are usually the result of multiple factors, including road conditions, traffic flow, driver behavior, and weather conditions. Therefore, the t-SNE dimensionality reduction method can be used to screen out important factors that affect the occurrence of accidents, such as vehicle speed, traffic flow, and weather conditions.

[0065] The advantages of the dimensionality reduction feature similarity coefficient are:

[0066] The core function of the dimensionality reduction feature similarity coefficient is to quantify the similarity between real-time traffic data of current intelligent connected vehicles and historical traffic data of accident-prone areas. By mapping historical data and real-time data to a low-dimensional space (through t-SNE dimensionality reduction) and calculating their Euclidean distance or similarity coefficient in the low-dimensional space, the degree of similarity between real-time data points and historical data points can be clearly represented.

[0067] Dimensionality reduction feature similarity coefficients are used to reduce the dimensionality of important features in historical data (such as traffic flow, vehicle speed, weather, etc.), retaining the features that are most helpful in distinguishing different data points and reducing irrelevant or redundant features. After dimensionality reduction, the visualization and analysis of data become more intuitive and easier to process.

[0068] Dimensionality reduction feature similarity coefficients can discover complex nonlinear relationships between features, while traditional linear dimensionality reduction methods (such as PCA) may not be able to capture these important complex patterns. After performing nonlinear dimensionality reduction on the multidimensional features of an accident, the similarity between the current state of an intelligent connected vehicle and historical accidents can be assessed more accurately.

[0069] The logic for obtaining the dimensionality reduction feature similarity coefficient is as follows: collect historical traffic data of accident-prone areas, extract features from the historical traffic data of accident-prone areas, including time features, traffic flow features, weather features and driving features, fuse the time features, traffic flow features, weather features and driving features, and obtain the historical comprehensive traffic feature vector of each accident in the accident-prone area through standardization processing, and obtain the real-time comprehensive traffic feature vector with the same dimension as the comprehensive traffic feature vector based on real-time traffic data.

[0070] It should be noted that time characteristics include time period, date, seasonality, etc., traffic flow characteristics include traffic volume, lane usage, and speed fluctuations in accident-prone areas, weather characteristics represent weather conditions in accident-prone areas, and driving characteristics represent driving data of vehicles involved in accidents in accident-prone areas, determined by onboard sensors and onboard communication systems, including driving behavior and vehicle type, etc.

[0071] The historical comprehensive traffic feature vector of each accident is input into the t-SNE algorithm for dimensionality reduction. The embedding coordinates of the historical comprehensive traffic feature vector of each accident in the low-dimensional space are represented as: X = [x1, x2, x3, ..., x...]. n ]; where n is the number of accidents in accident-prone areas;

[0072] By projecting the real-time integrated traffic feature vector onto the embedded coordinates of the historical integrated traffic feature vector in the low-dimensional space, the low-dimensional projection of the real-time integrated traffic feature vector is represented as: R, where the mapping function is learned by training the t-SNE algorithm, and the projection of the real-time integrated traffic feature vector is calculated using the trained mapping function. The mapping function is adjusted to the low-dimensional projection coordinates by the gradient descent algorithm to minimize the KL divergence between the real-time integrated traffic feature vector and the historical integrated traffic feature vector.

[0073] It should be noted that the real-time integrated traffic feature vector represents historical traffic data of accident-prone areas, while the real-time integrated traffic feature vector represents real-time data. The mapping function ensures that during the dimensionality reduction process of the t-SNE algorithm, the projection of real-time data points in the low-dimensional space is as consistent as possible with that of historical data points.

[0074] After dimensionality reduction using the t-SNE algorithm, the distance between the real-time integrated traffic feature vector and the historical integrated traffic feature vector is calculated using Euclidean distance. The calculation formula is as follows: Where i = 1, 2, 3, ..., n, x i This is the historical comprehensive traffic feature vector at the time of the i-th accident;

[0075] The similarity coefficient of the dimensionality-reduced features is calculated using the Gaussian kernel function, and the formula is as follows: Among them, TZ xs σ is the similarity coefficient for the dimensionality reduction features, and σ is the bandwidth parameter.

[0076] As can be seen from the formula, the larger the similarity coefficient of the dimensionality reduction feature, the more similar the real-time traffic data of the current intelligent connected vehicle is to the historical traffic data of accident-prone areas. This means that if the current intelligent connected vehicle does not change its traffic data when entering accident-prone areas, it is more likely to cause traffic accidents.

[0077] Whenever an intelligent connected vehicle approaches an accident-prone area, Bayesian inference is performed based on historical traffic data and real-time traffic data of the intelligent connected vehicle in the accident-prone area. By analyzing statistical data and constructing machine learning methods, the probability of an accident occurring in the accident-prone area based on the current real-time traffic data of the intelligent connected vehicle is determined, and the probability information of the traffic data is represented by the posterior probability coefficient of the accident.

[0078] The advantage of the posterior probability coefficient is that:

[0079] The post-accident probability coefficient combines historical traffic data and real-time traffic data from intelligent connected vehicles. Maximum likelihood estimation optimizes the feature weights through a logistic regression model, which can train the probability of an accident based on the relationship between features and provide conditional probabilities. Bayesian inference combines prior and posterior probabilities, and can still provide smoothness and stability for the model even when data changes, samples are insufficient, or noise is high.

[0080] Post-accident probability coefficients can assess the hazard level of the current traffic environment in real time. Through feature analysis of real-time traffic data, this coefficient can reflect whether the current driving environment is similar to historical accident hotspots, helping intelligent connected vehicles identify potentially high-risk areas.

[0081] The logic for obtaining the post-accident probability coefficient is as follows: Based on historical traffic data of accident-prone areas and current real-time traffic data of intelligent connected vehicles, determine the characteristics and accident labels of each accident in the historical traffic data, and determine the characteristics of the real-time traffic data. The prior probability P(A) is determined by the ratio of the number of accidents in the historical data to the total number of samples.

[0082] By training a logistic regression model, the predicted probability of an accident is calculated based on different combinations of features. The expression for logistic regression is: Where A is the accident label, A=1 indicates an accident has occurred, A=0 indicates no accident, and w1, w2, ..., w n are the weights of different features, and b is the bias term;

[0083] By training the weights by maximizing the log-likelihood function, and using the trained logistic regression model, the likelihood function at the time of the accident is obtained based on the predicted value of the accident probability of the logistic regression model, that is, the conditional probability of the accident under different features. Based on the features of real-time traffic data, the conditional probability of the accident between the features of real-time traffic data and the accident is determined, and the conditional probability of the accident between the features of real-time traffic data and the accident is denoted as: P(R|A).

[0084] It should be noted that the logistic regression model optimizes the model parameters through maximum likelihood estimation and directly trains the conditional probability of the accident to occur, while Bayesian inference uses posterior probability for inference. Combining the maximum likelihood estimation of logistic regression with the prior information of Bayesian inference can improve the stability and robustness of the model, enabling the model to run stably even when the data changes or the sample is insufficient.

[0085] Based on the current real-time traffic data of intelligent connected vehicles, the characteristics of the real-time traffic data are determined, and the evidence P(R) is determined by the probability of the real-time traffic data under the condition that the accident has occurred and has not occurred.

[0086] The posterior probability coefficient of the accident is calculated using the following formula: Among them, SG hy This represents the post-accident probability coefficient.

[0087] As can be seen from the formula, the larger the posterior probability coefficient of an accident, the higher the possibility that the real-time traffic data of current intelligent connected vehicles is at risk. This means that if current intelligent connected vehicles do not change their traffic data when entering accident-prone areas, they are more likely to cause traffic accidents.

[0088] By comprehensively analyzing similarity and probability information from traffic data, the posterior probability coefficients of accidents are weighted and calculated. When an intelligent connected vehicle approaches an accident-prone area, a traffic accident assessment model for the intelligent connected vehicle is generated in real time, producing a traffic accident assessment coefficient. The formula for calculating the traffic accident assessment coefficient is: PG jt =α1TZ xs +α2SG hy Among them, PG jt Let α1 and α2 be the posterior probability coefficients of the accident and the proportional coefficients of the posterior probability coefficients of the accident, respectively, and both α1 and α2 are greater than 0.

[0089] It should be noted that the magnitude of the traffic accident assessment coefficient quantifies the probability of an accident. It does not mean that an accident will definitely happen, but rather that it indicates a higher likelihood of occurrence. An accident is itself a probabilistic event and is influenced by many random factors.

[0090] Set a threshold for the traffic accident assessment coefficient, and mark the traffic accident assessment coefficient threshold as: PG yz When a connected vehicle is detected heading towards the next accident-prone area, the system obtains the traffic accident assessment coefficient of the connected vehicle for that area in real time. The system then compares the traffic accident assessment coefficient of the connected vehicle with a traffic accident assessment coefficient threshold. If the traffic accident assessment coefficient is greater than the threshold, a warning signal is generated and intervention measures are provided to the driver via broadcast. If the traffic accident assessment coefficient is less than the threshold, no warning signal is generated.

[0091] It should be noted that the traffic accident assessment coefficient is dynamic, meaning that it changes with traffic data. When an intelligent connected vehicle approaches an accident-prone area, the real-time traffic data continuously updates the assessment coefficient. When the probability of an accident occurs exceeds a certain threshold, the traffic accident assessment coefficient of the intelligent connected vehicle is considered to be greater than the traffic accident assessment coefficient threshold.

[0092] As intelligent connected vehicles enter accident-prone areas, their location and driving information are collected within the monitoring range of these areas. The location information of intelligent connected vehicles within the monitoring range is represented by a relative distance coefficient, and the driving information is represented by a risk behavior divergence coefficient.

[0093] The advantage of the relative distance coefficient is that:

[0094] The relative distance coefficient, obtained through real-time data collection (such as distance information acquired by onboard sensors), reflects the real-time changes in the relative position and distance between a connected vehicle and its surroundings. The level of this coefficient directly reflects the safety of the connected vehicle's operation. For example, a large fluctuation in distance (i.e., a high coefficient of variation) may indicate high uncertainty in the current driving environment, potentially reflecting unstable driver behavior or poor road conditions, suggesting a possible collision risk. Conversely, a smaller coefficient of variation means that the distance between the vehicle and other vehicles remains relatively stable, indicating higher safety.

[0095] The logic for obtaining the relative distance coefficient is as follows: After the intelligent connected vehicle enters an accident-prone area, the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is collected within the monitoring range, and the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is marked as: JL m,j Where m is the number of the other vehicles in the accident-prone area, m = 1, 2, 3, ..., M, M is a positive integer, j is the sampling number in the monitoring interval, j = 1, 2, 3, ..., J, J is a positive integer;

[0096] It should be noted that the monitoring range is set by professionals in the field. The relative distance data between intelligent connected vehicles and other vehicles is acquired in real time through onboard sensors and uploaded to the cloud through the onboard communication system.

[0097] Calculate the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval, and label the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas as: JL avg and JL std ;in,

[0098] The coefficient of variation of the relative distance between the intelligent connected vehicle and other vehicles in accident-prone areas within the monitoring interval is calculated using the following formula: Among them, JL by The coefficient of variation of the relative distance between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval;

[0099] Calculate the relative relationship coefficient of vehicle distance. Among them, CJ xd This is the relative relationship coefficient between vehicle distances.

[0100] As the formula shows, a larger relative distance coefficient indicates a greater variation in the relative distance between the intelligent connected vehicle and other vehicles, and more frequent interactions between them. This may lead to a greater risk of collision, especially in areas with dense traffic or rapidly changing traffic flow.

[0101] The advantage of the risky behavior divergence coefficient is that by comparing the distribution differences between real-time driving behavior and historical risky driving behavior, the risky behavior divergence coefficient can quantify the degree of abnormality between current driving behavior and driving behavior at the time of past accidents. By analyzing driving behavior based on divergence values, it can more accurately identify real risky driving behavior, rather than just based on some fixed rules or experience.

[0102] Dangerous driving behavior is not a single act, but rather a combination of various driving behaviors, such as speed, acceleration, following distance, and sudden braking. These behaviors have varying degrees of impact on the probability of an accident. Kernel density estimation can be used to estimate the probability distribution of each dangerous driving behavior based on historical data. By calculating the difference between real-time and historical data, kernel density estimation can capture the contribution of each driving behavior to the risk.

[0103] The logic for obtaining the divergence coefficient of the dangerous behavior is as follows: Based on historical traffic data from accident-prone areas, the dangerous driving behaviors of intelligent connected vehicles are determined. Based on historical data, the distribution of historical dangerous driving behaviors and the distribution of real-time dangerous driving behaviors are constructed using the kernel density estimation method. The distribution of historical dangerous driving behaviors is labeled as: P(h f The distribution of real-time dangerous driving behavior is labeled as: Q(h) f ), where f = 1, 2, 3, ..., F, F is a positive integer, f is the number of different dangerous driving behaviors, h f These are random variables representing different dangerous driving behaviors;

[0104] It should be noted that dangerous driving behaviors include vehicle speed, acceleration, following distance, and sudden braking. Different dangerous driving behaviors have different impacts on the probability of traffic accidents, and are usually determined based on their actual contribution to the occurrence of the accident.

[0105] The formula for calculating the divergence coefficient of dangerous behavior is as follows: Among them, SD wx Y is the divergence coefficient of dangerous behavior. f Weights for different dangerous driving behaviors.

[0106] As can be seen from the formula, the larger the divergence coefficient of dangerous behavior, the greater the difference between the current driving behavior and the known high-risk behaviors in the past, indicating that there is safe driving behavior.

[0107] The logic for obtaining the traffic accident risk coefficient is as follows: Based on the real-time generation of a traffic accident assessment model for intelligent connected vehicles when they approach accident-prone areas, the traffic accident assessment model of the intelligent connected vehicle after entering the accident-prone area is obtained, resulting in traffic accident assessment coefficients at different times within the monitoring interval. These coefficients are then compared with a threshold value to obtain the traffic accident assessment coefficients within the monitoring interval that exceed the threshold. These traffic accident assessment coefficients are then marked as PG. k ;

[0108] It should be noted that before an intelligent connected vehicle approaches an accident-prone area, the traffic accident assessment model is used to quantitatively assess the probability of the intelligent connected vehicle occurring in the accident-prone area, so as to facilitate the implementation of preventive measures in advance. However, after the intelligent connected vehicle approaches the accident-prone area, the traffic accident assessment model shifts to measuring the current risk status of the intelligent connected vehicle in real time.

[0109] The formula for calculating the traffic accident risk coefficient is as follows: Among them, FX pg This represents the risk coefficient for traffic accidents.

[0110] As the formula shows, the higher the traffic accident risk coefficient, the higher the risk of intelligent connected vehicles in accident-prone areas, and the more likely they are to be dangerous. Other vehicles should be notified to keep a relative distance from such intelligent connected vehicles.

[0111] Based on the comprehensive analysis of the location and driving information of intelligent connected vehicles entering accident-prone areas, and combined with traffic accident assessment information during the operation of intelligent connected vehicles in accident-prone areas, a risk assessment model is constructed by weighting the relative distance coefficient, the risk behavior divergence coefficient, and the traffic accident risk coefficient. This generates a risk assessment coefficient for intelligent connected vehicles, calculated using the formula: PG fx =β1CL xd -β2SD wx +β3FX pg Among them, PG fx β1, β2, and β3 are the relative relationship coefficients of vehicle distance, the divergence value coefficient of dangerous behavior, and the proportional coefficients of risk assessment coefficients, respectively, and β1, β2, and β3 are all greater than 0.

[0112] Set a risk assessment coefficient threshold, obtain the risk assessment coefficient of all intelligent connected vehicles in accident-prone areas, compare the risk assessment coefficient of intelligent connected vehicles with the risk assessment coefficient threshold, if the risk assessment coefficient is greater than the risk assessment coefficient threshold, then the intelligent connected vehicle is marked as a high-risk vehicle, and other intelligent connected vehicles are notified to avoid potential collisions with high-position vehicles. If the risk assessment coefficient is less than the risk assessment coefficient threshold, then the intelligent connected vehicle is not marked.

[0113] Example 2

[0114] Figure 2 This is a schematic diagram of the structure of an accident analysis and intelligent intervention system based on intelligent connected vehicles according to the present invention, including a data acquisition module, a traffic accident assessment module, and a high-risk vehicle identification module, with signal connections between the modules;

[0115] The data acquisition module is used to collect similarity and probability information of traffic data when intelligent connected vehicles approach accident-prone areas, and to collect location and driving information of intelligent connected vehicles when driving in accident-prone areas, and to determine traffic accident assessment information when intelligent connected vehicles are driving in accident-prone areas.

[0116] The traffic accident assessment module is used to generate a traffic accident assessment model for intelligent connected vehicles in real time by comprehensively analyzing similar and probabilistic information from traffic data, generating traffic accident assessment coefficients, and assessing the accident risk of intelligent connected vehicles when entering accident-prone areas.

[0117] The high-risk vehicle identification module is used to identify high-risk vehicles after a connected vehicle approaches an accident-prone area, based on the connected vehicle's location information, driving information, and traffic accident assessment model.

[0118] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0119] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0120] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0121] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0123] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0124] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0125] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for accident analysis and intelligent intervention based on intelligent connected vehicles, characterized in that, Includes the following steps: S1: Based on the sensors and vehicle communication system equipped in intelligent connected vehicles, whenever an intelligent connected vehicle approaches an accident-prone area, important features are extracted from the historical traffic data of the accident-prone area by using the t-SNE dimensionality reduction method, and real-time traffic data of the intelligent connected vehicle is obtained to determine the similarity information of the traffic data of the intelligent connected vehicle. S2: Based on historical traffic data and real-time traffic data of intelligent connected vehicles in areas with frequent accidents, Bayesian inference is performed. By analyzing statistical data and constructing machine learning methods, the probability information of intelligent connected vehicle traffic data is determined. S3: By comprehensively analyzing similar and probabilistic information from traffic data, construct a traffic accident assessment model for intelligent connected vehicles approaching accident-prone areas to assess the risk of intelligent connected vehicles entering accident-prone areas; S4: Based on the location and driving information of intelligent connected vehicles when driving in accident-prone areas, and by collecting traffic accident assessment models of intelligent connected vehicles when driving in accident-prone areas, determine the traffic accident assessment information of intelligent connected vehicles when driving in accident-prone areas, and identify intelligent connected vehicles that are high-risk vehicles in accident-prone areas.

2. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 1, characterized in that, Similar information regarding traffic data from intelligent connected vehicles includes: The similarity information of traffic data is represented by the dimensionality-reduced feature similarity coefficient; The logic for obtaining the dimensionality reduction feature similarity coefficient is as follows: collect historical traffic data of accident-prone areas, extract features from the historical traffic data of accident-prone areas, including time features, traffic flow features, weather features and driving features, fuse the time features, traffic flow features, weather features and driving features, and obtain the historical comprehensive traffic feature vector of each accident in the accident-prone area through standardization processing, and obtain the real-time comprehensive traffic feature vector with the same dimension as the comprehensive traffic feature vector based on real-time traffic data. The historical comprehensive traffic feature vector of each accident is input into the t-SNE algorithm for dimensionality reduction. The embedding coordinates of the historical comprehensive traffic feature vector of each accident in the low-dimensional space are represented as: X = [x1, x2, x3, ..., x...]. n ]; where n is the number of accidents in accident-prone areas; By projecting the real-time integrated traffic feature vector onto the embedded coordinates of the historical integrated traffic feature vector in the low-dimensional space, the low-dimensional projection of the real-time integrated traffic feature vector is represented as: R, where the mapping function is learned by training the t-SNE algorithm, and the projection of the real-time integrated traffic feature vector is calculated using the trained mapping function. The mapping function is adjusted to the low-dimensional projection coordinates by the gradient descent algorithm to minimize the KL divergence between the real-time integrated traffic feature vector and the historical integrated traffic feature vector. After dimensionality reduction using the t-SNE algorithm, the distance between the real-time integrated traffic feature vector and the historical integrated traffic feature vector is calculated using Euclidean distance. The calculation formula is as follows: Where i = 1, 2, 3, ..., n, x i This is the historical comprehensive traffic feature vector at the time of the i-th accident; The similarity coefficient of the dimensionality-reduced features is calculated using the Gaussian kernel function, and the formula is as follows: Among them, TZ xs σ is the similarity coefficient for the dimensionality reduction features, and σ is the bandwidth parameter.

3. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 2, characterized in that, Probabilistic information from traffic data for intelligent connected vehicles includes: The probabilistic information of traffic data is represented by post-accident probability coefficients; The logic for obtaining the post-accident probability coefficient is as follows: Based on historical traffic data of accident-prone areas and current real-time traffic data of intelligent connected vehicles, determine the characteristics and accident labels of each accident in the historical traffic data, and determine the characteristics of the real-time traffic data. The prior probability P(A) is determined by the ratio of the number of accidents in the historical data to the total number of samples. By training a logistic regression model, the predicted probability of an accident is calculated based on different combinations of features. The expression for logistic regression is: Where A is the accident label, A=1 indicates an accident has occurred, A=0 indicates no accident, and w1, w2, ..., w n are the weights of different features, and b is the bias term; By training the weights by maximizing the log-likelihood function, and using the trained logistic regression model, the likelihood function at the time of the accident is obtained based on the predicted value of the accident probability of the logistic regression model, that is, the conditional probability of the accident under different features. Based on the features of real-time traffic data, the conditional probability of the accident between the features of real-time traffic data and the accident is determined, and the conditional probability of the accident between the features of real-time traffic data and the accident is denoted as: P(R|A). Based on the current real-time traffic data of intelligent connected vehicles, the characteristics of the real-time traffic data are determined, and the evidence P(R) is determined by the probability of the real-time traffic data under the condition that the accident has occurred and has not occurred. The posterior probability coefficient of the accident is calculated using the following formula: Among them, SG hy This represents the post-accident probability coefficient.

4. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 3, characterized in that, Construct a traffic accident assessment model for intelligent connected vehicles approaching accident-prone areas to assess the risks associated with such vehicles entering such areas, including: The posterior probability coefficients of the accident are weighted and calculated. When the intelligent connected vehicle approaches an accident-prone area, a traffic accident assessment model for the intelligent connected vehicle is generated in real time, generating a traffic accident assessment coefficient. The formula for calculating the traffic accident assessment coefficient is: PG jt =α1TZ xs +α2SG hy Among them, PG jt Let α1 and α2 be the posterior probability coefficients of the accident and the proportional coefficients of the posterior probability coefficients of the accident, respectively, and both α1 and α2 are greater than 0. Set a threshold for the traffic accident assessment coefficient, and mark the traffic accident assessment coefficient threshold as: PG yz When it is detected that the intelligent connected vehicle is heading to the next accident-prone area, the traffic accident assessment coefficient of the intelligent connected vehicle for the next accident-prone area is obtained in real time, and the traffic accident assessment coefficient of the intelligent connected vehicle is compared with the traffic accident assessment coefficient threshold. If the traffic accident assessment coefficient is greater than the traffic accident assessment coefficient threshold, a warning signal will be generated; if the traffic accident assessment coefficient is less than the traffic accident assessment coefficient threshold, no warning signal will be generated.

5. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 4, characterized in that, Location information of intelligent connected vehicles when driving in accident-prone areas includes: The location information of intelligent connected vehicles when driving in accident-prone areas is represented by the relative distance coefficient. The logic for obtaining the relative distance coefficient is as follows: After the intelligent connected vehicle enters an accident-prone area, the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is collected within the monitoring interval, and the relative distance between the intelligent connected vehicle and other vehicles in the accident-prone area is marked as: JL m,j Where m is the number of the other vehicles in the accident-prone area, m = 1, 2, 3, ..., M, M is a positive integer, j is the sampling number in the monitoring interval, j = 1, 2, 3, ..., J, J is a positive integer; Calculate the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval, and label the mean and standard deviation of the relative distances between intelligent connected vehicles and other vehicles in accident-prone areas as: JL avg and JL std ;in, The coefficient of variation of the relative distance between the intelligent connected vehicle and other vehicles in accident-prone areas within the monitoring interval is calculated using the following formula: Among them, JL by The coefficient of variation of the relative distance between intelligent connected vehicles and other vehicles in accident-prone areas within the monitoring interval; Calculate the relative relationship coefficient of vehicle distance. Among them, CJ xd This is the relative relationship coefficient between vehicle distances.

6. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 5, characterized in that, Driving information of intelligent connected vehicles when driving in accident-prone areas includes: The driving information of intelligent connected vehicles when driving in accident-prone areas is represented by the risk behavior divergence coefficient. The logic for obtaining the divergence coefficient of the dangerous behavior is as follows: Based on historical traffic data from accident-prone areas, the dangerous driving behaviors of intelligent connected vehicles are determined. Based on historical data, the distribution of historical dangerous driving behaviors and the distribution of real-time dangerous driving behaviors are constructed using the kernel density estimation method. The distribution of historical dangerous driving behaviors is labeled as: P(h f The distribution of real-time dangerous driving behavior is labeled as: Q(h) f ), where f = 1, 2, 3, ..., F, F is a positive integer, f is the number of different dangerous driving behaviors, h f These are random variables representing different dangerous driving behaviors; The formula for calculating the divergence coefficient of dangerous behavior is as follows: Among them, SD wx Y is the divergence coefficient of dangerous behavior. f Weights for different dangerous driving behaviors.

7. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 6, characterized in that, Determine traffic accident assessment information when intelligent connected vehicles are driving in accident-prone areas, including: The traffic accident assessment information of intelligent connected vehicles driving in accident-prone areas will be represented by a traffic accident risk coefficient. The logic for obtaining the traffic accident risk coefficient is as follows: Based on the real-time generation of a traffic accident assessment model for intelligent connected vehicles when they approach accident-prone areas, the traffic accident assessment model of the intelligent connected vehicle after entering the accident-prone area is obtained, resulting in traffic accident assessment coefficients at different times within the monitoring interval. These coefficients are then compared with a threshold value to obtain the traffic accident assessment coefficients within the monitoring interval that exceed the threshold. These traffic accident assessment coefficients are then marked as PG. k ; The formula for calculating the traffic accident risk coefficient is as follows: Among them, FX pg This represents the risk coefficient for traffic accidents.

8. The accident analysis and intelligent intervention method based on intelligent connected vehicles according to claim 7, characterized in that, Intelligent connected vehicles identified as high-risk vehicles in accident-prone areas include: Based on the comprehensive analysis of the location and driving information of intelligent connected vehicles entering accident-prone areas, and combined with traffic accident assessment information during the operation of intelligent connected vehicles in accident-prone areas, a risk assessment model is constructed by weighting the relative distance coefficient, the risk behavior divergence coefficient, and the traffic accident risk coefficient. This generates a risk assessment coefficient for intelligent connected vehicles, calculated using the formula: PG fx =β1CL xd -β2SD wx +β3FX pg Among them, PG fx β1, β2, and β3 are the relative relationship coefficients of vehicle distance, the divergence value coefficient of dangerous behavior, and the proportional coefficients of risk assessment coefficients, respectively, and β1, β2, and β3 are all greater than 0. Set a risk assessment coefficient threshold, obtain the risk assessment coefficient of all intelligent connected vehicles in accident-prone areas, compare the risk assessment coefficient of each intelligent connected vehicle with the risk assessment coefficient threshold, if the risk assessment coefficient is greater than the risk assessment coefficient threshold, then mark the intelligent connected vehicle as a high-risk vehicle, if the risk assessment coefficient is less than the risk assessment coefficient threshold, then do not mark the intelligent connected vehicle.

9. An accident analysis and intelligent intervention system based on intelligent connected vehicles, used to implement the accident analysis and intelligent intervention method based on intelligent connected vehicles as described in any one of claims 1-8, characterized in that, It includes a data acquisition module, a traffic accident assessment module, and a high-risk vehicle identification module, with signal connections between the modules; The data acquisition module is used to collect similarity and probability information of traffic data when intelligent connected vehicles approach accident-prone areas, and to collect location and driving information of intelligent connected vehicles when driving in accident-prone areas, and to determine traffic accident assessment information when intelligent connected vehicles are driving in accident-prone areas. The traffic accident assessment module is used to generate a traffic accident assessment model for intelligent connected vehicles in real time by comprehensively analyzing similar and probabilistic information from traffic data, generating traffic accident assessment coefficients, and assessing the accident risk of intelligent connected vehicles when entering accident-prone areas. The high-risk vehicle identification module is used to identify high-risk vehicles based on the location information, driving information, and traffic accident assessment model of intelligent connected vehicles when they approach accident-prone areas.

Images