Method for automatic driving accident classification and severity prediction

Abstract

The application discloses a kind of automatic driving accident classification and severity prediction method, including the following steps: obtaining automatic driving accident in past years, establishes automatic driving accident mining and analysis dataset;Find out the main influencing factors leading to automatic driving accident through significance test, label the cluster class of each accident data;Establish neural network model and utilize convolutional neural network to predict the severity of automatic driving accident;Automatic driving accident dataset is input to neural network, and the predicted accident classification label is output, the matching rate with accident severity is calculated, and the automatic driving accident occurrence mode is analyzed.The application can extract the main features affecting the occurrence rate and severity of automatic driving collision accident, and the neural network model can accurately predict the severity of mass accident data, which helps traffic agencies to develop targeted engineering countermeasures for different accident patterns and improve the safety level of automatic driving vehicles in practical application.

Description

Technical Field

[0001] This invention belongs to the field of accident causation analysis technology, and more specifically, relates to a method for classifying and predicting the severity of autonomous driving accidents. Background Technology

[0002] With the development of intelligent vehicles and connected technologies, various levels of autonomous vehicles have gradually revealed numerous problems in practical applications. Besides factors such as roads, environment, and vehicles themselves affecting the safety of autonomous driving, the vehicle's own movement and interactions with surrounding vehicles also influence accident consequences. Utilizing advanced technologies and methods to analyze traffic accidents involving autonomous vehicles has become an urgent issue. Many countries and regions have established autonomous driving accident databases, containing details such as vehicle manufacturers, accident locations, and time and environment. In recent years, the rapid development of artificial intelligence, including machine learning and deep learning, has also provided efficient methods for analyzing autonomous driving accident data. Currently used prediction techniques include XGBoost, CATBoost, and SVM, but they still face challenges in learning from massive amounts of data in practical applications. Summary of the Invention

[0003] This invention utilizes convolutional neural networks to classify autonomous driving accidents and predict their occurrence patterns.

[0004] To address at least one of the aforementioned technical problems, according to one aspect of the present invention, a method for classifying and predicting the severity of autonomous driving accidents is provided, comprising the following steps:

[0005] S1. Obtain historical accident data for autonomous driving, perform joint preprocessing and feature engineering encoding on the multi-source data, and establish an autonomous driving accident mining and analysis dataset D. Collect relevant data for autonomous driving accidents, including the original accident report dataset D1 and the satellite map location dataset D2. Combine and preprocess the multi-source data D1 and D2 using accident locations, imput missing values ​​using the mode, delete dirty data, and encode the text into an unordered array of categorical variables to establish the accident mining and analysis dataset D.

[0006] S2. Identify the main influencing factors leading to autonomous driving accidents through significance testing, and select accident features with strong correlation to severity using Pearson correlation coefficient. The chi-square test formula (Equation 1) is used to test the significance of unordered categorical variables in the accident dataset, initially screening influencing factors. Equation 2 is used to calculate the Pearson correlation coefficient, eliminating variables with low correlation to accident severity, and the dataset is then divided into training and test sets.

[0007]

[0008] Among them, Ai E represents the frequency of feature i. i Let be the expected frequency of feature i, and n be the total frequency. Using χ² 2 Use the value to determine the significance of a variable and delete insignificant variables.

[0009]

[0010] S3. Combine principal component analysis and K-means clustering to cluster the feature set of autonomous driving accidents and label the cluster of each accident data. First, principal component analysis is used to reduce the dimensionality of the original data and find k vectors to project the original data to minimize the projection error (Equation 3). Then, the principal components are used as feature factors for clustering. The Euclidean distance between each sample and each cluster center is calculated, and each sample is assigned to the nearest cluster center until the sum of squared errors within the cluster (SSE) converges to a local minimum as the termination condition (Equation 4).

[0011]

[0012] Where M is the number of features, and the numerator represents the sum of the distances between the original point and the projected point. The smaller the error, the better.

[0013]

[0014] Where K is the number of cluster centers, C i It is the i-th cluster center; μ i It is cluster center C i Mean vector.

[0015] S4. Establish a neural network model to predict autonomous driving accident patterns using a convolutional neural network (CNN). Obtain the accuracy between the predicted values ​​of the predicted features and the cluster analysis labels to measure the network's predictive performance. The CNN uses one-dimensional convolution and one-dimensional pooling layers. One-dimensional convolution involves sliding a window along the length or width direction and multiplying and summing the results. The activation function is ReLU. The convolution operation process is shown in Equation 5:

[0016]

[0017] Among them, y l It is the output of the data after passing through the l-th convolutional layer, and g() is the activation function set in the neural network model. It is the data input of the m-th array convolution region in the l-th layer. These are the weights of the m-th convolutional region in the l-th layer. This represents the bias of the l-th layer, and * represents the convolution operation.

[0018] Pooling operations do not require parameter training. The pooling type, kernel size, and stride are specified, and the operation is as shown in Equation 6:

[0019]

[0020] in, It is the pooling output of the m-th array in the l-th layer. It is the p-th value of the m-th array in the l-th layer, and h() is the pooling function set in the neural network model.

[0021] S5. Input the autonomous driving accident test dataset into the neural network, output the predicted accident pattern label, calculate the matching rate with the severity of the accident, analyze the occurrence pattern of autonomous driving accidents, and judge the severity of the accident based on the relevant traffic information collected in real time and take corresponding control measures to improve the safety of autonomous driving.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for classifying and predicting the severity of autonomous driving accidents according to the present invention.

[0023] According to another aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method for classifying and predicting the severity of autonomous driving accidents according to the present invention.

[0024] Compared with the prior art, the present invention has at least the following beneficial effects:

[0025] This invention establishes a high-dimensional accident feature dataset by collecting autonomous driving accident data. After screening variables using methods such as significance testing and correlation calculation, principal component analysis and K-means clustering are introduced to cluster the features of the autonomous driving accident data, forming accident occurrence pattern labels corresponding to different severity levels. A convolutional neural network is used to predict the severity based on accident features, and the model's prediction accuracy is evaluated by comparing it with the clustered labels. The matching rate with accident severity is calculated, thereby achieving autonomous driving accident occurrence pattern classification and severity prediction, enabling targeted safety measures to improve autonomous driving safety. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments will be briefly described below. Obviously, the drawings described below only relate to some embodiments of the present invention and are not intended to limit the present invention.

[0027] Figure 1 This is a flowchart of the method for classifying and predicting the severity of autonomous driving accidents according to the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention.

[0029] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0030] Example 1:

[0031] A method for classifying and predicting the severity of autonomous driving accidents based on convolutional neural networks, such as... Figure 1 As shown, it includes the following steps:

[0032] Step 1: Data collection related to autonomous driving accidents, including raw accident report data and satellite map location data. Search the publicly available connected autonomous driving accident dataset D1 from the California Department of Motor Vehicles (DMV) online. Collect data on direct factors such as vehicle movement behavior, vehicle type, accident type, and accident time from 479 detailed accident reports from January 2015 to September 2022. Use Google Maps to match latitude and longitude at accident locations to collect data on road type, geometric design factors, and traffic facilities (D2). Combine and preprocess the multi-source data D1 and D2 using accident IDs to create an accident mining and analysis dataset D. Merge the raw dataset and location dataset row-wise according to the accident location index to form an accident set containing multiple factors such as roads, vehicles, and environment. Preprocess the accident dataset by filling missing values ​​with the mode, deleting dirty data, and encoding attribute values ​​under different variables into unordered categorical variables (Table 1).

[0033] Table 1 Accident Mining and Analysis Dataset D

[0034]

[0035]

[0036] Step 2: Identify the main influencing factors leading to autonomous driving accidents through significance testing, and select accident features with strong correlation to severity using Pearson correlation coefficient. The chi-square test formula (Equation 1) is used to conduct a significance test on the unordered categorical variables in the accident dataset to preliminarily screen influencing factors. Equation 2 is used to calculate the Pearson correlation coefficient, eliminating variables with low correlation to accident severity, and the dataset is divided into training and test sets at a ratio of 0.7.

[0037]

[0038] Among them, A i E represents the frequency of feature i. i Let be the expected frequency of feature i, and n be the total frequency. Using χ² 2 Use the value to determine the significance of a variable and delete insignificant variables.

[0039]

[0040] The main influencing factors on the severity of autonomous driving accidents were screened out through significance tests and Pearson correlation calculations, as shown in Table 2.

[0041] Table 2 Accident Training Dataset

[0042] Num Vec_type AV_Move Time … Weather Severity Crash_type … Bus_sta … Road_park 1 1 1 1 … 1 1 1 … 1 … 1 2 2 2 2 … 2 2 2 … 1 … 1 3 4 5 4 … 3 3 4 … 0 … 0 … … … … … … … … … … … … 379 2 7 1 … 5 2 5 … 0 … 1 380 3 9 3 … 4 4 6 … 0 … 0

[0043] S3. Combine principal component analysis and K-means clustering to cluster the feature set of autonomous driving accidents and label the cluster of each accident data. First, principal component analysis is used to reduce the dimensionality of the original data and find k vectors to project the original data to minimize the projection error (Equation 3). Then, the principal components are used as feature factors for clustering. The Euclidean distance between each sample and each cluster center is calculated, and each sample is assigned to the nearest cluster center until the sum of squared errors within the cluster (SSE) converges to a local minimum as the termination condition (Equation 4).

[0044]

[0045] Where M is the number of features, and the numerator represents the sum of the distances between the original point and the projected point. The smaller the error, the better.

[0046]

[0047] Where K is the number of cluster centers, C i It is the i-th cluster center; μ i It is cluster center C i Mean vector.

[0048] Step 4: Establish a neural network model. Use a convolutional neural network (CNN) to predict autonomous driving accident patterns, obtain the accuracy between the predicted values ​​of the predicted features and the cluster analysis labels, and measure the network's predictive performance. The CNN uses one-dimensional convolution and one-dimensional pooling layers. One-dimensional convolution involves sliding windows in the length or width direction and multiplying and summing the results, with ReLU as the activation function. After continuous experimentation, the best prediction results were achieved when the convolutional layer filter size was set to 128, the convolutional kernel size to 3*3, the pooling layer size to 1, the stride to 1, the learning rate to 1e-2, and the batch data processing size to 20.

[0049] The convolution operation process is as shown in Equation 5:

[0050]

[0051] Among them, y l It is the output of the data after passing through the l-th convolutional layer, and g() is the activation function set in the neural network model. It is the data input of the m-th array convolution region in the l-th layer. These are the weights of the m-th convolutional region in the l-th layer. This represents the bias of the l-th layer, and * represents the convolution operation.

[0052] Pooling operations do not require parameter training. The pooling type, kernel size, and stride are specified, and the operation is as shown in Equation 6:

[0053]

[0054] in, It is the pooling output of the m-th array in the l-th layer. It is the p-th value of the m-th array in the l-th layer, and h() is the pooling function set in the neural network model.

[0055] Step 5: Input the autonomous driving accident test dataset into the neural network, output the predicted accident pattern labels, calculate the matching rate with the accident severity, and analyze the autonomous driving accident occurrence patterns. This allows for the assessment of accident severity based on real-time collected traffic information and the implementation of corresponding control measures, thereby improving autonomous driving safety. Table 3 shows a comparison between the clustering labels and the autonomous driving accident occurrence pattern prediction results. The matching rate between the clustering labels and accident severity is 80%, indicating that the best clustering obtained by the PCA+K-Means clustering method is based on accident severity. The CNN-based accident occurrence pattern prediction accuracy is 0.72, far superior to the 0.32 accuracy obtained by the Support Vector Machine (SVM), and the matching rate with the accident severity labels is as high as 86%, demonstrating that the autonomous driving accident occurrence pattern classification method based on convolutional neural networks can more efficiently and accurately predict accident consequences.

[0056] Table 3 Comparison of Accident Patterns Predicted by Neural Networks for Autonomous Driving and Clustering Labels

[0057] Severity Vec_type HDV_move CAV_move Crash_type Location PCA+KMeans CNN 1 1 1 2 1 2 1 1 2 2 1 2 2 2 1 1 1 1 2 1 2 3 2 1 3 2 5 10 1 4 3 3 … … … … … … … … 4 1 3 4 1 2 4 4

[0058] Example 2:

[0059] The computer-readable storage medium of this embodiment stores a computer program that, when executed by a processor, implements the steps in the method for classifying and predicting the severity of autonomous driving accidents in Embodiment 1.

[0060] The computer-readable storage medium in this embodiment can be an internal storage unit of the terminal, such as the terminal's hard disk or memory; the computer-readable storage medium in this embodiment can also be an external storage device of the terminal, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc. equipped on the terminal; furthermore, the computer-readable storage medium can include both the terminal's internal storage unit and external storage devices.

[0061] The computer-readable storage medium of this embodiment is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0062] Example 3:

[0063] The computer device of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the method for classifying and predicting the severity of autonomous driving accidents in Embodiment 1.

[0064] In this embodiment, the processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The memory can include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0065] Those skilled in the art will understand that the content disclosed in the embodiments can be provided as a method, system, or computer program product. Therefore, this solution can take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this solution can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage) containing computer-usable program code.

[0066] This solution is described with reference to flowchart illustrations and / or block diagrams of methods and computer program products according to embodiments of this solution. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0067] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0068] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0069] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0070] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

Claims

1. A method for classifying and predicting the severity of autonomous driving accidents, characterized in that, Includes the following steps: S1. Obtain historical autonomous driving accidents, perform joint preprocessing and feature engineering coding on multi-source data, and establish an autonomous driving accident mining and analysis dataset. The multi-source data includes the original autonomous driving accident report dataset D1 and the satellite map location dataset D2. S2. Identify the main influencing factors leading to autonomous driving accidents through significance testing, and select accident features with strong correlation to severity through Pearson correlation coefficient. The main influencing factors, after being screened by Pearson correlation coefficient to select features with strong correlation to accident severity, are used for subsequent cluster analysis of accident feature set and training of neural network model. S3. Combine principal component analysis and K-means clustering to cluster the feature set of autonomous driving accidents and label the cluster of each accident data; S4. Establish a neural network model to predict autonomous driving accident patterns using a convolutional neural network, obtain the accuracy between the predicted values ​​of the predicted features and the cluster analysis labels, and measure the prediction performance of the network. S5. Input the autonomous driving accident test dataset into the neural network, output the predicted accident pattern label, calculate the matching rate with the severity of the accident, analyze the occurrence pattern of autonomous driving accidents, judge the severity of the accident based on the relevant traffic information collected in real time, and take corresponding control measures.

2. The method according to claim 1, characterized in that, Step S1 is as follows: Collect data related to autonomous driving accidents, including the original accident report dataset D1 and the satellite map location dataset D2; combine and preprocess the multi-source data D1 and D2 using the accident locations, fill missing values ​​with the mode, delete dirty data, encode the text into an unordered array of categorical variables, and establish the accident mining and analysis dataset D.

3. The method according to claim 1, characterized in that, Step S2 is as follows: Use chi-square test formula 1: （1） Significance tests were performed on the unordered categorical variables in the accident dataset to preliminarily screen the factors influencing the accidents. Pearson correlation coefficient was calculated using Formula 2, variables with low correlation to accident severity were removed, and the dataset was divided into training and test sets. （2） Among them, for Frequency of features Features Expected frequency The total frequency; using Use the value to determine the significance of a variable and delete insignificant variables.

4. The method according to claim 1, characterized in that, Step S3 specifically involves: first, using principal component analysis to reduce the dimensionality of the original data, finding k vectors to project onto the original data to minimize the projection error, as shown in Formula 3; then, using the principal components as feature factors for clustering, calculating the Euclidean distance between each sample and each cluster center, and assigning each sample to the nearest cluster center, until the sum of squared errors within the cluster (SSE) converges to a local minimum as the termination condition, as shown in Formula 4. （3） Where m is the number of features, and the numerator represents the sum of the distances between the original point and the projected point; the smaller the error, the better. （4） Where K is the number of cluster centers, It is the center of the i-th cluster; Cluster center Mean vector.

5. The method according to claim 1, characterized in that, Step S4 specifically involves: The CNN convolutional neural network uses one-dimensional convolution and one-dimensional pooling layers. One-dimensional convolution involves sliding a window along the length or width direction and multiplying and summing the results. The activation function is ReLU. The convolution operation process is shown in Formula 5. （5） in, It is the output of the data after passing through the l-th convolutional layer, and g() is the activation function set in the neural network model. It is the data input of the m-th array convolution region in the l-th layer. These are the weights of the m-th convolutional region in the l-th layer. This is the bias of the l-th layer, and * represents the convolution operation; Pooling operations do not require parameter training. The pooling type, kernel size, and stride are specified, and the operation is as shown in Formula 6: （6） in, It is the pooling output of the m-th array in the l-th layer. It is the p-th value of the m-th array in the l-th layer, and h() is the pooling function set in the neural network model.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, it implements the steps in the method for classifying and predicting the severity of autonomous driving accidents as described in any one of claims 1 to 5.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the method for classifying and predicting the severity of autonomous driving accidents as described in any one of claims 1 to 5.

Images