CN116631186A - Expressway traffic accident risk assessment method and system based on dangerous driving event data - Google Patents
Expressway traffic accident risk assessment method and system based on dangerous driving event data Download PDFInfo
- Publication number
- CN116631186A CN116631186A CN202310569137.8A CN202310569137A CN116631186A CN 116631186 A CN116631186 A CN 116631186A CN 202310569137 A CN202310569137 A CN 202310569137A CN 116631186 A CN116631186 A CN 116631186A
- Authority
- CN
- China
- Prior art keywords
- dangerous driving
- accident
- data
- driving event
- traffic accident
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 206010039203 Road traffic accident Diseases 0.000 title claims abstract description 111
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000012502 risk assessment Methods 0.000 title claims abstract description 43
- 230000001133 acceleration Effects 0.000 claims abstract description 52
- 238000013136 deep learning model Methods 0.000 claims abstract description 42
- 238000011144 upstream manufacturing Methods 0.000 claims abstract description 31
- 230000001419 dependent effect Effects 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000012545 processing Methods 0.000 claims abstract description 12
- 230000008859 change Effects 0.000 claims description 28
- 238000011176 pooling Methods 0.000 claims description 22
- 238000011160 research Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 16
- 210000002569 neuron Anatomy 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 12
- 230000000694 effects Effects 0.000 claims description 11
- 230000035945 sensitivity Effects 0.000 claims description 10
- 238000011156 evaluation Methods 0.000 claims description 7
- 230000006399 behavior Effects 0.000 claims description 5
- 238000002790 cross-validation Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 238000013527 convolutional neural network Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
- G08G1/054—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed photographing overspeeding vehicles
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention discloses a highway traffic accident risk assessment method and system based on dangerous driving event data, wherein the method comprises the following steps: acquiring expressway traffic accident data and four types of dangerous driving event data; searching and counting the frequency of dangerous driving events on road sections before accidents and when no accidents occur and the average value of the maximum acceleration and the maximum speed of vehicles in all the same type of dangerous driving events respectively; constructing a sample data set by taking the dangerous driving event information as an independent variable and taking whether an accident is a dependent variable; processing the data into an image form, constructing and training a deep learning model, and taking into consideration the internal relation of dangerous driving event data in different event types and upstream and downstream space positions, so as to mine the rule of dangerous driving events on a road section before an accident occurs; and finally, acquiring dangerous driving event data of the whole highway section in real time, and evaluating the traffic accident risk in real time according to the trained model.
Description
Technical Field
The invention provides a highway traffic accident risk assessment method and system based on dangerous driving event data, and belongs to the technical field of road traffic safety.
Background
Traffic safety issues on highways have been a major concern for governments. In recent years, real-time evaluation of highway traffic accident risk becomes a research hotspot, and has extremely important significance for developing active traffic safety management and reducing traffic accident occurrence rate.
The existing expressway traffic accident risk assessment method generally uses traffic flow data collected by fixed detectors (such as ETC (electronic toll collection) portal frames and ground annular coils) to judge accident risk, but the data are limited by the arrangement positions and the arrangement densities of the fixed detectors, so that traffic running states at all positions of a whole road section are difficult to dynamically reflect, and the accident risk assessment accuracy is affected. With the progress of data acquisition technology, dangerous driving behavior data of a vehicle are easier to acquire, and a series of dangerous driving behavior monitoring, risk assessment and early warning methods are presented. However, these methods are all methods for risk assessment by using a single individual vehicle as a study object, and lack a method for traffic accident risk assessment based on dangerous driving event data of vehicles on road segments by using road segment accident risks as study objects from the perspective of road traffic managers.
In the aspect of accident risk assessment models, most of the existing methods adopt statistical analysis or machine learning models to mine characteristic rules of factors such as road traffic flow before accidents occur, and some of the existing methods also adopt a Recurrent Neural Network (RNN) deep learning model to consider the time correlation of traffic flow data for accident risk modeling. However, in all the methods, the one-dimensional vector form data is used for representing factors such as road traffic flow before an accident occurs, and the internal connection of the data in multiple dimensions such as time, space and influence factors is difficult to consider. The Convolutional Neural Network (CNN) deep learning model can efficiently extract local modes in image data, comprehensively consider the correlation of the data in multiple dimensions to perform model training and prediction, and has much smaller calculation cost ratio RNN.
Because dangerous driving event data is acquired through vehicle navigation software, the acquisition range can cover all positions of the whole highway section of the highway and is not influenced by weather conditions. Therefore, dangerous driving event data of vehicles on the road are used and processed into image forms, the image forms are input into the CNN deep learning model, multidimensional correlation contained in the data can be comprehensively considered, and real-time refined assessment of traffic accident risks of all road sections of the expressway can be realized under the condition that new equipment is not required to be installed.
Disclosure of Invention
The invention aims to solve the technical problems that: aiming at the defects of the prior art, the invention uses dangerous driving event data of vehicles on the expressway, processes the dangerous driving event data into an image form according to different dangerous driving event types and spatial position relations, utilizes a deep learning model comprising a Convolutional Neural Network (CNN) and other modules to mine dangerous driving event rules on a road section before traffic accidents happen, and provides an expressway traffic accident risk assessment method and system based on the dangerous driving event data, which aims at solving the problems that the existing expressway traffic accident risk assessment method is difficult to accurately assess accident risks at each position of the expressway due to the limitation of the arrangement positions and the arrangement densities of fixed detectors, and the intrinsic relation of the data in different influence factors and spatial positions is difficult to extract and learn, so that the risk assessment precision is limited.
The invention adopts the following technical scheme for solving the technical problems:
the invention provides a highway traffic accident risk assessment method based on dangerous driving event data, which comprises the following steps:
s1: determining a research road section and a statistical time range, and respectively acquiring the following two types of information:
(1) And (3) counting dangerous driving event data of vehicles on the highway research road section within a time range, wherein the dangerous driving event data comprise the occurrence time of dangerous driving events, the longitude and latitude of the occurrence place and the uplink and downlink directions, the types of dangerous driving events (sudden acceleration, sudden deceleration, sudden left lane change and sudden right lane change), and the maximum acceleration and maximum speed of the vehicles in the dangerous driving events.
(2) And counting traffic accident data occurring on the highway research road section within the time range, wherein the traffic accident data comprise the starting time and the ending time of the accident, the longitude and the latitude of the accident place and the uplink and downlink directions.
S2, aiming at each traffic accident, searching dangerous driving event data of an upstream road section and a downstream road section of an accident site before the accident occurs, taking the dangerous driving event data as independent variables, and taking the accident as the dependent variables to construct an accident case data set;
s3, searching dangerous driving event data of an upstream road section and a downstream road section when no traffic accident occurs, taking the dangerous driving event data as an independent variable, taking the non-occurring accident as a dependent variable, constructing a non-accident case data set as a comparison group, and combining the accident case data set and the non-accident case data set to obtain a sample data set;
s4, carrying out standardized processing on the data: processing dangerous driving event data into an image form according to different dangerous driving event types and spatial position relations, constructing a deep learning model, and training and parameter adjustment until the model effect is optimal;
S5, acquiring dangerous driving event data of vehicles on all road sections of the expressway in real time, calculating to obtain occurrence frequency of various dangerous driving events and average value of maximum acceleration and maximum speed of the vehicles in the dangerous driving events according to the S2, inputting the average value into the deep learning model constructed in the S4, and calculating to obtain traffic accident risk assessment values and risk levels of all road sections.
Specifically, step S2 is specifically:
s2.1: and marking dangerous driving events and traffic accident positions on corresponding road sections of the expressway according to the longitude, latitude and uplink and downlink directions of the dangerous driving events and the traffic accident positions, and establishing an uplink and downlink relation between the road sections.
S2.2: and according to the space-time position of each traffic accident, four dangerous driving event data of rapid acceleration, rapid deceleration, rapid left lane change and rapid right lane change in the space range of the upstream and downstream of the accident site in the time range of t before the accident occurs are extracted.
S2.3: for each traffic accident, the following are calculated:
(1) Calculating occurrence frequencies of four dangerous driving events in the space-time range in the S2.2, and marking as follows:
Times＝[L_up,L_down,R_up,R_down,A_up,A_down,B_up,B_down]
wherein the first part of each variable represents the dangerous driving event type, namely L, R, A and B respectively represent the frequency of the occurrence of the driving events of sudden left lane change, sudden right lane change, sudden acceleration and sudden deceleration in the time range t before the accident occurs. The second part of each variable represents the location of the dangerous driving event, i.e. up and down represent road segments in the spatial range upstream and downstream of the accident location, respectively.
(2) Calculating the average value of the maximum acceleration and the maximum speed of the vehicle in the same type of dangerous driving event in the space-time range in S2.2, and marking as:
Acceleration＝[La_up,La_down,Ra_up,Ra_down,Aa_up,Aa_down,Ba_up,Ba_down]
Speed＝[Ls_up,Ls_down,Rs_up,Rs_down,As_up,As_down,Bs_up,Bs_down]
wherein La and Ls in the first portion of each variable represent the average of the maximum acceleration and maximum velocity of the vehicle in all the rapid left-hand lane-change events, respectively, and the remaining Ra, rs, aa, as, ba, bs are synonymous. The second portion of each variable represents a dangerous driving event occurrence location.
S2.4: and combining the time, acception and Speed data corresponding to each traffic accident to obtain dangerous driving event data with 24 characteristic variables. Summarizing dangerous driving event data corresponding to all traffic accidents, taking the dangerous driving event data as independent variables, and marking the independent variables as:
X i ＝[Times i ,Acceleration i ,Speed i ]
wherein, times i Representing the occurrence frequency of four dangerous driving events corresponding to the ith traffic accident i And Speed of i Meaning the same.
Each event case data is given a label y=1, and is denoted as y= [ Y ] as a dependent variable i ],y i =1. And matching the independent variable with the dependent variable to obtain an accident case data set.
Specifically, step S3 is specifically:
s3.1: for each traffic accident, extracting and calculating t before the accident according to the methods in S2.2, S2.3 and S2.4 0 、t 0 ～2t 0 、…、(n-1)t 0 ～nt 0 Dangerous driving event data in the space range of the upstream and downstream of the accident site in the time range are collected to obtain initial non-accident case data. Wherein n is a positive integer, controls the proportion of the control group selection, and can be selected according to actual conditions.
In order to avoid the influence caused by traffic accidents in the time range, checking the time of all the non-accident case data, and deleting the non-accident case data with the time between t time before any traffic accident and t' time after the accident. Deleting the case with information missing to obtain final non-accident case data, and recording the final non-accident case data as an independent variable:
X j ＝[Times j ,Acceleration j ,Speed j ]。
wherein, times j Represents the occurrence frequency of four dangerous driving events corresponding to the j-th non-accident case j And Speed of j Meaning the same.
For each non-accident case data, a label y=0 is given, and is denoted as y= [ Y ] as a dependent variable j ],y j =0. And matching the independent variable with the dependent variable to obtain the non-accident case data set.
S3.2: combining the accident case data set and the non-accident case data set to obtain a sample data set, and marking as:
X＝[Times k ,Acceleration k ,Speed k ]
＝[L_up k ,L_down k ,R_up k ,R_down k ,A_up k ,A_down k ,B_up k ,B_down k ,La_up k ,La_down k ,Ra_up k ,Ra_down k ,Aa_up k ,Aa_down k ,Ba_up k ,Ba_down k ,Ls_up k ,Ls_down k ,Rs_up k ,Rs_down k ,As_up k ,As_down k ,Bs_up k ,Bs_down k ]
Y＝[y k ],y k =0 or 1
Specifically, step S4 is specifically:
s4.1: the mean and standard deviation of each independent variable were calculated and all data were normalized as follows:
x norm ＝(x-μ)/σ
Where μ is the mean value of the independent variables and σ is the standard deviation of the independent variables.
S4.2: each dangerous driving event data is processed into a four-way one-dimensional image form as follows. The variables in each channel are as follows:
channels 1:L_up, la_up, ls_up, L_down, la_down, ls_down
Channel 2 R_up, ra_up, rs_up, R_down, ra_down, rs_down
Channels 3: A_up, aa_up, as_up, A_down, aa_down, as_down
Channels 4:B_up, ba_up, bs_up, B_down, ba_down, bs_down
Wherein 6 variables of the same driving behavior type are positioned in the same channel and are arranged according to the relation between the upstream road section and the downstream road section (from up to down) and the sequence of occurrence frequency, average maximum acceleration and average maximum speed of dangerous driving events. The variables for different types of dangerous driving events are located in 4 different lanes. The data form is beneficial to model extraction and learning of internal relations of the data in different dangerous driving event types and upstream and downstream space positions, and hidden modes in the data are mined.
S4.3: and establishing a deep learning model. And constructing a convolution layer, a pooling layer, a flattening layer, a full-connection layer and an output layer in sequence. The four-channel one-dimensional image data obtained in S4.2 is first input into the convolutional layer as input data of the deep learning model.
The input dimension of the convolution layer (Conv 1D layer) is I 1 The output dimension is O 1 Having m 1 Having a size n 1 Is a convolution kernel of (a). These convolution kernels are denoted as s 1 Along the direction of the variable alignment within the input data channel (e.g., from l_up to ls_down). The activation function of the convolutional layer is relu. The output of this module serves as the input to the pooling layer.
The input dimension of the pooling layer (AveragePooling 1D layer) is I 2 The output dimension is O 2 In s 2 Step size range n 2 Is a one-dimensional average pooling operation of (a). The output of the module serves as the input to the flattening layer.
The input dimension of the flattening layer (flat layer) is I 3 The output dimension is O 3 . The multi-channel one-dimensional image data output by the pooling layer is converted into a one-dimensional vector form, and a random discarding method with the ratio of p is adopted to discard the data so as to avoid overfitting. The output of the module serves as the input to the fully connected layer.
The input dimension of the full connection layer (Dense layer) is I 4 The output dimension is O 4 Having m 2 And neurons. The activation function of the fully connected layer is relu. At the same time adopt the ratio p drop The random discard method of (2) discards neurons. The output of the module serves as the input to the output layer.
The Output layer (Output layer) has 1 neuron. The activation function of the output layer is sigmoid, which converts the output value into a (0, 1) interval. Therefore, the meaning of the output data of the whole deep learning model is the probability of traffic accidents of the current research road section.
S4.4: training the established deep learning model based on the sample dataset. The Area (AUC), the Sensitivity (Sensitivity) and the Specificity (Specificity) Under the working characteristic Curve of the test subject are used as evaluation indexes, and a ten-fold cross validation method, a binary cross entropy loss function and a Nadam optimizer are adopted to optimize the established deep learning model based on a sample data set, so that the deep learning model with the optimal effect is obtained.
Specifically, step S5 is specifically:
s5.1: and acquiring four dangerous driving event data of sudden acceleration, sudden deceleration, sudden left lane change and sudden right lane change of the vehicle on the whole highway section in real time. And (2) calculating the occurrence frequency Times ' of four dangerous driving events in the time range of t before each position of the expressway and the average value of the maximum Acceleration Speed ' and the maximum Speed ' of the vehicle in all the dangerous driving events of the same type according to the S2. And summarizing to obtain input data.
S5.2: and inputting the data into the deep learning model constructed in the step S4 to obtain the probability p epsilon (0, 1) of the occurrence of the traffic accident at each position of the current expressway, namely the traffic accident risk assessment value.
When p epsilon (0, threshold), the traffic accident risk level is low risk, and no intervention is needed.
When p is epsilon [ threshold, 1), the traffic accident risk level is high risk, and corresponding measures are needed to be taken for controlling traffic safety risk.
Wherein threshold E (0, 1).
The invention also provides an expressway traffic accident risk assessment system based on dangerous driving event data, which comprises:
the data acquisition module is used for determining a research road section and a statistical time range and acquiring dangerous driving event data on the expressway research road section and traffic accident data on the research road section within the statistical time range;
the sample construction module is used for searching dangerous driving event data of an upstream road section and a downstream road section of an accident site before the accident occurs according to each traffic accident, taking the dangerous driving event data as independent variables and taking the accident as the dependent variables, and constructing an accident case data set; then, searching dangerous driving event data of an upstream road section and a downstream road section when no traffic accident occurs, taking the dangerous driving event data as independent variables, taking the non-occurring accident as dependent variables, constructing a non-accident case data set as a comparison group, and combining the accident case data set and the non-accident case data set to obtain a sample data set;
The model training module is used for carrying out standardized processing on the data, processing the data into an image form according to different dangerous driving event types and upstream and downstream spatial relations, constructing a deep learning model, and carrying out training and parameter adjustment until the model effect is optimal;
the road section accident risk assessment module is used for acquiring dangerous driving event data of vehicles on the whole road section of the expressway in real time, calculating the occurrence frequency of dangerous driving events and the average value of the maximum acceleration and the maximum speed of the vehicles in all the same type of dangerous driving events, inputting the average value into the constructed deep learning model, and calculating to obtain the traffic accident risk assessment value and the risk level of each road section.
Compared with the prior art, the invention has the following beneficial technical effects:
the highway traffic accident risk assessment method based on the dangerous driving event data is not limited by the layout positions and the layout densities of the fixed detectors, can cover the whole road section of the highway and is not influenced by weather conditions, and the traffic accident risk at each road section of the highway can be assessed in real time and in a refined manner under the condition that new equipment is not required to be installed.
Meanwhile, the method comprehensively considers the internal relation of dangerous driving event data in different event types and upstream and downstream space positions, evaluates the road accident risk according to two factors of the occurrence frequency of dangerous driving events on roads and the vehicle motion parameters in the dangerous driving events, can improve the accuracy of risk evaluation, and provides new theory and technical support for the management and control of traffic safety risks by expressway traffic management departments.
Drawings
FIG. 1 is a flow chart of a highway traffic accident risk assessment method in the present invention.
Fig. 2 is a schematic diagram of extraction rules of dangerous driving event data before occurrence of a traffic accident in the present invention.
FIG. 3 is a schematic diagram of a deep learning model framework according to the present invention.
Detailed Description
The technical solutions of the embodiments of the present invention will be fully described in detail below with reference to the accompanying drawings. Furthermore, the described embodiments are only some, but not all, embodiments of the invention. Based on the embodiments of the present invention, other embodiments that may be obtained by those of ordinary skill in the art without making any inventive effort are within the scope of the present invention.
Fig. 1 is a flow chart of a highway traffic accident risk assessment method based on dangerous driving event data, which can be applied to highway traffic accident risk assessment with traffic accident data and dangerous driving event data including vehicle motion parameters. The method comprises the following specific steps:
s1: determining a research road section and a statistical time range, and respectively acquiring the following two types of information:
(1) And (3) counting dangerous driving event data of vehicles on the highway research road section within a time range, wherein the dangerous driving event data comprise the occurrence time of dangerous driving events, the longitude and latitude of the occurrence place and the uplink and downlink directions, the types of dangerous driving events (sudden acceleration, sudden deceleration, sudden left lane change and sudden right lane change), and the maximum acceleration and maximum speed of the vehicles in the dangerous driving events.
(2) And counting traffic accident data occurring on the highway research road section within the time range, wherein the traffic accident data comprise the starting time and the ending time of the accident, the longitude and the latitude of the accident place and the uplink and downlink directions.
S2: and aiming at each traffic accident, searching dangerous driving event data of the road sections upstream and downstream of the accident site before the accident occurs. And constructing an accident case data set by taking dangerous driving event data as an independent variable and taking an accident as a dependent variable. The method comprises the following specific steps:
s2.1: and marking dangerous driving events and traffic accident positions on corresponding road sections of the expressway according to the longitude, latitude and uplink and downlink directions of the dangerous driving events and the traffic accident positions, and establishing an uplink and downlink relation between the road sections.
S2.2: as shown in fig. 2, according to the space-time position of each traffic accident, four dangerous driving event data, namely, the part circled by the oval dotted line frame in fig. 2, of rapid acceleration, rapid deceleration, rapid left lane change and rapid right lane change in the range of l meters upstream and downstream of the accident site in the range of t minutes before the accident occurs are extracted. Typically t is less than or equal to 30 and l is less than or equal to 1000.
S2.3: for each traffic accident, the following are calculated:
(1) The occurrence frequency of four dangerous driving events in the space-time range shown in fig. 2 is calculated and is recorded as:
Times＝[L_up,L_down,R_up,R_down,A_up,A_down,B_up,B_down]
Wherein the first part of each variable represents the dangerous driving event type, namely L, R, A and B respectively represent the frequency of the occurrence of the driving events of sudden left lane change, sudden right lane change, sudden acceleration and sudden deceleration in the time range before the accident. The second part of each variable represents the location of the dangerous driving event, i.e. up and down represent road segments in the spatial range upstream and downstream of the accident location, respectively.
(2) The average of the maximum acceleration and maximum speed of the vehicle in the same type of dangerous driving event in the space-time range shown in fig. 2 is calculated and is recorded as:
Acceleration＝[La_up,La_down,Ra_up,Ra_down,Aa_up,Aa_down,Ba_up,Ba_down]
Speed＝[Ls_up,Ls_down,Rs_up,Rs_down,As_up,As_down,Bs_up,Bs_down]
wherein La and Ls in the first portion of each variable represent the average of the maximum acceleration and maximum velocity of the vehicle in all the rapid left-hand lane-change events, respectively, and the remaining Ra, rs, aa, as, ba, bs are synonymous. The second portion of each variable represents a dangerous driving event occurrence location.
S2.4: and combining the time, acception and Speed data corresponding to each traffic accident to obtain dangerous driving event data with 24 characteristic variables. Summarizing dangerous driving event data corresponding to all traffic accidents, taking the dangerous driving event data as independent variables, and marking the independent variables as:
X i ＝[Times i ,Acceleration i ,Speed i ]
wherein, times i Representing the occurrence frequency of four dangerous driving events corresponding to the ith traffic accident i And Speed of i Meaning the same.
Each event case data is given a label y=1, and is denoted as y= [ Y ] as a dependent variable i ],y i =1. And matching the independent variable with the dependent variable to obtain an accident case data set.
S3: and searching dangerous driving event data of the upstream road section and the downstream road section when no traffic accident occurs. And constructing a non-accident case data set by taking dangerous driving event data as an independent variable and taking non-accident as a dependent variable as a control group. Combining the accident case data set and the non-accident case data set to obtain a sample data set. The method comprises the following specific steps:
s3.1: for each traffic accident, extracting and calculating t before the accident according to the methods in S2.2, S2.3 and S2.4 0 、t 0 ～2t 0 、…、(n-1)t 0 ～nt 0 Dangerous driving event data in the space range of the upstream and downstream of the accident site in the time range are collected to obtain initial non-accident case data. General t 0 Not less than 1 and t 0 Is a positive integer. n is a positive integer, controls the proportion of the control group selection, and can be selected according to actual conditions.
In order to avoid the influence caused by traffic accidents in the time range, checking the time of all the non-accident case data, and deleting the non-accident case data with the time between t time before any traffic accident and t' time after the accident. Deleting the case with information missing to obtain final non-accident case data, and recording the final non-accident case data as an independent variable:
X j ＝[Times j ,Acceleration j ,Speed j ]。
Wherein, times j Represents the occurrence frequency of four dangerous driving events corresponding to the j-th non-accident case j And Speed of j Meaning the same.
Each non-accident case data is given a label y=0, and is recorded as a dependent variableY＝[y j ],y j =0. And matching the independent variable with the dependent variable to obtain the non-accident case data set.
S3.2: combining the accident case data set and the non-accident case data set to obtain a sample data set, and marking as:
X＝[Times k ,Acceleration k ,Speed k ]
＝[L_up k ,L_down k ,R_up k ,R_down k ,A_up k ,A_down k ,B_up k ,B_down k ,La_up k ,La_down k ,Ra_up k ,Ra_down k ,Aa_up k ,Aa_down k ,Ba_up k ,Ba_down k ,Ls_up k ,Ls_down k ,Rs_up k ,Rs_down k ,As_up k ,As_down k ,Bs_up k ,Bs_down k ]
Y＝[y k ],y k =0 or 1
S4: and (5) carrying out standardization processing on the data. And processing dangerous driving event data into an image form according to different dangerous driving event types and spatial position relations, constructing a deep learning model, and training and parameter adjustment until the model effect is optimal. The method comprises the following specific steps:
s4.1: the mean and standard deviation of each independent variable were calculated and all data were normalized as follows:
x norm ＝(x-μ)/σ
where μ is the mean value of the independent variables and σ is the standard deviation of the independent variables.
S4.2: each dangerous driving event data is processed into a four-way one-dimensional image form as shown in the input layer of fig. 3 (part of the variable names are not labeled due to space limitations). The variables in each channel are as follows:
channels 1:L_up, la_up, ls_up, L_down, la_down, ls_down
Channel 2 R_up, ra_up, rs_up, R_down, ra_down, rs_down
Channels 3: A_up, aa_up, as_up, A_down, aa_down, as_down
Channels 4:B_up, ba_up, bs_up, B_down, ba_down, bs_down
Wherein 6 variables of the same driving behavior type are positioned in the same channel and are arranged according to the relation between the upstream road section and the downstream road section (from up to down) and the sequence of occurrence frequency, average maximum acceleration and average maximum speed of dangerous driving events. The variables for different types of dangerous driving events are located in 4 different lanes. The data form is beneficial to model extraction and learning of internal relations of the data in different dangerous driving event types and upstream and downstream space positions, and hidden modes in the data are mined.
S4.3: as shown in fig. 3, a convolution layer, a pooling layer, a flattening layer, a full connection layer, and an output layer are sequentially constructed. The four-channel one-dimensional image data obtained in S4.2 is first input into the convolutional layer as input data of the deep learning model.
The input dimension of the convolution layer (Conv 1D layer) is I 1 The output dimension is O 1 Having m 1 Having a size n 1 Is a convolution kernel of (a). These convolution kernels are denoted as s 1 Along the direction of variable alignment within the input data channel (the direction of the input layer arrow of fig. 3). The activation function of the convolutional layer is relu. The output of this module serves as the input to the pooling layer.
The input dimension of the pooling layer (AveragePooling 1D layer) is I 2 The output dimension is O 2 In s 2 Step size range n 2 Is used for reducing the dimension of data and avoiding overfitting. The output of the module serves as the input to the flattening layer.
The input dimension of the flattening layer (flat layer) is I 3 The output dimension is O 3 . The multi-channel one-dimensional image data output by the pooling layer is converted into a one-dimensional vector form, and the data is discarded by adopting a random discarding method with the ratio of p so as to avoid overfitting. The output of the module serves as the input to the fully connected layer.
The input dimension of the full connection layer (Dense layer) is I 4 The output dimension is O 4 Having m 2 And neurons. The activation function of the fully connected layer is relu. At the same time, adopt the ratio ofp drop The random discard method of (2) discards neurons. The output of the module serves as the input to the output layer.
The Output layer (Output layer) has 1 neuron. The activation function of the output layer is sigmoid, which converts the output value into a (0, 1) interval. Therefore, the meaning of the output data of the whole deep learning model is the probability of traffic accidents of the current research road section.
S4.4: training the established deep learning model based on the sample dataset. The effect of the deep learning model was evaluated using the Area Under the subject's work characteristic Curve (AUC, i.e., area index), sensitivity (Sensitivity), and Specificity (Specificity) as evaluation indexes. Specifically, the about sign index under each classification threshold is calculated by using the about sign index method, and the calculation formula is as follows:
Youden＝Sensitivity+Specificity-1
Where TP represents the number of incidents in practice and in prediction, FN represents the number of incidents in practice but in prediction, TN represents the number of non-incidents in practice and in prediction, and FP represents the number of incidents in practice but in prediction. The threshold corresponding to the maximum value of the about index (Youden) is selected as the classification threshold (threshold) of the model, and the sensitivity and specificity of the model under the threshold are used as model evaluation indexes. The closer the AUC, sensitivity and specificity values of the model are to 1, the better the model predictive performance.
Based on the three evaluation indexes, a ten-fold cross validation method, a binary cross entropy loss function and a Nadam optimizer are adopted to perform optimization on the established deep learning model based on the sample data set, so that the deep learning model with the optimal effect is obtained.
S5: dangerous driving event data of vehicles on all road sections of the expressway are acquired in real time, input into a deep learning model constructed in the step S4, and the risk assessment value and the risk level of traffic accidents on each road section are calculated. The method comprises the following specific steps:
s5.1: and acquiring four dangerous driving event data of rapid acceleration, rapid deceleration, rapid left lane change and rapid right lane change of vehicles on the whole road section of the expressway in real time, and establishing an upstream-downstream relationship between road sections. And (2) obtaining the occurrence frequency Times ' of four dangerous driving events in the time range of t before each position of the expressway and the average value of the maximum Acceleration Speed ' and the maximum Speed ' of the vehicle in all the dangerous driving events of the same type according to the calculation method in the S2, and collecting the input data X of the model.
S5.2: and inputting X into the deep learning model constructed in the S4 to obtain the probability p epsilon (0, 1) of the occurrence of the traffic accident at each position of the current expressway, namely the traffic accident risk assessment value.
When p epsilon (0, threshold), the traffic accident risk level is low risk, and no intervention is needed.
When p is epsilon [ threshold, 1), the traffic accident risk level is high risk, and corresponding measures are needed to be taken for controlling the safety risk.
Wherein threshold E (0, 1).
Specific cases
In order to show the applicability of the highway traffic accident risk assessment method based on the dangerous driving event data provided by the embodiment of the invention in a real scene, the following specific cases are given for further explanation:
taking the Liyang-Changxing section of the expressway of China as an example, the data collection time range is from 26 days of 9 months in 2020 to 2 days of 10 months in 2020. Dangerous driving event data of vehicles on the road section are collected, and the time interval for data collection is 1 second. The data fields include the time of occurrence of the dangerous driving event, the longitude, latitude and ascending and descending directions of the place where the event is located, the type of the dangerous driving event (sudden acceleration, sudden deceleration, sudden left lane change and sudden right lane change), and the motion parameters (maximum acceleration and maximum speed) of the vehicle in the dangerous driving event. Traffic accident data on the road section in the time range is collected, and the data fields comprise the starting time and the ending time of the accident, the longitude and the latitude of the accident place and the uplink and downlink directions. After arrangement, 371 traffic accidents are recorded on the research road section in the statistical range.
Because the recorded traffic accident occurrence time is slightly later than the actual occurrence time, the shorter time interval for summarizing the dangerous driving event data can influence the prediction accuracy. In addition, the summary time interval is shorter, the number of dangerous driving events in each interval is smaller, and the distribution difference in different intervals is larger, so that modeling is not facilitated. Thus, the present case selects 30 minutes as the time interval for data summarization. In addition, 250 meters are selected as the space interval selected by the dangerous driving event data, so that dangerous driving events with great influence on traffic accidents can be extracted finely. I.e. t=30, l=250.
According to the step of the specific embodiment S2, dangerous driving event data of the upstream and downstream road sections before the occurrence of the accident are extracted and data fusion is carried out, so that 265 accident cases are obtained. For each traffic accident, the steps according to the specific embodiment S3 are adopted to extract and calculate the time (i.e. t) 0-1, 1-2, 2-3 hours before the accident occurs 0 The dangerous driving event data in 250 meters upstream and downstream of the accident site in the range of 1) is taken as a corresponding non-accident case of the traffic accident, and initial non-accident case data is obtained by summarizing.
To avoid the impact of accident cases, the time of all non-accident cases is checked, and the non-accident case data is deleted within 30 minutes before any traffic accident to 180 minutes after the accident. And deleting the cases with the information missing to obtain final 722 pieces of non-accident case data, and forming a sample data set with the accident-to-non-accident ratio of about 1:3.
According to the step of the specific implementation mode S4, data are standardized, each dangerous driving event data are processed into four-channel one-dimensional image forms, and a deep learning model comprising a one-dimensional convolution module is established. The specific parameters of the model are as follows:
the input dimension of the convolution layer is (N, 6, 4) with 512 convolution kernels of size 3. These convolution kernels perform one-dimensional convolution operations in steps of 1. Thus, the output dimension is (N, 4,512). Wherein N is the number of training set samples.
The pooling layer performs a one-dimensional average pooling operation in a range of 3 in steps of 1. Thus, the output dimension is (N, 2,512).
The flattening layer converts the multi-channel one-dimensional image data output by the pooling layer into a one-dimensional vector form, and discards the data by adopting a random discarding method with the ratio of 0.5 so as to avoid overfitting. Thus, the output dimension is (N, 1024).
The fully connected layer has 32 neurons and the activation function is relu. Neurons were discarded using a random discard method with a ratio of 0.5. Thus, the output dimension is (N, 32).
The output layer has 1 neuron, and the output dimension is (N, 1), i.e., a one-dimensional vector. The activation function of the output layer is sigmoid. The meaning of the output data is the probability of traffic accidents of the current research road section.
And training a deep learning model by adopting a ten-fold cross validation method, a binary cross entropy loss function and a Nadam optimizer, and performing super-parameter tuning to obtain the deep learning model with the optimal effect.
And taking the verification set data as dangerous driving event data of the expressway full road section vehicle obtained in real time, and according to the step of the specific implementation mode S5, inputting the dangerous driving event data into a trained deep learning model to calculate a traffic accident risk assessment value. Experiments show that compared with three reference models (input data only supports one-dimensional vector form) of logistic regression, a support vector machine and an artificial neural network model, the AUC value of the deep learning model (input data is in an image form) provided by the invention is highest, 71.3% of accident conditions and 75.1% of non-accident conditions can be identified under an optimal threshold, and the model with sensitivity and specificity being more than 70% is the only model. Meanwhile, the deep learning model has the highest sensitivity under the condition of smaller false alarm rate (less than 30%). Experimental results of specific cases prove that the highway accident risk assessment method based on the dangerous driving event data has the best comprehensive performance, can extract and learn the internal connection of the dangerous driving event data in different event types and upstream and downstream space positions, effectively improves the accuracy of risk assessment, and has higher application value.
The foregoing description of the embodiments of the invention is provided to facilitate the understanding of the principles of the invention and its core ideas, and is not intended to limit the invention. Modifications, equivalents, and alternatives to the described embodiments of the invention will occur to those skilled in the art and are intended to be included within the spirit and principles of the invention.
Claims (10)
1. The highway traffic accident risk assessment method based on the dangerous driving event data is characterized by comprising the following steps of:
s1, determining a research road section and a statistical time range, and acquiring vehicle dangerous driving event related data on the expressway research road section and traffic accident data on the research road section in the statistical time range;
s2, aiming at each traffic accident, searching dangerous driving event data of an upstream road section and a downstream road section of an accident site before the accident occurs, taking the dangerous driving event data as independent variables, and taking the accident as the dependent variables to construct an accident case data set;
s3, searching dangerous driving event data of an upstream road section and a downstream road section when no traffic accident occurs, taking the dangerous driving event data as an independent variable, taking the non-occurring accident as a dependent variable, constructing a non-accident case data set as a comparison group, and combining the accident case data set and the non-accident case data set to obtain a sample data set;
S4, carrying out standardized processing on the data: processing dangerous driving event data into an image form according to different dangerous driving event types and spatial position relations, constructing a deep learning model, and training and parameter adjustment until the model effect is optimal;
s5, acquiring dangerous driving event data of vehicles on all road sections of the expressway in real time, calculating to obtain occurrence frequency of various dangerous driving events and average value of maximum acceleration and maximum speed of the vehicles in the dangerous driving events according to the S2, inputting the average value into the deep learning model constructed in the S4, and calculating to obtain traffic accident risk assessment values and risk levels of all road sections.
2. The highway traffic accident risk assessment method based on dangerous driving event data according to claim 1, wherein the data acquired in step S1 includes:
(1) Dangerous driving event related data of vehicles on an expressway, comprising: the occurrence time of dangerous driving events, the longitude, latitude and uplink and downlink directions of the occurrence place, the type of the dangerous driving events, and the maximum acceleration and maximum speed of the vehicle in the dangerous driving events;
(2) Highway traffic accident data comprising: the starting time and the ending time of the accident, the longitude and the latitude of the accident place and the uplink and downlink directions.
3. The highway traffic accident risk assessment method based on the dangerous driving event data according to claim 2, wherein the dangerous driving event type in step S1 includes: rapid acceleration, rapid deceleration, rapid left lane change and rapid right lane change.
4. The highway traffic accident risk assessment method based on dangerous driving event data according to claim 1, wherein step S2 specifically comprises the following sub-steps:
s201, marking dangerous driving events and traffic accident positions on corresponding road sections of the expressway according to longitude, latitude and uplink and downlink directions, and establishing an uplink and downlink relation between the road sections;
s202, according to the space-time position of each traffic accident, four dangerous driving event data of rapid acceleration, rapid deceleration, rapid left lane change and rapid right lane change in the space range of the upstream and downstream of the accident site in the time range of t before the accident occurs are extracted;
s203, for each traffic accident, respectively calculating the following contents:
(1) Calculating occurrence frequencies of four dangerous driving events in the space-time range in the step S202, and marking as:
Times＝[L_up,L_down,R_up,R_down,A_up,A_down,B_up,B_down]
wherein a first part of each variable represents a dangerous driving event type, namely L, R, A and B respectively represent the frequency of occurrence of sudden left lane change, sudden right lane change, sudden acceleration and sudden deceleration driving events in a time range t before an accident occurs, and a second part of each variable represents a dangerous driving event occurrence place, namely up and down respectively represent road sections in a space range of an upstream and a downstream of the accident place;
(2) Calculating the average value of the maximum acceleration and the maximum speed of the vehicle in the same type of dangerous driving event in the space-time range in the step S202, and marking as:
Acceleration＝[La_up,La_down,Ra_up,Ra_down,Aa_up,Aa_down,Ba_up,Ba_down]
Speed＝[Ls_up,Ls_down,Rs_up,Rs_down,As_up,As_down,Bs_up,Bs_down]
wherein La, ls in the first portion of each variable represent average values of maximum acceleration and maximum velocity of the vehicle in all rapid left lane change events, ra, rs represent average values of maximum acceleration and maximum velocity of the vehicle in all rapid right lane change events, aa, as represent average values of maximum acceleration and maximum velocity of the vehicle in all rapid acceleration events, ba, bs represent average values of maximum acceleration and maximum velocity of the vehicle in all rapid deceleration events, respectively;
s204: combining the time, acception and Speed data corresponding to each traffic accident to obtain dangerous driving event data with 24 characteristic variables, summarizing the dangerous driving event data corresponding to all the traffic accidents, and recording the dangerous driving event data as independent variables as: x is X i ＝[Times i ,Acceleration i ,Speed i ]Wherein, times i Representing the occurrence frequency of four dangerous driving events corresponding to the ith traffic accident i And Speed of i Respectively representing average values of maximum acceleration and maximum speed of the vehicle in four dangerous driving events corresponding to the ith traffic accident;
Each event case data is given a label y=1, and is denoted as y= [ Y ] as a dependent variable i ],y i =1, matching the independent and dependent variables, resulting in an incident case dataset.
5. The highway traffic accident risk assessment method based on dangerous driving event data according to claim 1, wherein step S3 specifically comprises the following sub-steps:
s301: for each traffic accident, t before the accident is extracted and calculated 0 、t 0 ～2t 0 、…、(n-1)t 0 ～nt 0 Dangerous driving event data in the space range of the upstream and downstream of the accident site in the time range are summarized to obtain initial non-accident case data, wherein n is a positive integer, and the positive integer is used for controlling the proportion selected by the comparison group;
checking the time of all initial non-accident cases, and deleting non-accident case data with the time between t time before any traffic accident and t' time after the accident; deleting the case with information missing to obtain final non-accident case data, and recording the final non-accident case data as an independent variable: x is X j ＝[Times j ,Acceleration j ,Speed j ]Wherein, times j Represents the occurrence frequency of four dangerous driving events corresponding to the j-th non-accident case j And Speed of j Respectively representing average values of maximum acceleration and maximum speed of the vehicle in four dangerous driving events corresponding to the j-th non-accident case;
For each non-accident case data, a label y=0 is given, and is denoted as y= [ Y ] as a dependent variable j ],y j =0, matching the independent variable and the dependent variable to obtain a non-accident case data set;
s302, combining the accident case data set and the non-accident case data set to obtain a sample data set, and marking as:
X＝[Times k ,Acceleration k ,Speed k ]
＝[L_up k ,L_down k ,R_up k ,R_down k ,A_up k ,A_down k ,B_up k ,B_down k ,La_up k ,La_down k ,Ra_up k ,Ra_down k ,Aa_up k ,Aa_down k ,Ba_up k ,Ba_down k ,Ls_up k ,Ls_down k ,Rs_up k ,Rs_down k ,As_up k ,As_down k ,Bs_up k ,Bs_down k ]
Y＝[y k ],y k =0 or 1.
6. The highway traffic accident risk assessment method based on dangerous driving event data according to claim 1, wherein step S4 specifically comprises the following sub-steps:
s401, press x norm Normalized data with = (x- μ)/σ, where μ is the average value and σ is the standard deviation;
s402, processing each dangerous driving event data into a four-channel one-dimensional image form, wherein variables in each channel are as follows:
channels 1:L_up, la_up, ls_up, L_down, la_down, ls_down
Channel 2 R_up, ra_up, rs_up, R_down, ra_down, rs_down
Channels 3: A_up, aa_up, as_up, A_down, aa_down, as_down
Channels 4:B_up, ba_up, bs_up, B_down, ba_down, bs_down
The method comprises the steps that 6 variables of the same driving behavior type are located in the same channel, and are arranged according to the relation between an upstream road section and a downstream road section and the sequence of occurrence frequency, average maximum acceleration and average maximum speed of dangerous driving events, and the variables of different types of dangerous driving events are located in 4 different channels;
S403, establishing a deep learning model: sequentially constructing a convolution layer, a pooling layer, a flattening layer, a full-connection layer and an output layer, and inputting the four-channel one-dimensional image data obtained in the step S402 into the convolution layer as input data of a deep learning model;
s404, training and optimizing the established deep learning model based on the sample data set: and taking the area, the sensitivity and the specificity under the working characteristic curve of the test subject as evaluation indexes, and adopting a ten-fold cross validation method, a binary cross entropy loss function and a Nadam optimizer for tuning and optimizing to obtain the deep learning model with the optimal effect.
7. According to claim 6The highway traffic accident risk assessment method based on the dangerous driving event data is characterized in that in step S403, the convolution layer has m 1 Having a size n 1 Is a convolution kernel of s 1 Performing one-dimensional convolution operation along the variable arrangement direction in the input data channel, wherein the activation function is relu, and the output of the convolution layer is used as the input of the pooling layer;
the pooling layer is s 2 Step size range n 2 The output of the pooling layer is used as the input of the flattening layer;
the flattening layer converts the multi-channel one-dimensional image data output by the pooling layer into a one-dimensional vector form, a random discarding method with the ratio of p is adopted to avoid overfitting, and the output of the flattening layer is used as the input of the full-connection layer;
The full connection layer has m 2 The neurons are activated as relu, and the random discarding method with the ratio of p is adopted to avoid overfitting, and the ratio of p is adopted drop The output of the full connection layer is used as the input of the output layer;
the output of the full-connection layer is provided with 1 neuron, the activation function of the output layer is sigmoid, and the meaning of the output data is the probability of traffic accidents in the current research road section.
8. The highway traffic accident risk assessment method based on dangerous driving event data according to claim 1, wherein step S5 specifically comprises:
s501, obtaining data of four dangerous driving events of rapid Acceleration, rapid deceleration, rapid left lane change and rapid right lane change of a vehicle on a whole highway section in real time, calculating the occurrence frequency Times ' of the four dangerous driving events in a time range of t before each position of the highway according to the step S2, and collecting average values of the maximum Acceleration accelization ' and the maximum Speed ' of the vehicle in all the same type of dangerous driving events to obtain input data of a model;
s502: inputting data into the deep learning model constructed in the step S4, and calculating the probability p E (0, 1) of the occurrence of the traffic accident at each position of the current expressway, namely, the traffic accident risk assessment value;
When p is E (0, threshold), the traffic accident risk level is low risk, and no intervention is needed;
when p is epsilon (threshold, 1), the traffic accident risk level is high risk, and corresponding measures are needed to be adopted to manage and control the safety risk;
wherein threshold E (0, 1).
9. A highway traffic accident risk assessment system based on dangerous driving event data, comprising:
the data acquisition module is used for determining a research road section and a statistical time range and acquiring dangerous driving event data on the expressway research road section and traffic accident data on the research road section within the statistical time range;
the sample construction module is used for searching dangerous driving event data of an upstream road section and a downstream road section of an accident site before the accident occurs according to each traffic accident, taking the dangerous driving event data as independent variables and taking the accident as the dependent variables, and constructing an accident case data set; searching dangerous driving event data of an upstream road section and a downstream road section when no traffic accident occurs, taking the dangerous driving event data as an independent variable, taking the non-occurring accident as a dependent variable, constructing a non-accident case data set as a comparison group, and combining the accident case data set and the non-accident case data set to obtain a sample data set;
The model training module is used for carrying out standardized processing on the data, processing the data into an image form according to different dangerous driving event types and upstream and downstream spatial relations, constructing a deep learning model, and carrying out training and parameter adjustment until the model effect is optimal;
the road section accident risk assessment module is used for acquiring dangerous driving event data of vehicles on the whole road section of the expressway in real time, calculating the occurrence frequency of dangerous driving events and the average value of the maximum acceleration and the maximum speed of the vehicles in all the same type of dangerous driving events, inputting the average value into the constructed deep learning model, and calculating to obtain the traffic accident risk assessment value and the risk level of each road section.
10. The highway traffic accident risk assessment system based on dangerous driving event data according to claim 9, wherein the model training module constructs a deep learning model by constructing a convolution layer, a pooling layer, a flattening layer, a full connection layer and an output layer in sequence, wherein
The convolution layer has m 1 Having a size n 1 Are convolved with s 1 Performing one-dimensional convolution operation along the variable arrangement direction in the input data channel, wherein the activation function is relu, and the output of the convolution layer is used as the input of the pooling layer;
The pooling layer is s 2 Step size range n 2 The output of the pooling layer is used as the input of the flattening layer;
the flattening layer converts the multi-channel one-dimensional image data output by the pooling layer into a one-dimensional vector form, a random discarding method with the ratio of p is adopted to avoid overfitting, and the output of the flattening layer is used as the input of the full-connection layer;
the full connection layer has m 2 The neurons are activated as relu, and the random discarding method with the ratio of p is adopted to avoid overfitting, and the ratio of p is adopted drop The output of the flattening layer is used as the input of the output layer;
the output of the full-connection layer is provided with 1 neuron, the activation function of the output layer is sigmoid, and the meaning of the output data is the probability of traffic accidents in the current research road section.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310569137.8A CN116631186A (en) | 2023-05-19 | 2023-05-19 | Expressway traffic accident risk assessment method and system based on dangerous driving event data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310569137.8A CN116631186A (en) | 2023-05-19 | 2023-05-19 | Expressway traffic accident risk assessment method and system based on dangerous driving event data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116631186A true CN116631186A (en) | 2023-08-22 |
Family
ID=87591311
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310569137.8A Pending CN116631186A (en) | 2023-05-19 | 2023-05-19 | Expressway traffic accident risk assessment method and system based on dangerous driving event data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116631186A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117273964A (en) * | 2023-11-21 | 2023-12-22 | 国任财产保险股份有限公司 | Intelligent vehicle insurance policy generation system and method for self-adaptive driving data |
| CN117351708A (en) * | 2023-10-08 | 2024-01-05 | 北京迈道科技有限公司 | Expressway safety operation management early warning method, system and storage medium |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355877A (en) * | 2016-08-23 | 2017-01-25 | 重庆大学 | Method of simulated estimate of expressway-traffic-accident-affected areas based on particle filter algorithm |
| CN108198415A (en) * | 2017-12-28 | 2018-06-22 | 同济大学 | A kind of city expressway accident forecast method based on deep learning |
| CN110210090A (en) * | 2019-05-22 | 2019-09-06 | 淮阴工学院 | Motor vehicle accident analysis method based on uncertainty theory and genetic algorithm |
| US20200062249A1 (en) * | 2018-08-22 | 2020-02-27 | Cubic Corporation | Connected and Autonomous Vehicle (CAV) Behavioral Adaptive Driving |
| CN110866677A (en) * | 2019-10-25 | 2020-03-06 | 东南大学 | Driver relative risk evaluation method based on benchmark analysis |
| CN112037513A (en) * | 2020-09-01 | 2020-12-04 | 清华大学 | Real-time traffic safety index dynamic comprehensive evaluation system and construction method thereof |
| CN112201038A (en) * | 2020-09-28 | 2021-01-08 | 同济大学 | Road network risk assessment method based on risk of bad driving behavior of single vehicle |
| CN112837533A (en) * | 2021-01-08 | 2021-05-25 | 合肥工业大学 | Highway accident frequency prediction method considering risk factor time-varying characteristics |
| CN112990545A (en) * | 2021-02-08 | 2021-06-18 | 东南大学 | Traffic safety state prediction method for expressway intersection area |
| WO2021133789A1 (en) * | 2019-12-27 | 2021-07-01 | Lyft, Inc. | Systems and methods for incident detection using inference models |
| CN113689071A (en) * | 2021-07-15 | 2021-11-23 | 东南大学 | Multi-driver risk evaluation-based active intervention method for bad driving behaviors |
| CN113762734A (en) * | 2021-08-17 | 2021-12-07 | 淮阴工学院 | Dangerous chemical vehicle highway driving risk assessment method and system |
| KR102416711B1 (en) * | 2022-02-21 | 2022-07-05 | 주식회사 카비 | Accident risk prediction device for each driver using artificial intelligence and accident risk prediction method using the same |
-
2023
- 2023-05-19 CN CN202310569137.8A patent/CN116631186A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355877A (en) * | 2016-08-23 | 2017-01-25 | 重庆大学 | Method of simulated estimate of expressway-traffic-accident-affected areas based on particle filter algorithm |
| CN108198415A (en) * | 2017-12-28 | 2018-06-22 | 同济大学 | A kind of city expressway accident forecast method based on deep learning |
| US20200062249A1 (en) * | 2018-08-22 | 2020-02-27 | Cubic Corporation | Connected and Autonomous Vehicle (CAV) Behavioral Adaptive Driving |
| CN110210090A (en) * | 2019-05-22 | 2019-09-06 | 淮阴工学院 | Motor vehicle accident analysis method based on uncertainty theory and genetic algorithm |
| CN110866677A (en) * | 2019-10-25 | 2020-03-06 | 东南大学 | Driver relative risk evaluation method based on benchmark analysis |
| WO2021133789A1 (en) * | 2019-12-27 | 2021-07-01 | Lyft, Inc. | Systems and methods for incident detection using inference models |
| CN112037513A (en) * | 2020-09-01 | 2020-12-04 | 清华大学 | Real-time traffic safety index dynamic comprehensive evaluation system and construction method thereof |
| CN112201038A (en) * | 2020-09-28 | 2021-01-08 | 同济大学 | Road network risk assessment method based on risk of bad driving behavior of single vehicle |
| CN112837533A (en) * | 2021-01-08 | 2021-05-25 | 合肥工业大学 | Highway accident frequency prediction method considering risk factor time-varying characteristics |
| CN112990545A (en) * | 2021-02-08 | 2021-06-18 | 东南大学 | Traffic safety state prediction method for expressway intersection area |
| CN113689071A (en) * | 2021-07-15 | 2021-11-23 | 东南大学 | Multi-driver risk evaluation-based active intervention method for bad driving behaviors |
| CN113762734A (en) * | 2021-08-17 | 2021-12-07 | 淮阴工学院 | Dangerous chemical vehicle highway driving risk assessment method and system |
| KR102416711B1 (en) * | 2022-02-21 | 2022-07-05 | 주식회사 카비 | Accident risk prediction device for each driver using artificial intelligence and accident risk prediction method using the same |
Non-Patent Citations (1)
| Title |
|---|
| 李静;王占永;蔡铭;: "信号交叉口左转待行区综合效益评估", 中山大学学报(自然科学版), no. 03, 15 May 2019 (2019-05-15) * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117351708A (en) * | 2023-10-08 | 2024-01-05 | 北京迈道科技有限公司 | Expressway safety operation management early warning method, system and storage medium |
| CN117273964A (en) * | 2023-11-21 | 2023-12-22 | 国任财产保险股份有限公司 | Intelligent vehicle insurance policy generation system and method for self-adaptive driving data |
| CN117273964B (en) * | 2023-11-21 | 2024-02-02 | 国任财产保险股份有限公司 | Intelligent vehicle insurance policy generation system and method for self-adaptive driving data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110570651B (en) | Road network traffic situation prediction method and system based on deep learning | |
| CN116631186A (en) | Expressway traffic accident risk assessment method and system based on dangerous driving event data | |
| CN112613225B (en) | Intersection traffic state prediction method based on neural network cell transmission model | |
| CN111275962B (en) | Vehicle track data aggregation effect prediction method and device | |
| CN109345832B (en) | Urban road overtaking prediction method based on deep recurrent neural network | |
| CN112990545B (en) | Traffic safety state prediction method for expressway intersection area | |
| CN114664091A (en) | Early warning method and system based on holiday traffic prediction algorithm | |
| CN113051811B (en) | Multi-mode short-term traffic jam prediction method based on GRU network | |
| CN116504076A (en) | Expressway traffic flow prediction method based on ETC portal data | |
| CN115148019A (en) | Early warning method and system based on holiday congestion prediction algorithm | |
| CN111724592B (en) | Highway traffic jam detection method based on charging data and checkpoint data | |
| CN113361825A (en) | Early warning method and system for trampling accident | |
| CN114882069A (en) | Taxi track abnormity detection method based on LSTM network and attention mechanism | |
| CN104217591B (en) | Dynamic road condition detection method and system | |
| CN116168356B (en) | Vehicle damage judging method based on computer vision | |
| CN106157657A (en) | A kind of moving state identification system and method for mobile subscriber | |
| CN116946183A (en) | Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment | |
| CN115719475B (en) | Three-stage trackside equipment fault automatic detection method based on deep learning | |
| CN116386020A (en) | Method and system for predicting exit flow of highway toll station by multi-source data fusion | |
| CN112201036B (en) | Urban expressway travel speed short-time prediction method based on inclusion-CNN | |
| Liu | Refined Judgment of Urban Traffic State Based on Machine Learning and Edge Computing | |
| CN116401944B (en) | Intelligent traffic management system based on big data | |
| AU2021100003A4 (en) | A deep transportation model to predict the human mobility for autonomous vehicle | |
| CN115035715B (en) | Expressway flow prediction method based on decision tree and multi-element auxiliary information | |
| CN113345220B (en) | Highway inspection vehicle tracking and prediction analysis system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |