CN112990563A - Real-time prediction method for rear-end collision accident risk of expressway - Google Patents

Real-time prediction method for rear-end collision accident risk of expressway Download PDF

Info

Publication number
CN112990563A
CN112990563A CN202110245304.4A CN202110245304A CN112990563A CN 112990563 A CN112990563 A CN 112990563A CN 202110245304 A CN202110245304 A CN 202110245304A CN 112990563 A CN112990563 A CN 112990563A
Authority
CN
China
Prior art keywords
time
trend
sequence
end collision
target vehicle
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.)
Granted
Application number
CN202110245304.4A
Other languages
Chinese (zh)
Other versions
CN112990563B (en
Inventor
苑仁腾
项乔君
甘婧
王明远
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Original Assignee
Southeast University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN202110245304.4A priority Critical patent/CN112990563B/en
Publication of CN112990563A publication Critical patent/CN112990563A/en
Application granted granted Critical
Publication of CN112990563B publication Critical patent/CN112990563B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G06Q50/265—Personal security, identity or safety

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Marketing (AREA)
  • Development Economics (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Computer Security & Cryptography (AREA)
  • Educational Administration (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a real-time prediction method for the risk of a highway rear-end collision accident, which comprises the steps of respectively acquiring the target vehicle in a historical time period and a future preset time period
Figure DEST_PATH_IMAGE002
Acquiring each rear-end collision accident in a target road section within a preset time period by using a seasonal time trend sequence, a periodic time trend sequence and a residual error item sequence of the time sequence
Figure DEST_PATH_IMAGE004
Within a preset time period from the occurrence time to the historical time
Figure DEST_PATH_IMAGE006
Seasonal time trend series, periodic time trend series, residual error of time seriesA sequence of items; and based on a time warping algorithm, acquiring the maximum value in the seasonal time trend similarity parameters, the maximum value in the periodic time similarity parameters and the maximum value in the residual error item similarity parameters of each rear-end collision accident and the target vehicle, and acquiring the predicted value of the rear-end collision accident of the target vehicle by combining with a prediction model. The method provided by the invention has the advantages that the obtained prediction result is accurate, and the rear-end collision analysis can be predicted in real time.

Description

Real-time prediction method for rear-end collision accident risk of expressway
Technical Field
The invention relates to the technical field of safe driving of vehicles, in particular to a real-time prediction method for rear-end collision risks on a highway.
Background
While promoting economic development, the highway is more likely to have major traffic accidents compared with urban roads due to the characteristics of large flow, high speed, large speed difference of different vehicle types and the like. Rear-end collisions are the main form of accidents on highways. In the driving process of the vehicle, the traffic accident risk needs to be identified and predicted actively, in real time and non-centrally, and active risk avoidance measures are taken to avoid the occurrence of accidents. The traffic safety based on the number of accidents belongs to long-term posterior evaluation, the requirement on the number of samples is high, the interpretability of the accident reasons is poor, and active prevention and control cannot be carried out.
To solve this problem, scholars at home and abroad propose a concept of traffic conflict, in which two or more road users approach each other at the same time and space under an observable condition, and if one of the road users takes an abnormal traffic behavior, such as changing direction, changing speed of a vehicle, suddenly stopping the vehicle, and the like, the collision occurs unless the other road user also takes a corresponding behavior. The traffic conflict technology can observe a large amount of data before an accident occurs, has the statistical advantages of large samples, short period, small area and high reliability, and is considered to be one of the most promising research directions in the future traffic safety field. However, at present, the traffic collision technology at home and abroad is used for predicting the rear-end collision risk of roads and mainly focusing on macroscopic traffic evaluation of areas or road sections, and the following problems exist: effective information of traffic conflict is not fully considered, prediction is not real-time and non-centralized, the specific relation between traffic conflict and accident is not clear in the prediction process, and the reliability of the prediction result is low.
Disclosure of Invention
The purpose of the invention is as follows: the method has high accuracy of prediction results and can predict the risk of the highway rear-end collision accident in real time.
The technical scheme is as follows: the invention provides a real-time prediction method for a rear-end collision risk of a highway, which is used for predicting the rear-end collision risk of a target vehicle at a target moment on a target road section, and comprises the following steps:
step 1: obtaining preset times of target vehicle in historical time period
Figure BDA0002963882330000011
Value, in turn, of the target vehicle over a historical period of time n
Figure BDA0002963882330000012
A time series; the historical time period is a preset time period from the current time t to the historical time direction, and the historical time period is
Figure BDA0002963882330000013
The value is the inverse of the vehicle collision time;
to the same
Figure BDA0002963882330000014
Performing trend decomposition on the time sequence to acquire the target vehicle in the historical time period n
Figure BDA0002963882330000015
Seasonal time trend sequence S (t) of the time sequence, periodic time trend sequence C (t) and residual error item sequence Q (t); then entering step 2;
step 2: acquiring accident data of I rear-end collisions occurring in a target road section within a preset time period;
respectively aiming at each rear-end collision I, I is more than or equal to 1 and less than or equal to I, executing the steps from 2.1 to 2.2, and further acquiring seasonal time trend similarity parameters, periodic time similarity parameters and residual error item similarity parameters of each rear-end collision in the I rear-end collision accidents:
step 2.1: acquiring the time of the rear-end collision accident i within a preset time period towards the historical time direction
Figure BDA0002963882330000021
Time series, thereby obtaining the
Figure BDA0002963882330000022
Seasonal time trend series S of time seriesiPeriodic time trend series CiSequence of residual terms Qi；
Step 2.2: based on time warping algorithm, respectively comparing S (t) and SiC (t) and CiQ (t) and QiAnd (4) judging the similarity of the time sequences, and further acquiring the seasonal time trend similarity parameter of the rear-end collision accident i
Figure BDA0002963882330000023
Periodic time similarity parameter
Figure BDA0002963882330000024
Residual term similarity parameter
Figure BDA0002963882330000025
Entering the step 3;
and step 3: obtaining the maximum value of seasonal time trend similarity parameters of all rear-end accidents
Figure BDA0002963882330000026
Maximum value in periodic time similarity parameter
Figure BDA0002963882330000027
Maximum value in residual item similarity parameter
Figure BDA0002963882330000028
Entering the step 4;
and 4, step 4: with target vehicle at current time t
Figure BDA0002963882330000029
Seasonal time trend sequence S (t) of time series, periodic time trend sequence C (t) and residual error item sequence Q (t)) Training an ARMA model, and acquiring the future time period P of the target vehicle by using the trained ARMA model
Figure BDA00029638823300000210
Seasonal time trend sequence S ' (t) of the time sequence, periodic time trend sequence C ' (t), and residual item sequence Q ' (t) prediction values;
according to
Figure BDA00029638823300000211
And S ' (t), C ' (t) and Q ' (t), and the final predicted value of the risk of the rear-end collision accident of the target vehicle is obtained by combining the prediction model.
As a preferable scheme of the present invention, the S ' (t), C ' (t), Q ' (t) respectively include seasonal time trend values, periodic time trend values, and residual term values of the target vehicle at each time in the future time period P;
in step 4, the prediction model is:
Figure BDA0002963882330000031
wherein, bpThe predicted value of the rear-end collision risk of the target vehicle at the time P in the future time period P is obtained;
sp、cp、qprespectively a seasonal time trend value, a periodic time trend value and a residual error term value of the target vehicle at the moment p.
As a preferable aspect of the present invention, after step 4, the method further comprises:
and 5: judging the predicted value bpIf the current time is less than the preset early warning threshold value, early warning is carried out if the current time is less than the preset early warning threshold value; otherwise, returning to the step 1.
As a preferable scheme of the present invention, the early warning threshold is 4 s.
As a preferred aspect of the present invention, in step 4, the method further comprises training the ARMA model according to the following method:
respectively carrying out stationarity treatment on S (t), C (t) and Q (t) to obtain S (t), C (t) and Q (t) after smoothing;
and (3) training the ARMA model by taking the smoothed S (t), C (t) and Q (t) as a sample set.
As a preferable aspect of the present invention, in step 4, the method further includes: according to the following formula:
Figure BDA0002963882330000032
obtaining target vehicle at current time t
Figure BDA0002963882330000033
A value;
where Δ l (t) represents a distance difference between the target vehicle and its preceding adjacent vehicle at the current time t; Δ v (t) represents the speed difference between the target vehicle and its preceding neighboring vehicle at the present time t.
As a preferred aspect of the present invention, in step 2, when the number I of rear-end collisions occurring within the preset time period on the road section in the preset range where the target vehicle is located is equal to zero, the seasonal time similarity parameter
Figure BDA0002963882330000034
Periodic time similarity parameter
Figure BDA0002963882330000035
Residual term similarity parameter
Figure BDA0002963882330000036
The values of (A) are all 1.
Has the advantages that: compared with the prior art, the method provided by the invention utilizes a trend decomposition method pair
Figure BDA0002963882330000037
The method overcomes the defect that the TTC value does not exist when two vehicles run at the same speed, and is more consistent with the following of the same speedThe invention establishes the phenomenon of the vertical fluctuation of the speed of the vehicle after driving and the speed of the vehicle before driving, and the prediction result can be gradually updated along with the change of time
Figure BDA0002963882330000041
Calculating the actual detection by time warping algorithm
Figure BDA0002963882330000042
Time series and before accident
Figure BDA0002963882330000043
Compared with the traditional method, the method improves the reliability of the traffic conflict technology, establishes the relation between the conflict indexes and the traffic accidents and enables the prediction result to be more accurate. In addition, the method starts from the perspective of a single vehicle, the final result can be used for guiding and standardizing driving behaviors, the occurrence of rear-end accidents is reduced, and the driving safety is improved.
Drawings
FIG. 1 is a schematic diagram of a scene simulation provided in accordance with an embodiment of the present invention;
FIG. 2 is a block flow diagram of a method provided in accordance with an embodiment of the present invention;
FIG. 3 is a block diagram of a flow chart of a method for calculating a similarity parameter according to an embodiment of the present invention;
FIG. 4 is a schematic diagram illustrating a relative distance change between two vehicles according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a relative speed change between two vehicles according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating a variation trend of 1/TTC provided by an embodiment of the present invention;
FIG. 7 is a schematic diagram of a seasonal trend provided in accordance with an embodiment of the present invention;
FIG. 8 is a schematic diagram of a cyclical trend provided in accordance with an embodiment of the present invention;
FIG. 9 is a schematic diagram of residual items provided in accordance with an embodiment of the present invention;
FIG. 10 is a graphical illustration of seasonal time trend ARMA training results provided in accordance with an embodiment of the present invention;
FIG. 11 is a diagram illustrating the result of the ARMA training with periodic time trend according to an embodiment of the present invention;
FIG. 12 is a schematic diagram of the residual term ARMA training result provided in accordance with an embodiment of the present invention;
FIG. 13 is a graph comparing predicted results and actual results provided according to an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
The method provided by the invention comprises the following steps:
step 1: collecting data, calculating and acquiring TTC (time to collision) values of target vehicles at preset moments in historical time periods, and calculating corresponding TTC values
Figure BDA0002963882330000044
A value; the historical time period is a preset time period from the current time t to the historical time direction, and the historical time period is
Figure BDA0002963882330000045
The value is the inverse of the vehicle collision time.
Specifically, speed and position data of adjacent front vehicles in the same lane at each preset time in a historical time period n are collected through vehicle-mounted equipment; according to the following formula:
Figure BDA0002963882330000051
obtaining target vehicle at current time t
Figure BDA0002963882330000052
A value;
wherein lp(t) represents the position of the preceding vehicle; lf(t) represents the position of the following vehicle at time t; v. off(t) represents the speed of the rear vehicle; v. ofp(t) represents the speed of the preceding vehicle at time t; Δ l (t) represents a distance difference between the target vehicle and its preceding adjacent vehicle at the present time t; Δ v (t) represents the speed difference between the target vehicle and its preceding neighboring vehicle at the present time t.
In one embodiment, n is 5min, the interval between preset times is 1s, and the method is obtained
Figure BDA0002963882330000053
The values are numbered to construct values of the target vehicle over a historical period of time n
Figure BDA0002963882330000054
Time series b (t):
B(t)＝{b1(t),b2(t),b3(t),...,b60n(t)}
if n is 5min and the interval between the predetermined times is 1s, then 60n data are included in the time sequence b (t), where b is60n(t) indicating time t
Figure BDA0002963882330000055
A value, numbered 60n in time series; b1(t) at a time corresponding to the time series number 1
Figure BDA0002963882330000056
A value; b (t) is updated gradually with the lapse of time.
To the same
Figure BDA0002963882330000057
Performing trend decomposition on the time sequence B (t) to obtain the historical time period n of the target vehicle
Figure BDA0002963882330000058
Seasonal time trend sequence S (t) of the time sequence, periodic time trend sequence C (t) and residual error item sequence Q (t); then step 2 is entered.
The method for performing trend decomposition on B (t) is as follows:
B(t)＝S(t)+C(t)+Q(t)
in the formula: s (t) { s ═ s1(t),…,s60n(t) } denotes historical statistics at time t, for a period of time prior to time t
Figure BDA0002963882330000059
The seasonal time trend of the sequence is such that,
Figure BDA00029638823300000510
respectively representing seasonal time trend values at each moment; c (t) { c1(t),…,c60n(t) } represents a periodic time trend; c. C1(t),...,c60n(t) respectively representing seasonal time trend values at respective times; q (n) { q ═ q1(t),…,q60n(t) } denotes the residual term, q1(t),...,q60n(t) represents the residual terms at each time instant.
Step 2: acquiring accident data of I rear-end collisions occurring in a target road section within a preset time period;
respectively aiming at each rear-end collision I, I is more than or equal to 1 and less than or equal to I, executing the steps from 2.1 to 2.2, and further acquiring seasonal time trend similarity parameters, periodic time similarity parameters and residual error item similarity parameters of each rear-end collision in the I rear-end collision accidents:
step 2.1: acquiring the time of the rear-end collision accident i within a preset time period towards the historical time direction
Figure BDA0002963882330000061
Time series, thereby obtaining the
Figure BDA0002963882330000062
Seasonal time trend series S of time seriesiPeriodic time trend series CiSequence of residual terms Qi；
Step 2.2: based on time warping algorithm, respectively comparing S (t) and SiC (t) and CiQ (t) and QiWhen in progressJudging the similarity of the sequences, and further acquiring the seasonal time trend similarity parameter of the rear-end collision accident i
Figure BDA0002963882330000063
Periodic time similarity parameter
Figure BDA0002963882330000064
Residual term similarity parameter
Figure BDA0002963882330000065
When the number I of rear-end accidents occurring on the target road section in the preset time period is equal to zero, the seasonal time similarity parameter
Figure BDA0002963882330000066
Periodic time similarity parameter
Figure BDA0002963882330000067
Residual term similarity parameter
Figure BDA0002963882330000068
The values of (A) are all 1.
With 1s as a time interval, establishing a database A of TTC time sequence values within 3min before each rear-end collision by using the statistical data of the rear-end collision accidents on the highway: a ═ A1,A2，....,Am}
Wherein A isi＝{ai1,ai2,….,aim}T，AiIndicates that the ith rear-end collision occurs within 3min before the occurrence of the rear-end collision
Figure BDA0002963882330000069
A time series value, wherein i is more than or equal to 0; m is a time series number; a isimIndicating m seconds before the occurrence of accident i
Figure BDA00029638823300000610
The value is obtained.
To AiPerforming trend decomposition:
Ai＝Si+Ci+Qi
wherein S isiIndicating the ith rear-end collision
Figure BDA00029638823300000611
Seasonal time trends of time series value decomposition; ciRepresenting a periodic time trend after decomposition, QiRepresenting the residual terms after decomposition.
Using a time warping algorithm (DTW) on SiAnd S (t), CiAnd C (t), QiAnd Q (t) performing time sequence similarity judgment to respectively obtain similarity parameters
Figure BDA00029638823300000612
And
Figure BDA00029638823300000613
and in the calculation process of the regularization algorithm (DTW), the obtained Euclidean distance matrix is subjected to standardization processing based on a min-max criterion.
And step 3: obtaining the maximum value of seasonal time trend similarity parameters of all rear-end accidents
Figure BDA0002963882330000071
Maximum value in periodic time similarity parameter
Figure BDA0002963882330000072
Maximum value in residual item similarity parameter
Figure BDA0002963882330000073
Entering the step 4;
and 4, step 4: with target vehicle at current time t
Figure BDA0002963882330000074
The seasonal time trend sequence S (t), the periodic time trend sequence C (t) and the residual error item sequence Q (t) of the time sequence train an ARMA model to obtain the time period P of the target vehicle in the future
Figure BDA0002963882330000075
Seasonal time trend series S ' (t) of the time series, periodic time trend series C ' (t), residual error term series Q ' (t).
The trained ARMA model was obtained according to the following method:
respectively carrying out stationarity treatment on the S (t), the C (t) and the Q (t) acquired in the step 1 by using a moving average method or a difference method to acquire S (t), C (t) and Q (t) after smoothing;
and (4) taking the smoothed S (t), C (t) and Q (t) as sample sets, respectively training the ARMA model to determine model parameters, and further acquiring the trained ARMA model.
And (3) predicting trend change values S ' (t), C ' (t) and Q ' (t) in the future 15S after the current time t respectively based on the trained ARMA model:
S'(t)＝{s60n+1(t),...,s60n+15(t)}
C'(t)＝{c60n+1(t),...,c60n+15(t)}
Q'(t)＝{q1(t),...,q60n+15(t)}
s ' (t), C ' (t) and Q ' (t) respectively comprise seasonal time trend values, periodic time trend values and residual error term values of the target vehicle at each moment in the future time period P;
according to
Figure BDA0002963882330000076
And S ' (t), C ' (t) and Q ' (t), and the predicted value b of the target vehicle in the rear-end collision accident is obtained by combining the prediction modelp。
The prediction model is as follows:
Figure BDA0002963882330000081
wherein, b60n+iIndicates the predicted value of the ith second, i is more than or equal to 1.
For prediction b60n+iReciprocal is removed to obtain the TTC value of the corresponding time, and the TTC value is expressed as TTC60n+i(ii) a When in useTTC60n+iAnd when the value is less than the fixed threshold value, the alarm indicating lamp is lightened to remind the driver of driving cautiously. In one embodiment, the pre-warning threshold is 4 s.
Taking the corresponding behavior of a car following a highway in a city as an example, a scene schematic diagram is shown in fig. 1, and the method for predicting the risk of the car rear-end collision accident on the highway in real time comprises the following steps, and related flow chart diagrams are shown in fig. 2 and fig. 3.
Acquiring data to obtain TTC sequence values within 15 minutes, and performing trend decomposition, wherein the method comprises the following steps:
by using vehicle-mounted radar and GPS positioning technology, the speed and the position between two adjacent vehicles can be obtained by using a formula delta v (t) ═ vf(t)-vp(t) and Δ l (t) ═ lf(t)-lp(t) a sequence of time-varying vehicles Δ v and Δ l may be obtained 900s before the current time, as shown in fig. 4 and 5.
The correspondence can be calculated by using Δ v/Δ l
Figure BDA0002963882330000082
Value, sequence B, made within 15 minutes
Figure BDA0002963882330000083
The trend over time is shown in fig. 6.
Performing trend decomposition on the time sequence B, and performing stationarity treatment by using a moving average method; drawing a seasonal time trend graph and a periodic time trend graph; residual error item maps are remained; as shown in fig. 7, 8 and 9, respectively.
The ARMA model is trained by seasonal time trend, periodic time trend and residual error items respectively to determine the parameters of the ARMA, and the training results are shown in FIG. 10, FIG. 11 and FIG. 12.
And (4) predicting values within 15s of the future of each trend by using the trained ARMA model, and taking the reciprocal of the values.
And (II) establishing a database of the time sequence value of the front 3minTTC before the collision by using the statistical data of the rear-end collision accidents on the highway, and performing trend decomposition.
(III) calculating actual acquisition
Figure BDA0002963882330000084
Time series and collision occurrence
Figure BDA0002963882330000085
The TTC prediction method is established by the similarity of sequences, and comprises the following steps:
computational collection using time warping algorithm
Figure BDA0002963882330000086
Before the time sequence and collision occur
Figure BDA0002963882330000087
Similarity of sequences;
screening to obtain
Figure BDA0002963882330000088
Corresponding similarity parameter
Figure BDA0002963882330000089
Weighting the trend value obtained in the step (one) to obtain a final prediction result; taking the reciprocal of the prediction to obtain a TTC prediction value; and when the predicted value of the TTC is smaller than the fixed threshold value, the alarm indicating lamp is turned on to remind the driver of driving cautiously.
The predicted results are plotted against the actual results, as shown in fig. 13.
The method provided by the invention can obtain the TTC time sequence value in real time by utilizing the vehicle-mounted equipment, and utilizes a trend decomposition method to correct
Figure BDA0002963882330000091
The method overcomes the defect that the TTC value does not exist when two vehicles run at the same speed, better conforms to the phenomena of the speed fluctuation of the following vehicles and the speed of the front vehicle, the prediction result can be gradually updated along with the change of time, and the essential reason of the rear-end collision accident is analyzed. The inventionBased on the historical statistical data of rear-end collision accidents, the method establishes the pre-collision time of the rear-end collision accidents
Figure BDA0002963882330000092
Calculating the actual detection by DTW algorithm
Figure BDA0002963882330000093
Time series and before accident
Figure BDA0002963882330000094
Compared with the traditional method, the method improves the reliability of the traffic conflict technology, establishes the relation between the conflict indexes and the traffic accidents and enables the prediction result to be more accurate. In addition, the method starts from the perspective of a single vehicle, the final result can be used for guiding and standardizing driving behaviors, the occurrence of rear-end accidents is reduced, and the driving safety is improved.
The above description is only a preferred embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be considered as the protection scope of the present invention.

Claims (7)

1. A real-time prediction method for the risk of the highway rear-end collision accident is characterized by being used for predicting the risk of the rear-end collision accident of a target vehicle at a target moment on a target road section, and comprising the following steps:
step 1: obtaining preset times of target vehicle in historical time period
Figure FDA0002963882320000011
Values to build the target vehicle's history of time periods
Figure FDA0002963882320000012
A time series; the historical time period is a preset time period from the current time t to the historical time direction, and the historical time period is
Figure FDA0002963882320000013
The value is the inverse of the vehicle collision time;
to the same
Figure FDA0002963882320000014
Performing trend decomposition on the time sequence to acquire the historical time period of the target vehicle
Figure FDA0002963882320000015
Seasonal time trend sequence S (t) of the time sequence, periodic time trend sequence C (t) and residual error item sequence Q (t); then entering step 2;
step 2: acquiring accident data of I rear-end collisions occurring in a target road section within a preset time period;
respectively aiming at each rear-end collision I, I is more than or equal to 1 and less than or equal to I, executing the steps from 2.1 to 2.2, and further acquiring seasonal time trend similarity parameters, periodic time similarity parameters and residual error item similarity parameters of each rear-end collision in the I rear-end collision accidents:
step 2.1: acquiring the time of the rear-end collision accident i within a preset time period towards the historical time direction
Figure FDA0002963882320000016
Time series, thereby obtaining the
Figure FDA0002963882320000017
Seasonal time trend series S of time seriesiPeriodic time trend series CiSequence of residual terms Qi；
Step 2.2: based on time warping algorithm, respectively comparing S (t) and SiC (t) and CiQ (t) and QiAnd (4) judging the similarity of the time sequences, and further acquiring the seasonal time trend similarity parameter of the rear-end collision accident i
Figure FDA0002963882320000018
Periodic timeSimilarity parameter
Figure FDA0002963882320000019
Residual term similarity parameter
Figure FDA00029638823200000110
Entering the step 3;
and step 3: obtaining the maximum value of seasonal time trend similarity parameters of all rear-end accidents
Figure FDA00029638823200000111
Maximum value in periodic time similarity parameter
Figure FDA00029638823200000112
Maximum value in residual item similarity parameter
Figure FDA00029638823200000113
Entering the step 4;
and 4, step 4: with target vehicle at current time t
Figure FDA00029638823200000114
The seasonal time trend sequence S (t), the periodic time trend sequence C (t) and the residual error item sequence Q (t) of the time sequence are used for training an ARMA model, and the trained ARMA model is used for acquiring the time period P of the target vehicle in the future
Figure FDA00029638823200000115
Seasonal time trend sequence S ' (t) of the time sequence, periodic time trend sequence C ' (t), and residual item sequence Q ' (t) prediction values;
according to
Figure FDA0002963882320000021
And S ' (t), C ' (t) and Q ' (t), and the final predicted value of the risk of the rear-end collision accident of the target vehicle is obtained by combining the prediction model.
2. The real-time prediction method for the risk of the highway rear-end collision accident according to claim 1, wherein the S ' (t), C ' (t) and Q ' (t) respectively comprise seasonal time trend values, periodic time trend values and residual error term values of the target vehicle at each moment in the future time period P;
in step 4, the prediction model is:
Figure FDA0002963882320000022
wherein, bpThe predicted value of the rear-end collision accident of the target vehicle at the time P in the future time period P is obtained;
sp、cp、qprespectively a seasonal time trend value, a periodic time trend value and a residual error term value of the target vehicle at the moment p.
3. The real-time highway rear-end accident risk prediction method according to claim 2, wherein after step 4, said method further comprises:
and 5: judging the predicted value bpIf the current time is less than the preset early warning threshold value, early warning is carried out if the current time is less than the preset early warning threshold value; otherwise, returning to the step 1.
4. The method according to claim 3, wherein the early warning threshold is 4 s.
5. The method for real-time prediction of risk of a highway rear-end collision accident according to claim 1, wherein in step 4, the method further comprises training an ARMA model according to the following method:
respectively carrying out stationarity treatment on S (t), C (t) and Q (t) to obtain S (t), C (t) and Q (t) after smoothing;
and (3) training the ARMA model by taking the smoothed S (t), C (t) and Q (t) as a sample set.
6. The real-time prediction method of the risk of a highway rear-end collision accident according to claim 1, wherein in step 4, the method further comprises the following formula:
Figure FDA0002963882320000023
obtaining target vehicle at current time t
Figure FDA0002963882320000024
A value;
where Δ l (t) represents a distance difference between the target vehicle and its preceding adjacent vehicle at the current time t; Δ v (t) represents the speed difference between the target vehicle and its preceding neighboring vehicle at the present time t.
7. The method according to claim 1, wherein in step 2, when the number I of rear-end collisions occurring in the road section of the preset range in which the target vehicle is located within the preset time period is equal to zero, the seasonal time similarity parameter is set to zero
Figure FDA0002963882320000031
Periodic time similarity parameter
Figure FDA0002963882320000032
Residual term similarity parameter
Figure FDA0002963882320000033
The values of (A) are all 1.
CN202110245304.4A 2021-03-05 2021-03-05 Real-time prediction method for rear-end collision accident risk of expressway Active CN112990563B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110245304.4A CN112990563B (en) 2021-03-05 2021-03-05 Real-time prediction method for rear-end collision accident risk of expressway

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110245304.4A CN112990563B (en) 2021-03-05 2021-03-05 Real-time prediction method for rear-end collision accident risk of expressway

Publications (2)

Publication Number Publication Date
CN112990563A true CN112990563A (en) 2021-06-18
CN112990563B CN112990563B (en) 2022-11-18

Family

ID=76353073

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110245304.4A Active CN112990563B (en) 2021-03-05 2021-03-05 Real-time prediction method for rear-end collision accident risk of expressway

Country Status (1)

Country Link
CN (1) CN112990563B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113610279A (en) * 2021-07-20 2021-11-05 中国石油大学(华东) Accident prediction method based on data set regularity
CN117874437A (en) * 2024-03-12 2024-04-12 广东电网有限责任公司 Accident trend identification method based on time sequence analysis

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103150930A (en) * 2013-02-01 2013-06-12 东南大学 Rear-end collision real-time prediction method aimed at frequently jammed section of expressway
CN111310981A (en) * 2020-01-20 2020-06-19 浙江工业大学 Reservoir water level trend prediction method based on time series
CN111695791A (en) * 2020-05-29 2020-09-22 腾讯科技（深圳）有限公司 Service index prediction method and device

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103150930A (en) * 2013-02-01 2013-06-12 东南大学 Rear-end collision real-time prediction method aimed at frequently jammed section of expressway
CN111310981A (en) * 2020-01-20 2020-06-19 浙江工业大学 Reservoir water level trend prediction method based on time series
CN111695791A (en) * 2020-05-29 2020-09-22 腾讯科技（深圳）有限公司 Service index prediction method and device

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113610279A (en) * 2021-07-20 2021-11-05 中国石油大学(华东) Accident prediction method based on data set regularity
CN117874437A (en) * 2024-03-12 2024-04-12 广东电网有限责任公司 Accident trend identification method based on time sequence analysis
CN117874437B (en) * 2024-03-12 2024-05-10 广东电网有限责任公司 Accident trend identification method based on time sequence analysis

Also Published As

Publication number Publication date
CN112990563B (en) 2022-11-18

Similar Documents

Publication Publication Date Title
CN107742193B (en) A kind of driving Risk Forecast Method based on time-varying state transition probability Markov chain
Berndt et al. Continuous driver intention recognition with hidden markov models
US20190155291A1 (en) Methods and systems for automated driving system simulation, validation, and implementation
CN110949398B (en) Method for detecting abnormal driving behavior of first-vehicle drivers in vehicle formation driving
CN110807436B (en) Dangerous driving behavior recognition and dangerous event prediction method, device and storage medium
EP2610836B1 (en) Device and method for the on-line prediction of the driving cycle in an automotive vehicle
CN107958269A (en) A kind of driving risk factor Forecasting Methodology based on hidden Markov model
CN111079590A (en) Peripheral vehicle behavior pre-judging method of unmanned vehicle
CN109733390A (en) A kind of adaptive lane-change method for early warning based on driver's characteristic
CN108877213B (en) Traffic accident identification method and system
CN110077398B (en) Risk handling method for intelligent driving
CN112990563B (en) Real-time prediction method for rear-end collision accident risk of expressway
CN113190921B (en) Automatic evaluation method and system for intelligent automobile driving performance test
CN111243338A (en) Vehicle acceleration-based collision risk evaluation method
CN112258841A (en) Intelligent vehicle risk assessment method based on vehicle track prediction
CN112991685A (en) Traffic system risk assessment and early warning method considering fatigue state influence of driver
CN113436432A (en) Method for predicting short-term traffic risk of road section by using road side observation data
CN109118787A (en) A kind of car speed prediction technique based on deep neural network
CN112560782A (en) Vehicle lane changing behavior identification method based on random forest algorithm
CN115630569A (en) Millimeter wave radar sensor modeling method applied to automatic driving
CN112863245A (en) Vehicle track change real-time prediction method based on deep neural network
CN116946183A (en) Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment
CN110610611B (en) Driving safety evaluation method for intelligent network-connected vehicle in mixed-driving traffic flow
CN113252057A (en) Method and system for identifying driving tendency based on high altitude navigation data
Yu et al. Drivers' car-following correlative behavior with preceding vehicles in multilane driving

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
GR01 Patent grant
GR01 Patent grant