CN108198415A - A kind of city expressway accident forecast method based on deep learning - Google Patents

A kind of city expressway accident forecast method based on deep learning Download PDF

Info

Publication number
CN108198415A
CN108198415A CN201711461082.XA CN201711461082A CN108198415A CN 108198415 A CN108198415 A CN 108198415A CN 201711461082 A CN201711461082 A CN 201711461082A CN 108198415 A CN108198415 A CN 108198415A
Authority
CN
China
Prior art keywords
accident
layer
traffic
section
flow
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
CN201711461082.XA
Other languages
Chinese (zh)
Other versions
CN108198415B (en
Inventor
杨奎
王雪松
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.)
Tongji University
Original Assignee
Tongji 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 Tongji University filed Critical Tongji University
Priority to CN201711461082.XA priority Critical patent/CN108198415B/en
Publication of CN108198415A publication Critical patent/CN108198415A/en
Application granted granted Critical
Publication of CN108198415B publication Critical patent/CN108198415B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • 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
    • G08G1/0133—Traffic data processing for classifying traffic situation
    • 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

Abstract

The present invention relates to a kind of city expressway accident forecast method based on deep learning, applied to intelligent transportation and active field of traffic control.The telecommunication flow information of accident section and upstream and downstream adjacent segments in 5 30 minutes before being occurred by obtaining urban freeway network accident, and acquire the telecommunication flow information in adjacent three sections of upstream and downstream in lower 25 minutes of non-accident condition, to build traffic flow bulk sample notebook data, city expressway accident risk assessment models are established using deep neural network technology, will arithmetic for real-time traffic flow characteristic parameter substitute into accident risk assessment models in calculate accident risk, according to compared with the danger threshold of setting so as to discriminate whether that traffic accident can occur.Institute's inventive method uses the deep neural network technology with powerful learning ability to realize that accident risk is predicted in real time, and with higher accuracy and reliability, overcome conventional method rely on sampling samples modeling and caused by shortcomings and deficiencies.

Description

A kind of city expressway accident forecast method based on deep learning
Technical field
The present invention relates to intelligent transportation and active field of traffic control, more particularly to a kind of city based on deep learning is fast Fast road accident forecast method.
Background technology
Through street has both the dual function of highway and urban arterial road, in mainly being undertaken in urban road system grow away from From quick trip requirements, there is large capacity, high speed limit, be the aorta of City road traffic system.City is fast Fast road traffic accident is multiple, it is big that traffic circulation is influenced；Sporadic congestion accounts for city expressway congestion as caused by traffic accident 50%-75%.The genesis mechanism of city expressway accident is explored, improves traffic safety status and is had become with traffic circulation efficiency The key subjects faced by vehicle supervision department.
To improve urban road to the management level of traffic accident, the American-European active traffic administration system of metropolis active adoption (Active Traffic Management System) is united to the real-time safety management of urban road progress.Active traffic Management system is based on traffic flow sensory perceptual system, passes through variable speed-limit (Variable Speed Limits), congestion warning The control means such as (Queue Warning) and ring road metering (Ramp Metering), realize the emergency evacuation of urban road, reach To road network safety, unobstructed maximization.
Invention content
The purpose of the present invention is：A kind of city expressway accident forecast method based on deep learning.By obtaining city The telecommunication flow information of accident section and upstream and downstream adjacent segments in 5-30 minutes before the generation of through street net accident, and acquire non-thing Therefore in lower 25 minutes of state adjacent three sections of upstream and downstream telecommunication flow information, to build traffic flow bulk sample notebook data, use depth Degree nerual network technique establishes city expressway accident risk assessment models, and arithmetic for real-time traffic flow characteristic parameter is substituted into accident wind Calculate accident risk in dangerous assessment models, according to compared with the danger threshold of setting so as to discriminate whether that traffic accident can occur, To carry out active traffic administration.
The technical solution adopted in the present invention is：
A kind of city expressway accident forecast method based on deep learning, step are as follows：
Step 1：Obtain before urban freeway network accident occurs accident section and upstream and downstream adjacent segments in 5-30 minutes Telecommunication flow information.It is divided within 5-30 minutes before accident is occurred 55 minutes segments, section, adjacent occurs for acquisition accident Upstream section, three, adjacent downstream section section are in the average vehicle speed (AS) of 5 time slices, the averagely stream per track It measures (AF), is average per lane occupancy ratio (AO), car speed standard deviation (SS), flow standard poor (SF), occupation rate standard deviation (SO), i.e. 6*5*3=90 traffic characteristic variable in total.
Step 2：Acquire the telecommunication flow information in adjacent three sections of upstream and downstream in lower 25 minutes of non-accident condition.Whole day is drawn It is divided into 288 5 minutes segments (60min*24h/5min=288), after the influence for excluding accident, for every section, acquisition connects The average vehicle speed (AS) in continuous 5 adjacent three sections in 5 minutes, it is average per track flow (AF), average occupy per track Rate (AO), car speed standard deviation (SS), flow standard poor (SF), occupation rate standard deviation (SO), i.e., 6*5*3=90 in total Traffic characteristic variable.If certain section whole day zero defects occurs, there are 288 non-accident samples.
Step 3：Build traffic flow bulk sample notebook data.According to identical traffic characteristic variable, by the accident condition in step 1 Under telecommunication flow information merge with the telecommunication flow information under accident condition non-in step 2, structure city expressway accident forecast Full sample data set.
Step 4：City expressway accident risk assessment models are established using deep neural network technology.Deep neural network It is made of input layer, hidden layer and output layer.It connects entirely between layers, i.e. any one neuron of preceding layer Centainly it is connected with any one neuron of later layer.Wherein, input layer can receive input data values, i.e. 90 traffic characteristics Variable X={ x1,x2,x3,…,x90}.These data values are propagated forward to deep neural network middle layer and (also commonly referred to as hide Layer, number of plies n >=2) neuron in.Hidden layer is gradually propagated forward to output layer, which can finally present deep to user The output of neural network is spent as a result, i.e. accident (" 1 ") or non-accident (" 0 ").Each layer can be using the output of preceding layer as function Input, brings into the activation primitive of this layer.
The activation primitive,
IfFor the deviation of l j-th of neuron of layer,For the activation value (Activation) of l j-th of neuron of layer, Then
Wherein summation is what is carried out on l-1 layers of all neurons.Each layer of l defines a weight (Weight) Matrix wl, it is in the element of jth row kth rowσ () is nonlinear activation function, such as Maxout activation primitives.
The Maxout activation primitives,
I.e. the expression of Maxout activation primitives asks some neuron of the layer (such as l layers) to correspond to last layer (l-1 layers) The activation value of all neurons is (i.e.) maximum value.
The hidden layer,
The number of plies of hidden layer and each layer of neuron number can be set according to actual conditions or using loss function (Loss Functions optimal network structure) is selected.
The loss function,
Wherein L (i), t(i), O(i)Represent respectively the corresponding penalty values of i-th of sample, output predicted value (i.e. target), True output.Y represents the neuron of output layer, and O represents output layer.
Step 5：Arithmetic for real-time traffic flow characteristic parameter is substituted into step 4 accident risk assessment models and calculates accident risk.It will Traffic characteristic parameters X={ the x currently acquired1,x2,x3,…,x90Substitute into deep neural network model, calculate accident Risk.
Step 6：Judge whether that traffic accident can occur.By the accident occurrence risk of prediction and the danger threshold ratio of setting Compared with so as to discriminate whether to occur traffic accident, output accident (" 1 ") or non-accident (" 0 ").
It is an advantage of the invention that：
The present invention proposes a kind of city expressway accident forecast method based on deep learning.It the advantage is that：1) lead to The telecommunication flow information in adjacent three sections of accident before acquisition accident occurs is crossed, builds the data set of bulk sample sheet rather than sampling sample This, information is more complete, accurate；2) city expressway accident risk assessment models are established using deep neural network technology, it will Arithmetic for real-time traffic flow characteristic parameter substitutes into accident risk assessment models and calculates accident risk, compares traditional technology model, learns Habit ability is stronger, precision higher, can handle bulk sample this big data.
Description of the drawings
Fig. 1 is the flow chart of Forecasting Methodology of the present invention.
Fig. 2 is the structure (totally 7 layers) for the deep neural network that embodiment contains 5 hidden layers.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings, and step is as follows：
Step 1：Obtain before urban freeway network accident occurs accident section and upstream and downstream adjacent segments in 5-30 minutes Telecommunication flow information.It is divided within 5-30 minutes before accident is occurred 55 minutes segments, section, adjacent occurs for acquisition accident Upstream section, three, adjacent downstream section section 5 time slices average vehicle speed (Average Speed, AS), it is average per track flow (Average Flow, AF), it is average per lane occupancy ratio (Average Occupancy, AO), Car speed standard deviation (Standard Deviation of Speed, SS), poor (the Standard Deviation of flow standard Of Flow, SF), occupation rate standard deviation (Standard Deviation of Occupancy, SO), i.e., 6*5*3=90 in total A traffic characteristic variable.
Step 2：Acquire the telecommunication flow information in adjacent three sections of upstream and downstream in lower 25 minutes of non-accident condition.Whole day is drawn It is divided into 288 5 minutes segments, after the influence for excluding accident, for every section, continuous 5 of acquisition is adjacent three in 5 minutes The average vehicle speed (AS) in section, average every track flow (AF), average every lane occupancy ratio (AO), car speed standard Poor (SS), flow standard poor (SF), occupation rate standard deviation (SO), i.e. 6*5*3=90 traffic characteristic variable in total.
Step 3：Build traffic flow bulk sample notebook data.According to identical traffic characteristic variable, by the accident condition in step 1 Under telecommunication flow information merge with the telecommunication flow information under accident condition non-in step 2, structure city expressway accident forecast Full sample data set.
Step 4：City expressway accident risk assessment models are established using deep neural network technology.Deep neural network (Deep Neural Network, DNN) is made of input layer, hidden layer and output layer, the depth god of structure chart such as Fig. 2 Through network, containing 5 hidden layers, 7 layers altogether.By 90 traffic characteristic variable X={ x1,x2,x3,…,x90As input layer Receive input data values.Each layer can be brought into using the output of preceding layer as the input of function in the activation primitive of this layer.
Maxout activation primitives are selected as activation primitive, then the activation primitive,
WhereinFor the deviation of l j-th of neuron of layer,Activation value (Activatio for l j-th of neuron of layer N), then.Summation is carried out on l-1 layers of all neurons.Each layer of l defines weight (Weight) matrix wl, it is in the element of jth row kth row
The number of plies of hidden layer and each layer of neuron number can be set according to actual conditions or using loss function (Loss Functions it is optimal so as to select) to compare different network structures.
Step 5：Arithmetic for real-time traffic flow characteristic parameter is substituted into step 4 accident risk assessment models and calculates accident risk.It will Traffic characteristic parameters X={ the x of current acquisition1,x2,x3,…,x90Substitute into deep neural network model, calculate accident hair Raw risk.
Step 6：Judge whether that traffic accident can occur.By the accident occurrence risk of prediction and the danger threshold ratio of setting Compared with so as to discriminate whether to occur traffic accident, output accident (" 1 ") or non-accident (" 0 ").
Embodiment
Utilize north and south viaduct, the true traffic flow parameter of inner ring road and accident number in the city expressway of Shanghai City, China According to application and test technical solution of the present invention.
Step 1 according to the present invention builds traffic flow bulk sample notebook data to step 3.It draws within 5-30 minutes before accident is occurred Be divided into 55 minutes segments, acquisition accident occur section, adjacent upstream section, three, adjacent downstream section section this 5 The average vehicle speed (AS) of a time slice, average every track flow (AF), average every lane occupancy ratio (AO), vehicle speed Spend standard deviation (SS), flow standard poor (SF), occupation rate standard deviation (SO).Whole day is divided into 288 5 minutes segments (60min*24h/5min=288), after the influence for excluding accident, for every section, continuous 5 of acquisition is adjacent three in 5 minutes The average vehicle speed (AS) in a section, average every track flow (AF), average every lane occupancy ratio (AO), car speed mark Accurate poor (SS), flow standard poor (SF), occupation rate standard deviation (SO).The bulk sample of final one month originally has 612 accidents and 597, 093 its accident.Preceding half a month is divided into training sample, including 277 accidents and 306,831 non-accidents；Half a month conduct afterwards Test samples, including 335 accidents and 290,262 non-accidents.The former is used to train deep neural network model, Hou Zheyong In the precision of prediction of test depth neural network model.
Combined training collection, it is pre- that the step 4 according to the present invention establishes the calibrated accident risk based on deep neural network Survey (totally 7 layers of model：5 hidden layers, input layer, output layers), the parameter of the deep neural network is as shown in table 1.
Table 1
Using the precision of test samples test model, area (AUC value) under the receiver operator curve of model is found Up to 0.928, illustrate that model accuracy is fabulous.When rate of false alarm is 5.0%, 63.0% accident can be correctly identified, it is whole to predict Precision is 94.9%；When rate of false alarm is 10.1%, 79.4% accident can be correctly identified, whole precision of prediction is 89.8%. Therefore, the present invention can be based on full sample data set, improve the precision of prediction of accident.The present invention has practical engineering application valency Value.

Claims (2)

  1. A kind of 1. city expressway accident forecast method based on deep learning, which is characterized in that step is as follows：
    Step 1：Obtain the traffic of accident section and upstream and downstream adjacent segments in 5-30 minutes before urban freeway network accident occurs Stream information.55 minutes segments are divided within 5-30 minutes before accident is occurred, section, adjacent upstream road occur for acquisition accident Section, three, adjacent downstream section section 5 time slices average vehicle speed (Average Speed, AS), average every Track flow (Average Flow, AF), average every lane occupancy ratio (Average Occupancy, AO), car speed standard Poor (Standard Deviation of Speed, SS), flow standard poor (Standard Deviation of Flow, SF), Occupation rate standard deviation (Standard Deviation of Occupancy, SO), i.e., 6*5*3=90 traffic characteristic becomes in total Amount；
    Step 2：Acquire the telecommunication flow information in adjacent three sections of upstream and downstream in lower 25 minutes of non-accident condition.Whole day is divided into 288 5 minutes segments (60min*24h/5min=288), after the influence for excluding accident, for every section, continuous 5 of acquisition The average vehicle speed (AS) in adjacent three sections, average every track flow (AF), average every lane occupancy ratio in 5 minutes (AO), car speed standard deviation (SS), flow standard poor (SF), occupation rate standard deviation (SO), i.e. 6*5*3=90 traffic in total Characteristic variable.If certain section whole day zero defects occurs, there are 288 non-accident samples；
    Step 3：Build traffic flow bulk sample notebook data
    According to identical traffic characteristic variable, by non-accident shape in the telecommunication flow information under the accident condition in step 1 and step 2 Telecommunication flow information under state, which corresponds to, to be merged, and builds the full sample data set of city expressway accident forecast；
    Step 4：City expressway accident risk assessment models are established using deep neural network technology
    Deep neural network (Deep Neural Network, DNN) is made of, layer and layer input layer, hidden layer and output layer Between connect entirely, i.e., any one neuron of preceding layer is centainly connected with any one neuron of later layer, wherein, Input layer can receive input data values, i.e. 90 traffic characteristic variable X={ x1,x2,x3,…,x90, to biography before these data values It is multicast in the neuron of deep neural network middle layer (also commonly referred to as hidden layer, number of plies n >=2)；The gradual propagated forward of hidden layer To output layer, which finally can be presented the output of deep neural network as a result, i.e. accident (" 1 ") or non-accident to user (“0”)；Each layer can be brought into using the output of preceding layer as the input of function in the activation primitive of this layer；
    The activation primitive,
    IfFor the deviation of l j-th of neuron of layer,For the activation value (Activation) of l j-th of neuron of layer, then
    Wherein summation is what is carried out on l-1 layers of all neurons.Each layer of l defines weight (Weight) matrix wl, it is in the element of jth row kth rowσ () is nonlinear activation function, is Maxout activation primitives；
    The Maxout activation primitives,
    I.e. the expression of Maxout activation primitives asks some neuron of the layer (such as l layers) to correspond to all of last layer (l-1 layers) The activation value of neuron is (i.e.) maximum value；
    The hidden layer,
    The number of plies of hidden layer and each layer of neuron number can be set according to actual conditions or using loss function (Loss Functions optimal network structure) is selected.
    The loss function is：
    Wherein L (i), t(i), O(i)The corresponding penalty values of i-th of sample are represented respectively, export predicted value (i.e. target), is true defeated Go out value.Y represents the neuron of output layer, and O represents output layer；
    Step 5：Arithmetic for real-time traffic flow characteristic parameter is substituted into step 4 accident risk assessment models and calculates accident risk
    Traffic characteristic parameters X={ the x that will currently acquire1,x2,x3,…,x90Substitute into deep neural network model, calculate accident Occurrence risk；
    Step 6：Judge whether that traffic accident can occur
    By the accident occurrence risk of prediction compared with the danger threshold set, so as to discriminate whether to occur traffic accident, output Accident (" 1 ") or non-accident (" 0 ").
  2. 2. the city expressway accident forecast method described in accordance with the claim 1 based on deep learning, which is characterized in that described Step 1 in, section refers to the main line part between two adjacent turn road entrances of city expressway, each section at least one Traffic flow detecting section.
CN201711461082.XA 2017-12-28 2017-12-28 A kind of city expressway accident forecast method based on deep learning Active CN108198415B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711461082.XA CN108198415B (en) 2017-12-28 2017-12-28 A kind of city expressway accident forecast method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711461082.XA CN108198415B (en) 2017-12-28 2017-12-28 A kind of city expressway accident forecast method based on deep learning

Publications (2)

Publication Number Publication Date
CN108198415A true CN108198415A (en) 2018-06-22
CN108198415B CN108198415B (en) 2019-07-05

Family

ID=62585277

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711461082.XA Active CN108198415B (en) 2017-12-28 2017-12-28 A kind of city expressway accident forecast method based on deep learning

Country Status (1)

Country Link
CN (1) CN108198415B (en)

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109117987A (en) * 2018-07-18 2019-01-01 厦门大学 Personalized street accidents risks based on deep learning predict recommended method
CN109671266A (en) * 2018-11-26 2019-04-23 北京交通大学 Based on the pilot accident's dynamic early-warning method for accelerating failure risk model
CN110264711A (en) * 2019-05-29 2019-09-20 北京世纪高通科技有限公司 A kind of traffic accident method of determining probability and device
CN110516549A (en) * 2019-07-25 2019-11-29 同济大学 A kind of traffic flow parameter acquisition methods based on binaryzation time-domain diagram
CN110910641A (en) * 2019-12-02 2020-03-24 北京航空航天大学 Neural network-based real-time risk assessment method for highway traffic operation
CN111724596A (en) * 2020-06-23 2020-09-29 上海电科智能系统股份有限公司 Method for intelligently, accurately and automatically identifying and prejudging bottleneck region of expressway
CN111724601A (en) * 2020-06-30 2020-09-29 西安电子科技大学 Motor vehicle traffic flow prediction method suitable for data loss
CN111862605A (en) * 2020-07-20 2020-10-30 腾讯科技（深圳）有限公司 Road condition detection method and device, electronic equipment and readable storage medium
CN112053561A (en) * 2020-09-11 2020-12-08 深兰人工智能芯片研究院（江苏）有限公司 Method, system and device for judging and positioning traffic accident on non-monitored road section
CN112562337A (en) * 2020-12-10 2021-03-26 之江实验室 Expressway real-time traffic accident risk assessment method based on deep learning
CN112836626A (en) * 2021-01-29 2021-05-25 北京百度网讯科技有限公司 Accident determination method and device, model training method and device, and electronic equipment
CN112990545A (en) * 2021-02-08 2021-06-18 东南大学 Traffic safety state prediction method for expressway intersection area
CN113744526A (en) * 2021-08-25 2021-12-03 江苏大学 Expressway risk prediction method based on LSTM and BF
CN113780641A (en) * 2021-08-31 2021-12-10 同济大学 Accident prediction method and device based on transfer learning
CN113919144A (en) * 2021-09-27 2022-01-11 同济大学 Mountain area highway accident cause analysis model modeling method and storage medium
CN113920725A (en) * 2021-09-30 2022-01-11 同济大学 Urban main road accident prediction method based on mesoscopic level and storage medium
CN114925994A (en) * 2022-05-07 2022-08-19 厦门大学 Urban village risk assessment and risk factor positioning method based on deep learning
CN115240408A (en) * 2022-06-15 2022-10-25 中国第一汽车股份有限公司 Vehicle accident early warning method and system based on block chain technology, electronic device and storage medium
CN115293255A (en) * 2022-08-02 2022-11-04 河北雄安荣乌高速公路有限公司 Highway traffic accident risk model construction and risk discrimination method
CN117057757A (en) * 2023-10-12 2023-11-14 深圳比特耐特信息技术股份有限公司 Digital data processing method and system for smart city

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101271625A (en) * 2008-04-03 2008-09-24 东南大学 Method for detecting freeway traffic event by integration supporting vector machine
CN101587644A (en) * 2009-06-02 2009-11-25 同济大学 Automatic detection method for traffic accident on urban expressway based on non-continuous sliding sequence
WO2009146960A3 (en) * 2008-06-03 2010-02-25 Siemens Aktiengesellschaft Method and device for determining loads on bridges
CN101976501A (en) * 2010-10-29 2011-02-16 天津市市政工程设计研究院 Principal component analysis and neural network based port road safety prediction method
CN103258422A (en) * 2013-01-24 2013-08-21 辽宁工程技术大学 Traffic route dynamic induction technology based on coupling of time recursion and neural network
CN105160866A (en) * 2015-08-07 2015-12-16 浙江高速信息工程技术有限公司 Traffic flow prediction method based on deep learning nerve network structure
CN106530714A (en) * 2016-12-21 2017-03-22 东南大学 Secondary traffic accident time prediction method based on traffic flow data

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101271625A (en) * 2008-04-03 2008-09-24 东南大学 Method for detecting freeway traffic event by integration supporting vector machine
WO2009146960A3 (en) * 2008-06-03 2010-02-25 Siemens Aktiengesellschaft Method and device for determining loads on bridges
CN101587644A (en) * 2009-06-02 2009-11-25 同济大学 Automatic detection method for traffic accident on urban expressway based on non-continuous sliding sequence
CN101587644B (en) * 2009-06-02 2012-06-06 同济大学 Automatic detection method for traffic accident on urban expressway based on non-continuous sliding sequence
CN101976501A (en) * 2010-10-29 2011-02-16 天津市市政工程设计研究院 Principal component analysis and neural network based port road safety prediction method
CN103258422A (en) * 2013-01-24 2013-08-21 辽宁工程技术大学 Traffic route dynamic induction technology based on coupling of time recursion and neural network
CN105160866A (en) * 2015-08-07 2015-12-16 浙江高速信息工程技术有限公司 Traffic flow prediction method based on deep learning nerve network structure
CN106530714A (en) * 2016-12-21 2017-03-22 东南大学 Secondary traffic accident time prediction method based on traffic flow data

Cited By (27)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109117987B (en) * 2018-07-18 2020-09-29 厦门大学 Personalized traffic accident risk prediction recommendation method based on deep learning
CN109117987A (en) * 2018-07-18 2019-01-01 厦门大学 Personalized street accidents risks based on deep learning predict recommended method
CN109671266B (en) * 2018-11-26 2020-11-06 北京交通大学 Driver accident dynamic early warning method based on accelerated failure risk model
CN109671266A (en) * 2018-11-26 2019-04-23 北京交通大学 Based on the pilot accident's dynamic early-warning method for accelerating failure risk model
CN110264711A (en) * 2019-05-29 2019-09-20 北京世纪高通科技有限公司 A kind of traffic accident method of determining probability and device
CN110516549A (en) * 2019-07-25 2019-11-29 同济大学 A kind of traffic flow parameter acquisition methods based on binaryzation time-domain diagram
CN110910641A (en) * 2019-12-02 2020-03-24 北京航空航天大学 Neural network-based real-time risk assessment method for highway traffic operation
CN111724596A (en) * 2020-06-23 2020-09-29 上海电科智能系统股份有限公司 Method for intelligently, accurately and automatically identifying and prejudging bottleneck region of expressway
CN111724601A (en) * 2020-06-30 2020-09-29 西安电子科技大学 Motor vehicle traffic flow prediction method suitable for data loss
CN111862605B (en) * 2020-07-20 2022-03-08 腾讯科技（深圳）有限公司 Road condition detection method and device, electronic equipment and readable storage medium
CN111862605A (en) * 2020-07-20 2020-10-30 腾讯科技（深圳）有限公司 Road condition detection method and device, electronic equipment and readable storage medium
CN112053561A (en) * 2020-09-11 2020-12-08 深兰人工智能芯片研究院（江苏）有限公司 Method, system and device for judging and positioning traffic accident on non-monitored road section
CN112562337A (en) * 2020-12-10 2021-03-26 之江实验室 Expressway real-time traffic accident risk assessment method based on deep learning
CN112836626A (en) * 2021-01-29 2021-05-25 北京百度网讯科技有限公司 Accident determination method and device, model training method and device, and electronic equipment
CN112836626B (en) * 2021-01-29 2023-10-27 北京百度网讯科技有限公司 Accident determining method and device, model training method and device and electronic equipment
CN112990545A (en) * 2021-02-08 2021-06-18 东南大学 Traffic safety state prediction method for expressway intersection area
CN113744526A (en) * 2021-08-25 2021-12-03 江苏大学 Expressway risk prediction method based on LSTM and BF
CN113744526B (en) * 2021-08-25 2022-12-23 贵州黔通智联科技股份有限公司 Highway risk prediction method based on LSTM and BF
CN113780641A (en) * 2021-08-31 2021-12-10 同济大学 Accident prediction method and device based on transfer learning
CN113919144A (en) * 2021-09-27 2022-01-11 同济大学 Mountain area highway accident cause analysis model modeling method and storage medium
CN113920725A (en) * 2021-09-30 2022-01-11 同济大学 Urban main road accident prediction method based on mesoscopic level and storage medium
CN114925994A (en) * 2022-05-07 2022-08-19 厦门大学 Urban village risk assessment and risk factor positioning method based on deep learning
CN115240408A (en) * 2022-06-15 2022-10-25 中国第一汽车股份有限公司 Vehicle accident early warning method and system based on block chain technology, electronic device and storage medium
CN115293255A (en) * 2022-08-02 2022-11-04 河北雄安荣乌高速公路有限公司 Highway traffic accident risk model construction and risk discrimination method
CN115293255B (en) * 2022-08-02 2023-05-05 河北雄安荣乌高速公路有限公司 Expressway traffic accident risk model construction and risk discrimination method
CN117057757A (en) * 2023-10-12 2023-11-14 深圳比特耐特信息技术股份有限公司 Digital data processing method and system for smart city
CN117057757B (en) * 2023-10-12 2023-12-22 深圳比特耐特信息技术股份有限公司 Digital data processing method and system for smart city

Also Published As

Publication number Publication date
CN108198415B (en) 2019-07-05

Similar Documents

Publication Publication Date Title
CN108198415B (en) A kind of city expressway accident forecast method based on deep learning
CN105225500B (en) A kind of traffic control aid decision-making method and device
CN104732075B (en) A kind of Urban Road Traffic Accidents risk real-time predicting method
CN103971520B (en) A kind of traffic flow data restoration methods based on temporal correlation
CN109448369A (en) Highway real time execution Risk Calculation method
CN106650913A (en) Deep convolution neural network-based traffic flow density estimation method
CN109840660A (en) A kind of vehicular characteristics data processing method and vehicle risk prediction model training method
CN106652483A (en) Method for arranging traffic information detection points in local highway network by utilizing detection device
CN102346964A (en) Real-time jam prediction and intelligent management system for road traffic network area
CN107437339A (en) Variable information advices plate control method for coordinating and system under a kind of information guidance
CN109272746A (en) A kind of MFD estimating and measuring method based on BP neural network data fusion
CN107767011B (en) Pedestrian characteristic acquisition system and service level dynamic evaluation method for railway station
CN103456163A (en) City expressway interchange traffic capacity and running status discrimination method and system
CN100481153C (en) Method for automatically inspecting highway traffic event based on offset minimum binary theory
CN109455178A (en) A kind of road vehicles traveling active control system and method based on binocular vision
CN107230350A (en) A kind of urban transportation amount acquisition methods based on bayonet socket Yu mobile phone flow call bill data
CN107180534A (en) The express highway section average speed method of estimation of support vector regression fusion
CN106530714A (en) Secondary traffic accident time prediction method based on traffic flow data
CN113535816A (en) Driving performance evaluation method and system for intelligent network cloud control vehicle
CN111986483A (en) Method and device for studying and judging road congestion state based on electric alarm data collision and storage medium
CN105551241B (en) A kind of real-time jamming analysis method based on FCD and EP multi-source datas
Prontri et al. Traffic signal control using fuzzy logic
Kukadapwar et al. Modeling of traffic congestion on urban road network using fuzzy inference system
CN116597642A (en) Traffic jam condition prediction method and system
CN107067723A (en) A kind of method of estimation of Urban road hourage

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