CN110517484A - Diamond interchange area planar crossing sign occlusion prediction model building method - Google Patents

Diamond interchange area planar crossing sign occlusion prediction model building method Download PDF

Info

Publication number
CN110517484A
CN110517484A CN201910721114.8A CN201910721114A CN110517484A CN 110517484 A CN110517484 A CN 110517484A CN 201910721114 A CN201910721114 A CN 201910721114A CN 110517484 A CN110517484 A CN 110517484A
Authority
CN
China
Prior art keywords
traffic
volume
car
vehicle
crossing
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
Application number
CN201910721114.8A
Other languages
Chinese (zh)
Inventor
沈强儒
施佺
曹阳
田源
葛文璇
曹慧
汤天培
周儒夏
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.)
Nantong University
Original Assignee
Nantong 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 Nantong University filed Critical Nantong University
Priority to CN201910721114.8A priority Critical patent/CN110517484A/en
Publication of CN110517484A publication Critical patent/CN110517484A/en
Pending legal-status Critical Current

Links

Classifications

    • 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
    • G06N3/045—Combinations of 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
    • G08G1/065—Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/09—Arrangements for giving variable traffic instructions

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a kind of diamond interchange area planar crossing sign occlusion prediction model building methods, include the following steps: Step 1: identifying stadia computation；Step 2: the hypothesis and framework establishment of Traffic Sign Recognition condition: a. Traffic Sign Recognition condition hypothesis；B. landmark identification framework establishment；Step 3: the foundation of traffic volume forecast model: a. road network function is established；B. traffic volume forecast model construction；Step 4: traffic sign occlusion prediction model construction: (1) fast lane traffic sign occlusion prediction；(2) kerb lane traffic sign occlusion prediction；(3) trolley appears in occlusion prediction in the region AB；(4) prediction of integrated condition signs blocking is as follows.The present invention identifies sighting distance by analysis driver, it is proposed the Probabilistic Prediction Model that the mark based on identification sighting distance and the volume of traffic is blocked, by simulation calculation carry out fail-safe analysis, to improve main line on across level-crossing fingerpost setting validity have certain meaning.

Description

Diamond interchange area planar crossing sign occlusion prediction model building method
Technical field
The invention belongs to traffic application fields, and in particular to a kind of diamond interchange area planar crossing sign occlusion prediction mould Type construction method.
Background technique
Sighting distance is to guarantee the important indicator of traffic safety, and associated documents are research shows that related between traffic accident and sighting distance Property, wherein traffic accident occurs in sighting distance to driver is directly or indirectly relevant to the identification of road sign information accounts for 80% Undesirable section 5%-15% more than normal section, wherein the line-of-sight problem in intersection especially highlights.In highway condition Limited diamond interchange region, ring road with level-crossing is usually arranged by cross-channel, when level-crossing is located at below main line, be easy out The excessively close situation of bridge spacing now on across main line, flag information easily form identification blind area, meanwhile, when running at high speed to intersection The influence that vehicle in the domain of mouth region is blocked vulnerable to front vehicles, cannot accurately identify flag information.Hu Shaorong, Wu et al. are established Autoroute trackside traffic sign is blocked model, with the detection skill of deep approach of learning research road signs information Art；Zhu Zhibang, Guo Tangyi et al. establish last block fingerpost model of expressway exit, and it is public which solves high speed Last block fingerpost of road setting apart from the problem of.Study of recognition about traffic information has Agarwal, when Vivek is based on Between and space to Traffic Sign Recognition problem, Costa, Marco et al. are to the speed between the distance of visual cognition and big compact car of mark The method of poor, time headway etc. research traffic sign placement is spent, Tang Zhihui et al. is with Kohonen neural network method to real-time Traffic flow data carry out clustering, by neural network model in conjunction with Unscented kalman filtering method, while merging interaction Formula method realizes the prediction of comprehensive traffic stream.Forecasting research of the China about diamond interchange region intersection signs blocking probability at present It is less, it is limited to qualitative research.
Summary of the invention
Goal of the invention: in order to solve the deficiencies in the prior art, the present invention provides a kind of diamond interchange area planar intersections Signs blocking prediction model construction method, the present invention identify sighting distance by analysis driver, propose based on identification sighting distance and traffic The Probabilistic Prediction Model that the mark of amount is blocked, by simulation calculation carry out fail-safe analysis, to improve main line on across plane Intersecting fingerpost setting validity has certain meaning.
Technical solution: a kind of diamond interchange area planar crossing sign occlusion prediction model building method, including walk as follows It is rapid:
Step 1: identification stadia computation；
Step 2: the hypothesis and framework establishment of Traffic Sign Recognition condition: a. Traffic Sign Recognition condition hypothesis；B. indicate Identification framework building；
Step 3: the foundation of traffic volume forecast model: a. road network function is established；B. traffic volume forecast model construction；
Step 4: traffic sign occlusion prediction model construction: (1) fast lane traffic sign occlusion prediction；(2) outside vehicle Road traffic sign occlusion prediction；(3) trolley appears in occlusion prediction in the region AB；(4) integrated condition signs blocking is predicted.
As optimization: the step one, identification stadia computation；Traffic sign setting is in road E point, and vehicle driving is to flat Face intersection region finds the traffic information of E point when vehicle driving to A point；This section of time driver carries out traffic information Traffic information is read in identification since B point, is read until C point completes traffic information, vehicle claims in the distance that this time is passed through For recognize reading distance (l')；According to intersection condition traffic condition and road conditions after driver, which recognizes, runs through flag information Make corresponding decision, vehicle driving to D point, the distance that this section of process vehicle passes through is known as judging distance (j)；From action limit D to dynamic Make to complete point F referred to as action distance (l)；Guarantee the safe distance (S) between vehicle and front vehicles；Driver runs through traffic Information simultaneously can complete the distance that necessary driving behavior is passed through safely, swimmingly as identification sighting distance；
Therefore, identification stadia computation is as follows:
Ls=l'+j+l+s (1)；
Wherein: l'=vt1；J=vt2；
Wherein: t1Recognize read time, reading process of the driver to road signs information, usually desirable 3s；t2When judging Between, study and judge process of the driver to road signs information, usually desirable 2s；uaoNormal sections of road speed, can be set by intersection Meter speed degree value (m/s)；us0Vehicle speed after studying and judging can use the 50% of desin speed；
Work as τ2 ”2When → 0,It can approximate value τ2', 0.2~0.9s of value range；τ1' found for driver, identification The time required to making decisions after road signs information (s)；τ1" it is that driver takes the time required to deceleration action (s)；S is to guarantee The distance of vehicle safety state, can value be 50-100m.
As optimization: in the hypothesis and framework establishment of the step two, Traffic Sign Recognition condition, a. traffic sign is known Other condition hypothesis is as follows:
Flag information, which is blocked in prediction model, should fully consider traffic volume forecast, it is assumed that condition is as follows:
1) level-crossing is nearby straightway, and the region maximum longitudinal grade is not more than 3%；
2) compact car, in-between car, large car convert by " specification of the highway route design " (D20-2017 JTG) requirement；
3) traffic sign should meet mark specification setting requirements；
4) a variety of models reach intersection region and obey Poisson distribution, and probability of the large car in front of compact car is 50%；
5) mainly there are three kinds of compact car, in-between car and large car vehicles in intersection region, and proportion is respectively i1, i2, i3。
As optimization: in the hypothesis and framework establishment of the step two, Traffic Sign Recognition condition, b. landmark identification frame Framework is built as follows:
Vehicle drives into level-crossing region, when having large car in front of compact car, should comprehensively consider front oversize vehicle Influence identification of the driver to traffic information；Small vehicle pilot's line of vision height and traffic sign text height height difference are denoted as h (h=h1-h2), road signs information and compact car horizontal line of sight height angle are denoted as φ, φ can value be 15 °, driver visual angle When more than this threshold values, road signs information is easy to be missed, then is known according to geometrical relationship:
Wherein: (m) at a distance between traffic information is put in the discovery of D- road signs information.
As optimization: in the foundation of step three, the traffic volume forecast model, a. road network function is established as follows: being located at Suffered traffic impact node is G (V, E, W) in diamond interchange level-crossing region, and wherein V is diamond interchange region and this The relevant node set of level-crossing；E is diamond interchange level-crossing region adjacent node set；W is each node and hands over Prong distance weighting；By using network of highways forecasting traffic flow model learning formula, volume of traffic history or current data are mapped extremely Future transportation amount, the mapping process of prediction are as follows:
Wherein, XtFor the volume of traffic of the observation of t moment, h (⊙) is diamond interchange area planar intersection design Model learning formula.
As optimization: in the foundation of step three, the traffic volume forecast model, b. traffic volume forecast model construction is such as Under: settling time sequence volume of traffic Q (t) is decomposed into dominant term: time cycle q (t) and volume of traffic trend term f (t), auxiliary Help item: road-net node exponential term ψ (t) and the volume of traffic change item ε (t) at random；
There is duplicate mode according to the volume of traffic in predicted time sequence in t moment, third index flatness can be used Prognosis traffic volume dominant term q (t)+f (t), obtainsRoad network is determined with space-time autoregressive moving average model again Node prognosis traffic volume
Traffic flow within January have certain similitude, by prediction January in four periods time related sequence into Row prediction:
In formula:Volume of traffic variation in 1 day, 5 days, 7 days and 28 days in January can be predicted with third index flatness；θi It (t) is weighted index；
By the volume of traffic of predictionI indicates the seasonal effect factor, can indicate are as follows:
In formula: k is cycle length；siIt (t) is on time step i (i-th of time point) by smoothed out value；tiTo work as The non-smooth value of preceding trend, to be current smoothed value (siAnd a upper smooth value (s (t))i-1(t)) difference；piRefer to that diamond shape is vertical The intersection accidents amount period of change of friendship；
si(t)=α (Gi(t)-pi-k)+(1-α)(si-1(t)+ti-1) (7)；
ti=β (si(t)-si-1(t))+(1-β)ti-1(8)；
pj i=γ (Gi(t)-si)+(1-γ)pi-k(9)；
In formula: α, β, γ are smoothing constant；GiIt (t) is true value, i.e.,
Gi(t)=Q (t) (10)；
Therefore,The weight coefficient θ at placei(t) forgetting factor least square method of recursion is used, i.e.,
With the fundamental quantity that is predicted as in one month, respectively according to for 24 hours, 120h, 168h and 720h are period forecasting, then
θ (t)=[θ1(t) θ2(t) θ3(t) θ4(t)]T(14)；
0 < λ < 1 is forgetting factor, and p (t) is error covariance；
Road network node prognosis traffic volumeIt is main to consider influence of the adjacent traffic stream to the intersection under space-time condition, it can Regard the multiple assigning process of the traffic flow of upstream to downstream road section as, the correlation between road-net node with node increase Or reduce and be varied, according to this theory, mobility model STARIMA is returned by time and is determined in intersection due to road The net node factor volume of traffic；
With STARIMA model, with W(h)Traffic flow interior joint weight expresses the correlation between section, determines The volume of traffic, i.e.,
In formula:Weight in-road network Spatial weight matrix between section i and section j, and meet:
Therefore, the road network exponential term volume of traffic of the transport node as caused by time and spatial coherence are as follows:
Based on this, the numerical value of the major event of prognosis traffic volume and auxiliary item in diamond interchange level-crossing can get,
As optimization: the step four, traffic sign occlusion prediction model construction: compact car is in A point to B point range When being less than AB length range with front cart distance, driver cannot accurately identify road signs information, it is assumed that large car occurs In the region AC, after car appears in oversize vehicle, following 3 aspects can be divided into:
(1) fast lane traffic sign occlusion prediction:
Assuming that the traffic density within the scope of the certain length for the level-crossing that two-way traffic ring road enters is λ (λ=Q (t)/V), then large car quantity before compact car in AC length range under ideal conditions are as follows: the ω AC (λ of ω=0.5 i3/ (i1+i2+i3)), then in this region, large car blocks the Probability p of compact car1Are as follows:
(3) kerb lane traffic sign occlusion prediction:
The distribution of the diamond interchange level-crossing region lane Zhong Ge vehicle and the volume of traffic, left turn lane setting, right-hand bend Factor correlation is arranged in lane, when intersection Regional Traffic Flow is in stationary flow or freestream conditions, point of the vehicle in lane Cloth is affected by speed difference, and small-sized vehicle speed is higher, and large-scale vehicle speed is lower, while large car driver habit is on the outside Lanes, kerb lane cart occupation rate can be calculated by τ times of normal lane, then fast lane is large-scale in AC length range Vehicle occupation rate are as follows: the Probability p of large car occurs in τ λ AC2Are as follows:
(3) trolley appears in occlusion prediction in the region AB:
Kerb lane of the compact car in the region AB, large car appear in the region BC, are easy to happen large car and block generally Rate are as follows:
(4) prediction of integrated condition signs blocking is as follows:
The event that being located in AB length range has one or more compact car is J1, compact car lanes on the outside, in AC In length range without cart event be J2, compact car do not occur the event of large car in inside lanes in the region AC For J3, compact car is not blocked by any in this regional scope, accurate to obtain road signs information probability are as follows:
P=P [J1∩(J2∪J3)]=P1·(P2+P3-P2·P3) (23)；
Wherein p1,p2,p3With running velocity near level-crossing, driver characteristics, the factor phase of vehicle feature It closes.
As optimization:
The utility model has the advantages that the present invention is to determine that diamond inter change level-crossing area flag is blocked probability, use Automobile power theory obtains the identification sighting distance value of diamond interchange level-crossing, is united on this basis by geometry and probability Meter learns principle and establishes typical vehicle signs blocking frame in identification horizon range, with long-term and short period sequence to diamond shape The volume of traffic of stereo region intersection predicted, forms that diamond interchange level-crossing region is long-term and short period sequence Signs blocking Probabilistic Prediction Model, with measured value to block Probabilistic Prediction Model emulation demarcate and examine its reliability.
The result shows that: under long term time sequence prediction, volume of traffic size has more apparent phase with signs blocking probability Guan Xing, related coefficient is up to 0.849；In week age, the confidence interval of signs blocking probability 95% in short period sequence Measured value accounts for the 87.65% of predicted value, and prediction model has preferable reliability, blocks the biggish region of probability to prediction indication Diamond interchange level-crossing is considered as reinforcing mark flexibility setting or the division of lane function.
Detailed description of the invention
Fig. 1 is identification sighting distance schematic diagram of the invention；
Fig. 2 is signs blocking schematic diagram of the invention；
Fig. 3 is weight relationship schematic diagram of the invention；
Fig. 4 is random entry combination frequency distribution schematic diagram of the invention；
Fig. 5 is the volume of traffic of the invention and blocks probability relativity schematic diagram；
Fig. 6 is of the invention to block probabilistic forecasting schematic diagram based on time series.
Specific embodiment
The technical scheme in the embodiments of the invention will be clearly and completely described below, so that the technology of this field Personnel can better understand advantages and features of the invention, to make apparent boundary to protection scope of the present invention It is fixed.Embodiment described in the invention is only a part of the embodiment of the present invention, instead of all the embodiments, based on the present invention In embodiment, those of ordinary skill in the art's every other implementation obtained without making creative work Example, shall fall within the protection scope of the present invention.
Embodiment
Across the level-crossing section on diamond interchange area about main line, large, medium and small vehicle mixing staggeredly travel, space headway Smaller, intersection two sides traffic sign and upcoming traffic information are blocked vulnerable to bridge or large car, it is not easy to timely by driver Identification.In the previous segment limit for entering intersection, driver's main working tasks are to causing visual confusion in road environment Or potential source of traffic information is judged, select reasonable measure with safe and efficient completion traveling task.
1, stadia computation is identified
Driver to traffic information processing be typically required for identification information there are situation, read the information content, judge information The processes such as effect, operation vehicle (deceleration, acceleration, steering, straight trip) to driving task.Identify that sighting distance is that driver has found traffic Information to completion corresponding operating and vehicle and upcoming traffic information keeps certain safe distance (I), vehicle institute in this time The distance that need to pass through, as shown in Figure 1.
Traffic sign is arranged in road E point, as shown in Figure 1.Vehicle by direction running in figure to level-crossing region, when When vehicle driving to A point, the traffic information of E point is found；This time, driver identified traffic information, since B point Traffic information is read, is read until C point completes traffic information, vehicle is referred to as to recognize reading distance in the distance that this time is passed through (l')；Corresponding decision is made according to intersection condition traffic condition and road conditions after driver, which recognizes, runs through flag information, To D point, the distance that this section of process vehicle passes through is known as judging distance (j) vehicle driving；Point F is completed from action limit D to movement to claim For action distance (l)；Guarantee the safe distance (S) between vehicle and front vehicles.Driver runs through traffic information and can pacify Entirely, necessary driving behavior is swimmingly completed, as: the distance that changing Lane, Reduced Speed Now etc. are passed through is identification sighting distance.
Therefore, identification stadia computation is as follows:
Ls=l'+j+l+s (1)
Wherein: l'=vt1；J=vt2；
Wherein: t1Recognize read time, reading process of the driver to road signs information, usually desirable 3s；t2When judging Between, study and judge process of the driver to road signs information, usually desirable 2s；uaoNormal sections of road speed, can be set by intersection Meter speed degree value (m/s)；us0Vehicle speed after studying and judging can use the 50% of desin speed；
Work as τ2 ”2When → 0,It can approximate value τ2', 0.2~0.9s of value range；τ1' found for driver, identification The time required to making decisions after road signs information (s)；τ1" it is that driver takes the time required to deceleration action (s)；S is to guarantee The distance of vehicle safety state, can value be 50-100m.
2, the hypothesis and framework establishment of Traffic Sign Recognition condition
2.1 condition hypothesis
Road signs information is identified by front oversize vehicle in diamond interchange level-crossing region compact car driver Or main line overpass beam collective effect influence, it is related to large car incorporation rate, the volume of traffic etc., at the same also with large car and compact car Time headway is related, wherein the volume of traffic and traffic composition are blocked and played an important role, and therefore, flag information is blocked prediction Traffic volume forecast should be fully considered in model, it is assumed that condition is as follows:
1) level-crossing is nearby straightway, and the region maximum longitudinal grade is not more than 3%；
2) compact car, in-between car, large car convert by " specification of the highway route design " (D20-2017 JTG) requirement；
3) traffic sign should meet mark specification setting requirements；
4) a variety of models reach intersection region and obey Poisson distribution, and probability of the large car in front of compact car is 50%；
5) mainly there are three kinds of compact car, in-between car and large car vehicles in intersection region, and proportion is respectively i1, i2, i3。
2.2 landmark identification framework establishments
Traffic sign placement is considered as the reasonability of intersection signal, driver's identification and installation site.Vehicle drives into plane friendship Prong region, when having large car in front of compact car, should comprehensively consider front oversize vehicle influences driver to traffic information Identification is constituted as shown in Figure 2.Small vehicle pilot's line of vision height and traffic sign text height height difference are denoted as h (h=h1- h2), road signs information and compact car horizontal line of sight height angle are denoted as φ (φ can 15 ° of value), and driver visual angle is more than this When threshold values, road signs information is easy to be missed, as shown in Fig. 2, then being known according to geometrical relationship:
Wherein: (m) at a distance between traffic information is put in the discovery of D- road signs information.
3, traffic volume forecast model
The description of 3.1 problems
Highway diamond interchange discrepancy highway sector speed is higher, is unable to reasonable utilization time series to traffic Amount prediction easily causes road signs information identification problem, and then congestion and accident phenomenon occurs.
Long-term and short period sequence (Long and Short-term Time-series, LST) prediction model is to solve Traffic problems caused by time series.LST uses convolutional neural networks (e Convolution Neural Network, CNN) The traffic volume forecast index between recurrent neural network (Recurrent Neural Networks, RNN) extraction variable, meets and hands over Driver time's control demand under the conditions of prong zone time.
3.2 road network functions are established
Being located at traffic impact node suffered in diamond interchange level-crossing region is G (V, E, W), and wherein V is diamond shape Stereo region node set relevant to this level-crossing.E is diamond interchange level-crossing region adjacent node set；W It is each node and intersection distance weighting.By using network of highways forecasting traffic flow model learning formula, volume of traffic history is mapped Or current data is to future transportation amount, the mapping process of prediction are as follows:
Wherein, XtFor the volume of traffic of the observation of t moment, h (⊙) is diamond interchange area planar intersection design Model learning formula.
3.3 traffic volume forecast model constructions
CNN, RNN model use convolutional Neural and recurrent neural network connect current amount of traffic state, when establishing Between sequence volume of traffic Q (t), be decomposed into dominant term: time cycle q (t) and volume of traffic trend term f (t) assists item: road network Node index item ψ (t) and the volume of traffic change item ε (t) at random.
There is duplicate mode according to the volume of traffic in predicted time sequence in t moment, third index flatness can be used Prognosis traffic volume dominant term q (t)+f (t), obtainsRoad network is determined with space-time autoregressive moving average model again Node prognosis traffic volumeSince the volume of traffic that the intersection near diamond interchange is mainly subject to is up and down at a high speed and adjacent to road The influence on road, the factors such as pedestrian are negligible, but will receive weather, festivals or holidays, socio-economic activity, commercial activity layout etc. The influence of factor.Traffic flow has certain similitude within January, passes through the correlation time sequence in four periods in prediction January Column are predicted:
In formula:Volume of traffic variation in 1 day, 5 days, 7 days and 28 days in January can be predicted with third index flatness；θi It (t) is weighted index.
By the volume of traffic of predictionI indicates the seasonal effect factor, can indicate are as follows:
In formula: k is cycle length；siIt (t) is on time step i (i-th of time point) by smoothed out value；tiTo work as The non-smooth value of preceding trend, to be current smoothed value (siAnd a upper smooth value (s (t))i-1(t)) difference；piRefer to that diamond shape is vertical The intersection accidents amount period of change of friendship.
si(t)=α (Gi(t)-pi-k)+(1-α)(si-1(t)+ti-1) (7)
ti=β (si(t)-si-1(t))+(1-β)ti-1 (8)
pj i=γ (Gi(t)-si)+(1-γ)pi-k (9)
In formula: α, β, γ are smoothing constant；GiIt (t) is true value, i.e.,
Gi(t)=Q (t) (10)
Therefore,The weight coefficient θ at placei(t) forgetting factor least square method of recursion is used, i.e.,
With the fundamental quantity that is predicted as in one month, respectively according to for 24 hours, 120h, 168h and 720h are period forecasting, then
θ (t)=[θ1(t) θ2(t) θ3(t) θ4(t)]T (14)
0 < λ < 1 is forgetting factor, and p (t) is error covariance.
Road network node prognosis traffic volumeIt is main to consider influence of the adjacent traffic stream to the intersection under space-time condition, it can Regard the multiple assigning process of the traffic flow of upstream to downstream road section as, the correlation between road-net node with node increase Or reduce and be varied, according to this theory, mobility model STARIMA (Space-time is returned by time Autoregressive Integrated Moving Average) it determines in intersection due to road-net node factor traffic Amount.
With the STARIMA model of Wu et al., with W(h)Traffic flow interior joint weight expresses the correlation between section, It determinesThe volume of traffic, i.e.,
In formula:Weight in-road network Spatial weight matrix between section i and section j, and meet:
Therefore, the road network exponential term volume of traffic of the transport node as caused by time and spatial coherence are as follows:
Based on this, the numerical value of the major event of prognosis traffic volume and auxiliary item in diamond interchange level-crossing can get,
4, traffic sign occlusion prediction model construction
When compact car is less than AB length range with front cart distance in A point to B point range, driver cannot accurately know Other road signs information, as shown in Figure 2.Assuming that large car appears in the region AC, after car appears in oversize vehicle, can divide It is discussed for 3 aspects.
(1) fast lane traffic sign occlusion prediction
Assuming that the traffic density within the scope of the certain length for the level-crossing that two-way traffic ring road enters is λ (λ=Q (t)/V), then large car quantity before compact car in AC length range under ideal conditions are as follows: the ω AC (λ of ω=0.5 i3/ (i1+i2+i3)), then in this region, large car blocks the Probability p of compact car1Are as follows:
(2) kerb lane traffic sign occlusion prediction
The distribution of the diamond interchange level-crossing region lane Zhong Ge vehicle and the volume of traffic, left turn lane setting, right-hand bend The factors such as lane setting are related, and when intersection Regional Traffic Flow is in stationary flow or freestream conditions, vehicle is in lane Distribution is affected by speed difference, and small-sized vehicle speed is higher, and large-scale vehicle speed is lower, while large car driver habit is outside Side lanes, kerb lane cart occupation rate can be calculated by τ times of normal lane, then fast lane is big in AC length range Type vehicle occupation rate are as follows: the Probability p of large car occurs in τ λ AC2Are as follows:
(3) trolley appears in occlusion prediction in the region AB
Kerb lane of the compact car in the region AB, large car appear in the region BC, are easy to happen large car and block generally Rate are as follows:
(4) integrated condition signs blocking is predicted
The event that being located in AB length range has one or more compact car is J1, compact car lanes on the outside, in AC In length range without cart event be J2, compact car do not occur the event of large car in inside lanes in the region AC For J3, compact car is not blocked by any in this regional scope, accurate to obtain road signs information probability are as follows:
P=P [J1∩(J2∪J3)]=P1·(P2+P3-P2·P3) (23)
Wherein p1,p2,p3With running velocity near level-crossing, driver characteristics, the factors phase such as vehicle feature It closes.
5, emulation experiment and interpretation of result
The acquisition of 5.1 experimental datas
This experiment acquires 2018-05-21 to the Zhejiang 2018-06-20 and Shaanxi, while in 2019-02-11 to 2019- The diamond interchange data of 03-10 Melbourne, AUS 12 main line overpass beams of totally 2 middle of the month, ring road are that two-way traffic enters and leaves Mouthful, intersection is unidirectional two-way traffic, and the speed of service is with measured data V85, annual average daily traffic is by surveying acquisition, various vehicles Type arrival rate obeys Poisson distribution, and large car is in interior kerb lane proportion by practical investigation, traffic flow statistics time interval For 15min, calculating comments equal data as shown in table 1.
1 12 diamond interchange investigational datas of table
Note: the speed of service 1 is that car speed is surveyed before level-crossing；The speed of service 2 is apart from level-crossing 100m surveys speed after slowing down.
5.2 emulation experiment
By means of VISSIM and MATLAB software, with the calculation method of above-mentioned identification sighting distance and based on the mark of time series Occlusion prediction model prediction diamond interchange level-crossing regional traffic amount, time interval uses 15min sequence, with season three times A period, i.e. s were used as using one day when sex index exponential smoothing is predicted1(t)=96 α can, be obtained1,β1,γ1, same s2(t)=480, α can be obtained2,β2,γ2, s3(t)=672 α can, be obtained3,β3,γ3, s4(t)=2688 α can, be obtained4,β4,γ4, to it since t > 96 Smoothing prediction three times, using being compared as predicted value and true value for the 3rd week of surveying in 4 weeks and Parameter analysis.With α1,β1,γ1For Example due to speed and the volume of traffic and blocks and has biggish correlation between probability, using speed as parameter foundation is divided, i.e.,
α1=vmin+(ο(h)-0.5)(vmax-vmin)/H (24)
Wherein ο (h) is 1,2 ... ..., H, vminAnd vmaxIt is divided into evaluate the speed maximum value in intersection in diamond interchange And minimum value, it herein can be according to desin speed value.ParameterValue can be according to section The H equal part in section, whereinPass through calculating Value and the mean absolute error between predicted value and actual value can use the value as α when error is less than 5% among it1,β1, γ1Value, other are according to said method calculated.
(1) fail-safe analysis of forecast power
When calculating weight using band forgetting factor, forgetting factor λ value is typically in the range of 0.95-0.99, takes intermediate value in text, i.e., λ=0.97, θi(t)=[0 00 0]T, n is respectively 1,2,3,4 in curve.Known by Fig. 3, θ in the week1(t) and θ2 (t) certain trend is kept substantially, in Saturday and to θ between Sunday3(t) it is apparently higher than θ1(t) and θ2(t), illustrate the traffic at weekend Trend and Mon-Fri have otherness, θ4(t) on every Saturdays it is being to have relative drop trend on Sunday, is showing Saturday and Sunday pair Mon-Fri volume of traffic degree of dependence is larger, therefore, four kinds of weights is used in text, cover traffic volume forecast process substantially In each link node variation, have certain reasonability.
(2) analysis of simulation result
One week magnitude of traffic flow random entry in 28 days time serieses is carried out with the combinatorial forecast proposed above pre- It surveys, shares 2688 time prediction node, 327 random entry predicted values, the volume of traffic random entry frequency predicts error, such as by histogram Shown in Fig. 4, random entry error Normal Distribution rule illustrates under the conditions of time series random variable quantity in the result It is little that variation blocks impact probability to global flag.
Since the 3rd week correlation in 28 days is more obvious, it was used as traffic volume forecast object using the 3rd week, is used MATLAB predicted, the probability that vehicle is blocked in the predicted time sequence volume of traffic to block probability have it is preferable related Property, as shown in Figure 5.Although blocking probability and the volume of traffic in 300-400 time series section has deviation, overall to reach unanimity Property.
It is compared, leads to signs blocking probability and true value compactness of the signs blocking probabilistic model to time series It crosses with mean absolute error MAD, the accuracy of root-mean-square error RMSE and mean square deviation MSE evaluation and foreca result, such as 2 institute of table Show, with Poisson regression relative coefficient, obtaining the related coefficient between predicted value and measured value is 0.849, correlation with higher Property.
2 prediction result error analysis of table
In order to illustrate the correlation of prediction result and measured result, by vertical with diamond shape of the prediction model to time series The prediction of level-crossing signs blocking probability is handed over, as shown in fig. 6, with image J and Origin data processing method, it is right The mark probability that is blocked surveys 95% confidence interval and prediction 95% confidence interval comparison and determines actual measurement with Area processing method Signs blocking probability value and prediction indication block the ratio of probability value area, and then determine between the two in 95% confidence interval Correlation, be 87.65% according to lap both is calculated, it is wider to illustrate that prediction result is distributed than measured result, it is real The result of survey is included among prediction result, has preferable reliability.
When probability higher diamond interchange level-crossing region is blocked in prediction, traffic sign placement is considered as blocking probability demands Flexibility setting.
Specific conclusion of the invention is as follows:
1) for diamond interchange entrance, nearby the volume of traffic is big, and signs blocking brings security risk to traffic safety, proposes Consider that the traffic sign of identification sighting distance blocks probabilistic model.First choice considers in existing identification stadia computation that there are problems, right Driver's driving behavior decomposition analysis obtains landmark identification sighting distance value.
2) with the traffic volume forecast model based on time series to regional traffic amount in diamond interchange level-crossing into Row prediction, and traffic volume forecast is carried out to 15min and 28day with convolutional neural networks theory, and then analyze the volume of traffic and mark Will is blocked the significant correlation of probability.
3) excavating and analyzing influences the volume of traffic to four influence factors for blocking probability, and one by one to its weight, that is, reliable Property analyzed, shown with the 3rd week by prediction result within 4 weeks as in short term blocking probabilistic model subjects, it is logical It crosses simulation result and shows that mark probability and the volume of traffic variation in one week that is blocked has certain regularity, and and circumstance of occlusion It is positively correlated, there is important reference significance to the setting that research indicates under the conditions of Different Traffic Flows.
4) on the level-crossing position in diamond interchange and main line across bridge spacing it is closer when, traffic sign position is taken an examination Consider and shifts to an earlier date bulletin driver intersection information；Intersection biggish for the volume of traffic is considered as reinforcement repetition flag on dividing strip and sets It sets or the division of lane function.In text in simulation process to large car be mixed into ratio carried out it is preliminary it is assumed that can also basis Actual traffic composition is calculated, to instruct the setting of specific diamond interchange intersection mark.

Claims (7)

1. a kind of diamond interchange area planar crossing sign occlusion prediction model building method, it is characterised in that: including walking as follows It is rapid:
Step 1: identification stadia computation；
Step 2: the hypothesis and framework establishment of Traffic Sign Recognition condition: a. Traffic Sign Recognition condition hypothesis；B. landmark identification Framework establishment；
Step 3: the foundation of traffic volume forecast model: a. road network function is established；B. traffic volume forecast model construction；
Step 4: traffic sign occlusion prediction model construction: (1) fast lane traffic sign occlusion prediction；(2) kerb lane is handed over Logical signs blocking prediction；(3) trolley appears in occlusion prediction in the region AB；(4) integrated condition signs blocking is predicted.
2. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 1, feature It is: the step one, identification stadia computation；Traffic sign is arranged in road E point, vehicle driving to level-crossing mouth region Domain finds the traffic information of E point when vehicle driving to A point；This time, driver identified traffic information, from B point Start to read traffic information, be read until C point completes traffic information, vehicle the distance that this time is passed through be referred to as to recognize reading away from From (l')；Make corresponding determine according to intersection condition traffic condition and road conditions after driver, which recognizes, runs through flag information Plan, vehicle driving to D point, the distance that this section of process vehicle passes through are known as judging distance (j)；Point F is completed from action limit D to movement Referred to as action distance (l)；Guarantee the safe distance (S) between vehicle and front vehicles；Driver runs through traffic information and energy Safety completes the distance that necessary driving behavior is passed through swimmingly as identification sighting distance；
Therefore, identification stadia computation is as follows:
Ls=l'+j+l+s (1)；
Wherein: l'=vt1；J=vt2；
Wherein: t1Recognize read time, reading process of the driver to road signs information, usually desirable 3s；t2Judge the time, drives Study and judge process of the person of sailing to road signs information, usually desirable 2s；uaoNormal sections of road speed can design speed by intersection It spends value (m/s)；us0Vehicle speed after studying and judging can use the 50% of desin speed；
Work as τ2”2When → 0,It can approximate value τ2', 0.2~0.9s of value range；τ1' found for driver, identification traffic The time required to making decisions after flag information (s)；τ1" it is that driver takes the time required to deceleration action (s)；S is to guarantee vehicle The distance of safe condition, can value be 50-100m.
3. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 1, feature Be: in the hypothesis and framework establishment of the step two, Traffic Sign Recognition condition, a. Traffic Sign Recognition condition hypothesis is such as Under:
Flag information, which is blocked in prediction model, should fully consider traffic volume forecast, it is assumed that condition is as follows:
1) level-crossing is nearby straightway, and the region maximum longitudinal grade is not more than 3%；
2) compact car, in-between car, large car convert by " specification of the highway route design " (D20-2017 JTG) requirement；
3) traffic sign should meet mark specification setting requirements；
4) a variety of models reach intersection region and obey Poisson distribution, and probability of the large car in front of compact car is 50%；
5) mainly there are three kinds of compact car, in-between car and large car vehicles in intersection region, and proportion is respectively i1, i2, i3。
4. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 3, feature Be: in the hypothesis and framework establishment of the step two, Traffic Sign Recognition condition, b. landmark identification framework establishment is as follows:
Vehicle drives into level-crossing region, and when having large car in front of compact car, should comprehensively consider front oversize vehicle influences Identification of the driver to traffic information；Small vehicle pilot's line of vision height and traffic sign text height height difference are denoted as h (h= h1-h2), road signs information and compact car horizontal line of sight height angle are denoted as φ, φ can value be 15 °, driver visual angle is more than When this threshold values, road signs information is easy to be missed, then is known according to geometrical relationship:
Wherein: (m) at a distance between traffic information is put in the discovery of D- road signs information.
5. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 1, feature Be: in the foundation of step three, the traffic volume forecast model, a. road network function is established as follows: being located at diamond interchange plane Suffered traffic impact node is G (V, E, W) in the region of intersection, and wherein V is diamond interchange region and this level-crossing phase The node set of pass；E is diamond interchange level-crossing region adjacent node set；W is each node and intersection distance weighting； By using network of highways forecasting traffic flow model learning formula, volume of traffic history or current data are mapped to future transportation amount, in advance The mapping process of survey are as follows:
Wherein, XtFor the volume of traffic of the observation of t moment, h (⊙) is diamond interchange area planar intersection design model Practise formula.
6. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 5, feature Be: in the foundation of step three, the traffic volume forecast model, b. traffic volume forecast model construction is as follows: settling time sequence Column volume of traffic Q (t) is decomposed into dominant term: time cycle q (t) and volume of traffic trend term f (t) assists item: road-net node Exponential term ψ (t) and the volume of traffic change item ε (t) at random；
There is duplicate mode according to the volume of traffic in predicted time sequence in t moment, can be predicted with third index flatness The volume of traffic dominant term q (t)+f (t), obtainsRoad network node is determined with space-time autoregressive moving average model again Prognosis traffic volume
Traffic flow has certain similitude within January, is carried out by the time related sequence in four periods in prediction January pre- It surveys:
In formula:Volume of traffic variation in 1 day, 5 days, 7 days and 28 days in January can be predicted with third index flatness；θi(t) it is Weighted index；
By the volume of traffic of predictionI indicates the seasonal effect factor, can indicate are as follows:
In formula: k is cycle length；siIt (t) is on time step i (i-th of time point) by smoothed out value；tiTo work as precursor The non-smooth value of gesture, to be current smoothed value (siAnd a upper smooth value (s (t))i-1(t)) difference；piRefer to diamond interchange Intersection accidents amount period of change；
si(t)=α (Gi(t)-pi-k)+(1-α)(si-1(t)+ti-1) (7)；
ti=β (si(t)-si-1(t))+(1-β)ti-1(8)；
pj i=γ (Gi(t)-si)+(1-γ)pi-k(9)；
In formula: α, β, γ are smoothing constant；GiIt (t) is true value, i.e.,
Gi(t)=Q (t) (10)；
Therefore,The weight coefficient θ at placei(t) forgetting factor least square method of recursion is used, i.e.,
With the fundamental quantity that is predicted as in one month, respectively according to for 24 hours, 120h, 168h and 720h are period forecasting, then
θ (t)=[θ1(t) θ2(t) θ3(t) θ4(t)]T(14)；
0 < λ < 1 is forgetting factor, and p (t) is error covariance；
Road network node prognosis traffic volumeIt is main to consider influence of the adjacent traffic stream to the intersection under space-time condition, it can regard as It is the multiple assigning process of the traffic flow of upstream to downstream road section, the correlation between road-net node adds deduct with the increasing of node It is varied less, according to this theory, mobility model STARIMA is returned by time and is determined in intersection due to road network section The point factor volume of traffic；
With STARIMA model, with W(h)Traffic flow interior joint weight expresses the correlation between section, determinesTraffic Amount, i.e.,
In formula:Weight in-road network Spatial weight matrix between section i and section j, and meet:
Therefore, the road network exponential term volume of traffic of the transport node as caused by time and spatial coherence are as follows:
Based on this, the numerical value of the major event of prognosis traffic volume and auxiliary item in diamond interchange level-crossing can get,
7. diamond interchange area planar crossing sign occlusion prediction model building method according to claim 6, feature Be: the step four, traffic sign occlusion prediction model construction: compact car in A point to B point range with the big spacing in front When from being less than AB length range, driver cannot accurately identify road signs information, it is assumed that large car appears in the region AC, small vapour After vehicle appears in oversize vehicle, following 3 aspects can be divided into:
(1) fast lane traffic sign occlusion prediction:
Assuming that the traffic density within the scope of the certain length for the level-crossing that two-way traffic ring road enters is λ (λ=Q (t)/V), Then large car quantity before compact car in AC length range under ideal conditions are as follows: the ω AC (λ of ω=0.5 i3/(i1+i2+ i3)), then in this region, large car blocks the Probability p of compact car1Are as follows:
(2) kerb lane traffic sign occlusion prediction:
The distribution of the diamond interchange level-crossing region lane Zhong Ge vehicle and the volume of traffic, left turn lane setting, right-turn lane Setting factor is related, when intersection Regional Traffic Flow is in stationary flow or freestream conditions, distribution of the vehicle in lane by Speed difference is affected, and small-sized vehicle speed is higher, and large-scale vehicle speed is lower, while large car driver habit lane on the outside Traveling, kerb lane cart occupation rate can be calculated by τ times of normal lane, then fast lane large car in AC length range accounts for There is rate are as follows: the Probability p of large car occurs in τ λ AC2Are as follows:
(3) trolley appears in occlusion prediction in the region AB:
Kerb lane of the compact car in the region AB, large car appear in the region BC, are easy to happen large car and block probability are as follows:
(4) prediction of integrated condition signs blocking is as follows:
The event that being located in AB length range has one or more compact car is J1, compact car lanes on the outside, in AC length model Enclose it is interior without cart event be J2, for compact car in inside lanes, the event for not occurring large car in the region AC is J3, In Compact car is not blocked by any in this regional scope, accurate to obtain road signs information probability are as follows:
P=P [J1∩(J2∪J3)]=P1·(P2+P3-P2·P3) (23)；
Wherein p1,p2,p3It is related with the factor of running velocity near level-crossing, driver characteristics, vehicle feature.
CN201910721114.8A 2019-08-06 2019-08-06 Diamond interchange area planar crossing sign occlusion prediction model building method Pending CN110517484A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910721114.8A CN110517484A (en) 2019-08-06 2019-08-06 Diamond interchange area planar crossing sign occlusion prediction model building method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910721114.8A CN110517484A (en) 2019-08-06 2019-08-06 Diamond interchange area planar crossing sign occlusion prediction model building method

Publications (1)

Publication Number Publication Date
CN110517484A true CN110517484A (en) 2019-11-29

Family

ID=68624549

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910721114.8A Pending CN110517484A (en) 2019-08-06 2019-08-06 Diamond interchange area planar crossing sign occlusion prediction model building method

Country Status (1)

Country Link
CN (1) CN110517484A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233428A (en) * 2020-10-10 2021-01-15 腾讯科技（深圳）有限公司 Traffic flow prediction method, traffic flow prediction device, storage medium and equipment
CN114937359A (en) * 2022-05-20 2022-08-23 四川大学 Method, system, terminal and medium for positioning and analyzing cascade fault of traffic infrastructure
CN115202534A (en) * 2022-07-11 2022-10-18 南昌航空大学 Positioning-based driver and digital traffic sign interaction method and system

Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070146162A1 (en) * 2005-12-22 2007-06-28 Nissan Technical Center North America, Inc. Vehicle communication system
CN101727743A (en) * 2009-11-27 2010-06-09 东南大学 Method for setting roadside speed limit signs on ordinary road
CN101739838A (en) * 2009-11-27 2010-06-16 东南大学 Determination method for continuously setting speed limit signs at roadside of general road
CN102693633A (en) * 2012-06-07 2012-09-26 浙江大学 Short-term traffic flow weighted combination prediction method
CN106448174A (en) * 2016-12-05 2017-02-22 南通大学 Calculation method of traffic sign shielding probability on plane intersection region of trunk highway
US20170116850A1 (en) * 2015-10-27 2017-04-27 Hyundai Motor Company Traffic prediction system, vehicle-mounted display apparatus, vehicle, and traffic prediction method
CN106709599A (en) * 2016-12-15 2017-05-24 中山大学 Whole region guide maximization-based road guide sign laying method
CN107230351A (en) * 2017-07-18 2017-10-03 福州大学 A kind of Short-time Traffic Flow Forecasting Methods based on deep learning
US20170323448A1 (en) * 2016-05-09 2017-11-09 Iteris, Inc. Pedestrian tracking at a traffic intersection to identify vulnerable roadway users for traffic signal timing, pedestrian safety, and traffic intersection control
CN109493603A (en) * 2018-11-29 2019-03-19 东北林业大学 A kind of highway gantry sign large car blocks probability determination method
CN109544911A (en) * 2018-10-30 2019-03-29 中山大学 A kind of city road net traffic state prediction technique based on LSTM-CNN
CN109741604A (en) * 2019-03-05 2019-05-10 南通大学 Based on tranquilization shot and long term memory network model prediction intersection traffic method of flow
CN109903557A (en) * 2019-03-04 2019-06-18 南京邮电大学 Based on the freeway traffic flow prediction technique for improving independent loops neural network
CN110021165A (en) * 2019-03-18 2019-07-16 浙江工业大学 A kind of traffic flow forecasting method based on Autoencoder-LSTM Fusion Model
CN110033629A (en) * 2018-01-11 2019-07-19 丰田自动车株式会社 Signal light identifier and automated driving system
CN110047291A (en) * 2019-05-27 2019-07-23 清华大学深圳研究生院 A kind of Short-time Traffic Flow Forecasting Methods considering diffusion process

Patent Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070146162A1 (en) * 2005-12-22 2007-06-28 Nissan Technical Center North America, Inc. Vehicle communication system
CN101727743A (en) * 2009-11-27 2010-06-09 东南大学 Method for setting roadside speed limit signs on ordinary road
CN101739838A (en) * 2009-11-27 2010-06-16 东南大学 Determination method for continuously setting speed limit signs at roadside of general road
CN102693633A (en) * 2012-06-07 2012-09-26 浙江大学 Short-term traffic flow weighted combination prediction method
US20170116850A1 (en) * 2015-10-27 2017-04-27 Hyundai Motor Company Traffic prediction system, vehicle-mounted display apparatus, vehicle, and traffic prediction method
US20170323448A1 (en) * 2016-05-09 2017-11-09 Iteris, Inc. Pedestrian tracking at a traffic intersection to identify vulnerable roadway users for traffic signal timing, pedestrian safety, and traffic intersection control
CN106448174A (en) * 2016-12-05 2017-02-22 南通大学 Calculation method of traffic sign shielding probability on plane intersection region of trunk highway
CN106709599A (en) * 2016-12-15 2017-05-24 中山大学 Whole region guide maximization-based road guide sign laying method
CN107230351A (en) * 2017-07-18 2017-10-03 福州大学 A kind of Short-time Traffic Flow Forecasting Methods based on deep learning
CN110033629A (en) * 2018-01-11 2019-07-19 丰田自动车株式会社 Signal light identifier and automated driving system
CN109544911A (en) * 2018-10-30 2019-03-29 中山大学 A kind of city road net traffic state prediction technique based on LSTM-CNN
CN109493603A (en) * 2018-11-29 2019-03-19 东北林业大学 A kind of highway gantry sign large car blocks probability determination method
CN109903557A (en) * 2019-03-04 2019-06-18 南京邮电大学 Based on the freeway traffic flow prediction technique for improving independent loops neural network
CN109741604A (en) * 2019-03-05 2019-05-10 南通大学 Based on tranquilization shot and long term memory network model prediction intersection traffic method of flow
CN110021165A (en) * 2019-03-18 2019-07-16 浙江工业大学 A kind of traffic flow forecasting method based on Autoencoder-LSTM Fusion Model
CN110047291A (en) * 2019-05-27 2019-07-23 清华大学深圳研究生院 A kind of Short-time Traffic Flow Forecasting Methods considering diffusion process

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
ALEXANDER SHUSTANOV，PAVEL YAKIMOV: "《CNN Design for Real-Time Traffic Sign Recognition》", 《PROCEDIA ENGINEERING》 *
张勇，关伟: "《预测交通流量时间序列的组合动态建模方法》", 《吉林大学学报（工学版）》 *
胡绍荣，马荣国，李苗苗， 邓亚娟: "《多车道高速公路路侧交通标志遮挡概率研究》", 《中国公路学报》 *
韩卫国，王劲峰，高一鸽，胡建军: "《区域交通流的时空预测与分析》", 《公路交通科技》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112233428A (en) * 2020-10-10 2021-01-15 腾讯科技（深圳）有限公司 Traffic flow prediction method, traffic flow prediction device, storage medium and equipment
CN112233428B (en) * 2020-10-10 2023-09-22 腾讯科技（深圳）有限公司 Traffic flow prediction method, device, storage medium and equipment
CN114937359A (en) * 2022-05-20 2022-08-23 四川大学 Method, system, terminal and medium for positioning and analyzing cascade fault of traffic infrastructure
CN115202534A (en) * 2022-07-11 2022-10-18 南昌航空大学 Positioning-based driver and digital traffic sign interaction method and system
CN115202534B (en) * 2022-07-11 2023-06-23 南昌航空大学 Interaction method and system of driver and digital traffic sign based on positioning

Similar Documents

Publication Publication Date Title
Skabardonis et al. Real-time estimation of travel times on signalized arterials
Zhao et al. Truck traffic speed prediction under non-recurrent congestion: Based on optimized deep learning algorithms and GPS data
CN100456335C (en) Visual evaluating method for urban traffic system state based on traffic flow phase character istic and its application
CN104732075B (en) A kind of Urban Road Traffic Accidents risk real-time predicting method
Alhajyaseen et al. Left-turn gap acceptance models considering pedestrian movement characteristics
CN110517484A (en) Diamond interchange area planar crossing sign occlusion prediction model building method
CN110310516A (en) A kind of highway merging area traffic conflict method of discrimination based on vehicle driving trace prediction
CN109816977A (en) A kind of integrative design intersection evaluation system of data-driven
CN101739823A (en) Road-section average travel time measuring method suitable for low-frequency sampling
KR101277007B1 (en) Method for offering passing time of expresshighway passing auto mobile and system thereof
Wang et al. Analysis of the driver’s behavior characteristics in low volume freeway interchange
Bae et al. Multicontextual machine-learning approach to modeling traffic impact of urban highway work zones
Kieu et al. Benefits and issues of bus travel time estimation and prediction
Chen et al. Using automatic passenger counter data in bus arrival time prediction
CN113487872B (en) Bus transit time prediction method based on big data and artificial intelligence
Chen et al. Exploring spatiotemporal mobilities of highway traffic flows for precise travel time estimation and prediction based on electronic toll collection data
Mamidipalli et al. Probit-based pedestrian gap acceptance model for midblock crossing locations
Nohekhan et al. Investigating the impacts of I-66 Inner Beltway dynamic tolling system
Yu et al. GPS data mining at signalized intersections for congestion charging
Osei et al. Modelling of segment level travel time on urban roadway arterials using floating vehicle and GPS probe data
Dixon et al. Quantifying safety performance of driveways on state highways
Barchański Analysis of critical gap times and follow-up times at selected, median, uncontrolled T-intersections differentiated by the nature of the surrounding
Monsere et al. Operational guidance for bicycle-specific traffic signals in the United States
Mallikarjuna Operational analysis of roundabouts under mixed traffic flow condition
Huo et al. Development of Level-of-Service Criteria based on a Single Measure for BRT in China

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20191129

RJ01 Rejection of invention patent application after publication