CN106846804B - The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain - Google Patents
The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain Download PDFInfo
- Publication number
- CN106846804B CN106846804B CN201710124151.1A CN201710124151A CN106846804B CN 106846804 B CN106846804 B CN 106846804B CN 201710124151 A CN201710124151 A CN 201710124151A CN 106846804 B CN106846804 B CN 106846804B
- Authority
- CN
- China
- Prior art keywords
- state
- time headway
- mrow
- time
- headway
- 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.)
- Active
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
- G08G1/0108—Measuring and analyzing of parameters relative to traffic conditions based on the source of data
- G08G1/0116—Measuring and analyzing of parameters relative to traffic conditions based on the source of data from roadside infrastructure, e.g. beacons
-
- 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
Landscapes
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of real-time saturation volume rate method of estimation in intersection based on hidden Markov chain, comprise the following steps:A, collection influence the traffic flow data of saturation headway, comprising saturation headway in itself;B, is trained the parameter for implying Markov chain model by the traffic flow data of collection;C, gathers the time headway before signal intersection parking line, the implicit time headway state of time headway sequence behind is estimated, obtains continuously stablizing time headway sequence in real time;D, is averaging to stablizing time headway sequence, and is used as the corresponding saturation volume rate of the stop line by the use of the inverse of average value.The present invention can estimate saturation volume rate in real time, and provide parameter basis for traffic control.
Description
Technical field
It is specific next the present invention relates to a kind of method of estimation for the real-time saturation volume rate in intersection being used in urban traffic control
Say, be to relate to the use of detector data and signal information, the state of the time headway of intersection is modeled, and training mould
Type, estimates a kind of method of saturation volume rate in real time with model afterwards.
Background technology
Saturation volume rate is one of key parameter in urban traffic control system, is referred to when the green light signals of signal lamp open
After bright, fleet crosses flow rate during stop line with saturation state.The flow rate changes with the change in time and space, and very
It has impact on the control effect of intersection in big degree.At present, in mainstream control system mostly using fixed saturation volume rate or
Person simply sets saturation volume rate with quantile.The method that this comparison ossifys is difficult to reflect actual conditions.So as to have
Necessity proposes the method for estimation of new saturation volume rate for the traffic stream characteristics of intersection.
The content of the invention
The technical problem to be solved in the present invention is overcome existing method accurately to estimate crossing inlet road saturation in real time
The defects of flow rate.The present invention proposes a kind of saturation volume rate method of estimation based on hidden Markov chain.To solve technical problem, this
The solution of invention is:For some entrance driveway, when green light opens it is bright after, fleet crosses stop line, so as to produce a system
Arrange successive time headway.Time headway is divided into five classes by the present invention according to the state of rear car:Acceleration mode, stable state, subtracts
Fast state 1, deceleration regime 2 are not parking to cross totally five kinds of state.Stable state therein is exactly fleet with stable saturation headstock
When away from the state for crossing stop line, the average value of all time headways can serve as stop line within the cycle under stable state
Estimate.
In order to identify the state of time headway, the present invention is based on Markov Chain, it is proposed that the hidden Ma Erke of time headway
Husband's chain model.Wherein, hidden state is exactly five kinds of states, and observable parameter is time headway.
The technical scheme is that:1 establishes the Hidden Markov chain model of time headway during green light signals;2 collections
Data, that is, a series of data of the time headway during green lights are gathered, include the corresponding state of time headway and headstock
When the value away from itself;3 are trained model using EM algorithms, obtain the value of model parameter;4 utilize the hidden Ma Er after training
Section husband time headway model carries out state estimation to time headway, obtains the corresponding status switch of time headway;5 just with shape
Stable state in state sequence, is averaging it, its estimate reciprocal as saturation volume rate.
Saturation volume rate can be estimated in real time by the method for the present invention, and parameter basis is provided for traffic control.
Brief description of the drawings
Fig. 1 is that the detector of stop line sets figure;
Fig. 2 is the definition figure of time headway;
Fig. 3 is time headway variation diagram ideally;
Fig. 4 is time headway state transition diagram.
Embodiment
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1 is that the detector of stop line sets figure.Wherein, detector is arranged at stop line, so that each car is by stopping
It can be detected after fare by stop line.Fig. 2 is the computational methods of time headway, and Adjacent vehicles pass through the time of stop line
Detected value of the difference as time headway.Fig. 3 be after the green light of signal lamp opens bright, ideally by the headstock of stop line when
Away from variation diagram.At the initial stage of green light, since vehicle acceleration is not yet complete, so that time headway is larger, with subsequent vehicle
By, car speed gradually increases, therefore time headway is less and less, to the 5th or the 6th car or so, time headway base
Stablize in sheet, the time headway of this stabilization is exactly the corresponding saturation headway of the stop line.The saturation headway falls
Number is exactly saturation volume rate.Due to reasons such as the aging on road surface, the composition of vehicle, weather, which can constantly change, so that
It is necessary to estimate which time headway is to stablize time headway in real time, and utilizes saturation headway estimation saturation volume rate.
For time headway all during some green light, its state has five kinds of Fig. 4:Acceleration mode, stable state,
Deceleration regime 1, deceleration regime 2 and not parking passes through state.Acceleration mode is the state that green light opens bright initial stage, and stable state is tight
And then acceleration mode.When green light terminates, and is converted to amber light, vehicle deceleration is simultaneously parked at stop line.Deceleration regime 1 is corresponding
It is that upstream has enough fleet's demands, corresponding deceleration regime 2 is that the vehicle that upstream reaches not is saturation.Do not slow down and pass through
State refers to that vehicle does not pass through the state of intersection directly by deceleration regime.
The transfer of five kinds of states is as shown in Figure 4.
The present invention utilizes following symbols:
The state value set of time headway, shares five kinds of states:S=s1,s2,s3,s4,s5;
Observe value set:M∈R+, which is a positive real number;
The observation sequence of a cycle:O=o1,o2,o3,o4,...ok....;
Hidden state sequence in a cycle:I=i1,i2,i3,i4,....ik..., each of which hidden state take
It is worth for a certain kind in S.
The step of the present invention
The step of the present invention includes following three step:
Time headway sequence is modeled using hidden Markov chain
Gathered data is trained the parameter of model
Time headway sequence is identified in real time using trained model, and utilizes stable state time headway therein
Estimate of the average inverse of observation as saturation volume rate.
It is specific as follows:
First step:Hidden Markov chain models
Implicit markovian modeling includes:
State-transition matrix
Probabilistic relation between emission probability, namely hidden state and observation
Initial hidden state distribution
Hidden state refers to five kinds of states, when different conditions are shifted, is described with state-transition matrix.State turns
Move matrix and be defined as A=[aij].Wherein aijRepresent that the state of time headway is transferred to the probability of state j from i, for example, a12Represent
The probability of stable state is transformed into from acceleration mode.
Emission probability, namely the relation of each state of time headway and the observation of time headway in itself with logarithm just
State is distributed to express.When k-th of time headway and time headway state are m, the observation of time headway is distributed as:
Wherein, m is the numbering of state, and value range is 1~5;K is the time headway in whole time headway sequence
Numbering, value range be 1~N, N for time headway total during green light quantity.
Initial hidden state distribution π={ π1,π2,π3,π4,π5, represent that first hidden state is a certain for five kinds of states
The probability of kind.
Second step:Model parameter determines
Imply parameter lambda=(A, { μ that markovian parameter Estimation refers to estimation modelk,m;σk,m},π),
A=[aij]
·μk,m;σk,m
π={ π1,π2,π3,π4,π5}
Estimate that above-mentioned parameter needs some data, the data that data acquisition is detected using the detector of stop line, and hand
The time headway status information of work identification, specifically the fleet with video camera shooting by stop line, passes through manual identified, system afterwards
Count time headway and its corresponding state.If be still within by the vehicle of stop line in acceleration, time headway is acceleration
State;If vehicle accelerates to maximal rate by stop line from totally stationary state, which is to stablize shape
State;If vehicle does not pass through by parking by stop line, time headway state directly during green light to be not parking
State;If red light opens the bright rear vehicle stopped and does not stop in a upper red light, for deceleration regime 2;Red light open it is bright after stop
If vehicle only in a upper red parking, stops, is then deceleration regime 1 again when this red light opens bright.
Assuming that acquiring w green light, the time headway during each green light and its state are O=o1,o2... and I=i1,
i2... estimates that the overall step of model parameter is using EM algorithms:
The parameter Estimation of hidden Markov chain uses EM algorithms, and this method is set out by setting initial parameter, circulated
Generation, obtained argument sequence converge to optimal value gradually.It is assumed that given initial parameter value λ ', the whole description of process are:
1. the initialization value λ ' of setup parameter:Here each [a is setij] and each { μk,m;σk,mAnd it is each
A πi;
2. construct auxiliary function Q:
3. maximizing function Q obtains an iterationAnd make λ '=λ*;
4. repeat the above process by several times, then obtaining parameter lambda=(A, { μk,m;σk,m, π) estimate.
In above-mentioned second step, the computational methods of each single item are in auxiliary function Q:
·
·
In both the above formula
So as to which the auxiliary function in second step can be expressed as:
The parameter π that maximization Q in third step is obtainediAnd aijComputational methods are respectively:
Parameter μk,mAnd σk,mComputational methods method obtained by solving following Nonlinear System of Equations:
Third step:The estimation of time headway status switch and saturation volume rate calculate
The estimation of time headway status switch is estimated using Viterbi algorithm.Refer to given a cycle when away from
Observation sequence O=o1,o2,o3,o4,...oT, status switch I=i that estimating system implies1,i2,i3,i4,....iT, wherein, no
The length T of synperiodic observation sequence is different.For given cycle observation sequence, it is believed that T is a constant.Herein,
The parameter lambda of hidden Markov chain=(A, { μk,m;σk,m, π) provided by the algorithm of previous step, so as to as known
's.
It estimates that flow is:
Step 1:Initialization, to 1≤i≤N, δ1(i)=- ln (πi)-ln(b1,i(o1));ψ1(i)=0;
Step 2:Iterative calculation, to 1≤j≤N and
Step 3:Iteration ends,
Step 4:Optimum state is recalled, for t=T-1, T-2 ... 1, it *=ψt+1(i* t+1);
Assuming that have passed through system decoding, the status switch of system has been obtained:Acceleration mode-acceleration mode-acceleration mode-add
Fast state-stable state-... stable state-deceleration regime-deceleration regime.Wherein there are n stable state, corresponding observation sequence
It is classified as os,1,os,2....os,nIt is so as to the time headway computational methods in the cycle:
So far, the saturation volume rate in crossing inlet road just estimates.
Claims (3)
1. the real-time saturation volume rate method of estimation in a kind of intersection based on hidden Markov chain, it is characterised in that this method includes
Following steps:
1) the Hidden Markov chain model of time headway during green light signals is established, comprising:
1. the probabilistic relation between emission probability, i.e. hidden state and observation, the hidden state is to be passed through according to vehicle
The state of stop line moment is classified as five kinds:Acceleration mode, stable state, it is not parking cross state, deceleration regime 1 and slow down
State 2;Emission probability is expressed with logarithm normal distribution:When k-th of time headway and time headway state are m, headstock
When away from observation be distributed as:
<mrow>
<msub>
<mi>b</mi>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msqrt>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&pi;&sigma;</mi>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
<mn>2</mn>
</msubsup>
<mi>t</mi>
</mrow>
</msqrt>
</mfrac>
<msup>
<mi>exp</mi>
<mrow>
<mo>-</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mi>n</mi>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>&mu;</mi>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msub>
<mi>&sigma;</mi>
<mrow>
<mi>k</mi>
<mo>,</mo>
<msup>
<mi>m</mi>
<mn>2</mn>
</msup>
</mrow>
</msub>
</mrow>
</mfrac>
</mrow>
</msup>
</mrow>
Wherein, m is the numbering of state, and value range is 1~5;K is volume of the time headway in whole time headway sequence
Number, value range is 1~N, the quantity of time headway total during being green light N;
2. state-transition matrix A=[aij], wherein aijRepresent that the state of time headway is transferred to the probability of state j from i;
3. initial hidden state distribution π={ π1,π2,π3,π4,π5, represent that first hidden state is five kinds of state a certain kind
Probability;
Parameter lambda wherein to be determined=(A, { μk,m;σk,m},π);
2) data of the time headway during a series of green lights of collection, include the corresponding state of time headway and time headway sheet
The value of body;
3) model is trained using EM algorithms, obtains the value of model parameter;
4) state estimation is carried out to time headway using the Hidden Markov time headway model after training, obtains time headway
Corresponding status switch;
5) just with the stable state in status switch, it is averaging, its estimate reciprocal as saturation volume rate.
2. the real-time saturation volume rate method of estimation in the intersection according to claim 1 based on hidden Markov chain, its feature
It is, step 2) is specially:The data detected using the detector of stop line, and the time headway state letter identified by hand
Breath, specifically the fleet with video camera shooting by stop line, afterwards by manual identified, counts time headway and its corresponding shape
State:If be still within by the vehicle of stop line in acceleration, time headway is acceleration mode;If vehicle is from totally stationary
State accelerates to maximal rate by stop line, then the time headway state is stable state;If vehicle is not by parking
And by stop line directly during green light, then time headway state crosses state to be not parking;Red light opens the bright rear car stopped
If do not stopped in a upper red light, for deceleration regime 2;If red light opens the bright rear vehicle stopped in a upper red light
Parking, stops, is then deceleration regime 1 again when this red light opens bright.
3. the real-time saturation volume rate method of estimation in the intersection according to claim 1 based on hidden Markov chain, its feature
It is, step 5) is specially:After obtaining the corresponding status switch of time headway, to the time headway of stable state in status switch
Sequence os,1,os,2......os,nIt is averaging, and the estimate using the inverse of the average value as saturation volume rate, i.e.,:
Wherein, n is the number of stable state in time headway hidden state.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710124151.1A CN106846804B (en) | 2017-03-03 | 2017-03-03 | The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710124151.1A CN106846804B (en) | 2017-03-03 | 2017-03-03 | The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106846804A CN106846804A (en) | 2017-06-13 |
CN106846804B true CN106846804B (en) | 2018-04-20 |
Family
ID=59137347
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710124151.1A Active CN106846804B (en) | 2017-03-03 | 2017-03-03 | The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106846804B (en) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107564290B (en) * | 2017-10-13 | 2021-02-19 | 公安部交通管理科学研究所 | Urban road intersection saturated flow rate calculation method |
CN107909825B (en) * | 2017-11-03 | 2020-08-25 | 航天科工广信智能技术有限公司 | Method for detecting regression saturation flow rate in Gaussian process |
CN108597239B (en) * | 2018-04-28 | 2021-02-09 | 重庆大学 | Traffic light control system and method based on Markov decision |
CN109389826B (en) * | 2018-12-14 | 2021-03-16 | 武汉理工大学 | Real-time calculation method for saturation flow rate of signalized intersection |
CN111613051B (en) * | 2019-02-25 | 2022-02-25 | 北京嘀嘀无限科技发展有限公司 | Method and device for estimating saturated headway |
CN109993983B (en) * | 2019-04-09 | 2020-10-27 | 西安交通大学 | Intelligent transportation vehicle self-adaptive induction method based on Markov chain |
CN110910661B (en) * | 2019-11-13 | 2021-07-20 | 北京百度网讯科技有限公司 | Intersection message control machine control method and device |
CN111028509B (en) * | 2019-12-19 | 2020-12-15 | 河北万方中天科技有限公司 | Traffic saturation flow obtaining method and device and terminal equipment |
CN111882860B (en) * | 2020-06-05 | 2022-03-04 | 阿波罗智联(北京)科技有限公司 | Lane adjusting method and device |
CN113380026B (en) * | 2021-05-28 | 2022-07-22 | 南京理工大学 | HMM-based highway traffic prediction method |
CN113643531B (en) * | 2021-07-20 | 2022-09-20 | 东北大学 | Intersection lane saturation flow rate calculation method based on small time zone division statistics |
CN115131971A (en) * | 2022-06-30 | 2022-09-30 | 北京图盟科技有限公司 | Control delay determination method based on track data and related components |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101727746A (en) * | 2009-09-18 | 2010-06-09 | 苏州大学 | Method for dynamically estimating travel time of motor vehicle on urban road under control of signal lamps |
JP2010238247A (en) * | 2010-05-27 | 2010-10-21 | Sumitomo Electric Ind Ltd | Vehicle control system, onboard device and vehicle |
CN102157064A (en) * | 2011-05-26 | 2011-08-17 | 东南大学 | Method for designing signal intersection of bus lanes |
JP2012155769A (en) * | 2007-06-07 | 2012-08-16 | Sumitomo Electric Ind Ltd | Stop position determination device, method and computer program, and traffic index calculation device, method and computer program |
CN104123848A (en) * | 2014-07-14 | 2014-10-29 | 昆明理工大学 | Single intersection oversaturation signal timing method in consideration of widening segment length |
CN104835319A (en) * | 2015-04-07 | 2015-08-12 | 同济大学 | Method for estimating vehicle import behavior on high-grade road bottleneck zone on-ramp |
-
2017
- 2017-03-03 CN CN201710124151.1A patent/CN106846804B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2012155769A (en) * | 2007-06-07 | 2012-08-16 | Sumitomo Electric Ind Ltd | Stop position determination device, method and computer program, and traffic index calculation device, method and computer program |
CN101727746A (en) * | 2009-09-18 | 2010-06-09 | 苏州大学 | Method for dynamically estimating travel time of motor vehicle on urban road under control of signal lamps |
JP2010238247A (en) * | 2010-05-27 | 2010-10-21 | Sumitomo Electric Ind Ltd | Vehicle control system, onboard device and vehicle |
CN102157064A (en) * | 2011-05-26 | 2011-08-17 | 东南大学 | Method for designing signal intersection of bus lanes |
CN104123848A (en) * | 2014-07-14 | 2014-10-29 | 昆明理工大学 | Single intersection oversaturation signal timing method in consideration of widening segment length |
CN104835319A (en) * | 2015-04-07 | 2015-08-12 | 同济大学 | Method for estimating vehicle import behavior on high-grade road bottleneck zone on-ramp |
Non-Patent Citations (1)
Title |
---|
信号交叉口饱和流率动态提取方法;杨东援 等;《交通运输工程学报》;20170228;第13卷(第1期);98-103 * |
Also Published As
Publication number | Publication date |
---|---|
CN106846804A (en) | 2017-06-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106846804B (en) | The real-time saturation volume rate method of estimation in intersection based on hidden Markov chain | |
CN106355907B (en) | Signalized crossing queue length real-time estimation method based on track of vehicle | |
CN106384509A (en) | Urban road driving time distribution estimation method considering taxi operation states | |
CN100573618C (en) | A kind of traffic intersection four-phase vehicle flow detection method | |
CN111368687A (en) | Sidewalk vehicle illegal parking detection method based on target detection and semantic segmentation | |
CN106650913A (en) | Deep convolution neural network-based traffic flow density estimation method | |
CN109859468A (en) | Multilane traffic volume based on YOLOv3 counts and wireless vehicle tracking | |
CN104835319B (en) | Method for estimating vehicle import behavior on high-grade road bottleneck zone on-ramp | |
CN110070734B (en) | Signalized intersection saturated headway estimation method based on Gaussian mixture model | |
CN104809895A (en) | Adjacent intersection arterial road coordinate control model and optimization method thereof | |
CN108492562A (en) | Intersection vehicles trajectory reconstruction method based on fixed point detection with the alert data fusion of electricity | |
CN107133974A (en) | The vehicle type classification method that Gaussian Background modeling is combined with Recognition with Recurrent Neural Network | |
CN106803347B (en) | Urban intersection traffic state judging method based on RFID data | |
CN106297296B (en) | A kind of fine granularity hourage distribution method based on sparse track point data | |
CN109190444A (en) | A kind of implementation method of the lane in which the drivers should pay fees vehicle feature recognition system based on video | |
CN109389826A (en) | A kind of real-time computing technique of signalized intersections saturation volume rate | |
CN102890862B (en) | Traffic condition analyzing device and method based on vector mode | |
CN108701396A (en) | A kind of detection alarm method, storage medium and the server of Chinese herbaceous peony accumulated snow and icing | |
CN103810854B (en) | A kind of based on the artificial intelligent transportation parameter detection method demarcated | |
CN109658694A (en) | A kind of high-risk vehicular traffic stream prediction technique of convolutional neural networks and system | |
CN110444015A (en) | Intelligent network based on no signal crossroad subregion joins car speed decision-making technique | |
CN105551250A (en) | Method for discriminating urban road intersection operation state on the basis of interval clustering | |
CN110738852B (en) | Intersection steering overflow detection method based on vehicle track and long and short memory neural network | |
CN112116813B (en) | State recognition model establishing method, road traffic state recognition method and road traffic state recognition device | |
CN103605960A (en) | Traffic state identification method based on fusion of video images with different focal lengths |
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 |