CN107293118A - A kind of traffic speed motion interval Forecasting Approach for Short-term - Google Patents

A kind of traffic speed motion interval Forecasting Approach for Short-term Download PDF

Info

Publication number
CN107293118A
CN107293118A CN201710584159.6A CN201710584159A CN107293118A CN 107293118 A CN107293118 A CN 107293118A CN 201710584159 A CN201710584159 A CN 201710584159A CN 107293118 A CN107293118 A CN 107293118A
Authority
CN
China
Prior art keywords
mrow
traffic speed
interval
msub
order difference
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
CN201710584159.6A
Other languages
Chinese (zh)
Other versions
CN107293118B (en
Inventor
聂庆慧
邓社军
周扬
肖枭
于世军
刘路
张鹏鹏
谈圣
黄佳宇
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.)
Yangzhou University
Original Assignee
Yangzhou 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 Yangzhou University filed Critical Yangzhou University
Priority to CN201710584159.6A priority Critical patent/CN107293118B/en
Publication of CN107293118A publication Critical patent/CN107293118A/en
Application granted granted Critical
Publication of CN107293118B publication Critical patent/CN107293118B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/01Detecting movement of traffic to be counted or controlled
    • G08G1/0104Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125Traffic data processing
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation, e.g. linear programming, "travelling salesman problem" or "cutting stock problem"

Abstract

The present invention discloses a kind of traffic speed motion interval Forecasting Approach for Short-term of high reliability, comprises the following steps:(10) traffic speed time series is obtained:Obtain target section traffic speed time series observation on road;(20) stationary time series is obtained:By first-order difference computing, traffic speed time series is converted into stationary time series;(30) first-order difference predictor calculation:According to traffic speed first-order difference time series predicting model, traffic speed first-order difference predicted value in each current time interval is calculated;(40) residual error standard deviation predictor calculation:According to the comprehensive generilized auto regressive conditional heteroskedastic forecast model of residual error, residual error standard deviation predicted value in each current time interval is calculated;(50) target section traffic speed forecast interval is determined:According to traffic speed observation, traffic speed first-order difference predicted value and residual error a standard deviation predicted value, traffic speed forecast interval of the target section in each time interval is determined.

Description

A kind of traffic speed motion interval Forecasting Approach for Short-term
Technical field
The invention belongs to short-term traffic flow prediction technical field, the high traffic speed motion interval of particularly a kind of reliability Forecasting Approach for Short-term.
Background technology
The traffic flow speed of service is one of important technology index of road traffic operation, management and control.Accurately, reliably Traffic speed short-term prediction is had become in the important research of City ITS such as paths chosen, active traffic control Hold.
Numerous studies have been carried out to road section traffic speed short-term prediction technology both at home and abroad.Based on statistical model and people The Forecasting Methodology of the technologies such as work intelligence is constantly suggested, and the accuracy predicted also is continuously available raising.
However, most of researchs all just for traffic speed first moment video sequence value carry out the modelings of Forecasting Methodology with Assess, ignore the wave characteristic of traffic flow speed second moment, it is difficult to effectively quantify traffic speed forecasting reliability.Although There is a small amount of volatility modeling researched and proposed to the second moment of traffic speed, but using the mould for formulating parameter more than existing method Type, strongly limit the description to traffic speed second moment dynamic structure.
In a word, the problem of prior art is present be:The reliability of traffic speed short-term prediction is low.
The content of the invention
It is an object of the invention to provide a kind of traffic speed motion interval Forecasting Approach for Short-term, reliability is high.
The technical solution for realizing the object of the invention is:
A kind of traffic speed motion interval Forecasting Approach for Short-term, comprises the following steps:
(10) traffic speed time series is obtained:Obtain target section traffic speed time series observation on road;
(20) stationary time series is obtained:By first-order difference computing, when traffic speed time series is converted into steady Between sequence;
(30) first-order difference predictor calculation:According to traffic speed first-order difference time series predicting model, calculate respectively when Traffic speed first-order difference predicted value in preceding time interval t;
(40) residual error standard deviation predictor calculation:According to the comprehensive generilized auto regressive conditional heteroskedastic prediction mould of residual error Type, calculates residual error standard deviation predicted value in each current time interval t;
(50) target section traffic speed forecast interval is determined:According to the traffic speed in each previous interval (t-1) Traffic speed first-order difference predicted value and each current time interval residual error standard deviation are pre- in observation, each current time interval t Measured value, determines traffic speed forecast interval of the target section in each time interval t.
Compared with prior art, its remarkable advantage is the present invention:
Reliability is high.The present invention is built in road target section traffic speed first moment video sequence short-time forecasting model On the basis of, the wave characteristic of traffic speed second moment is further extracted and quantified, i.e., the second moment of residual sequence is constructed Conditional heterosedasticity forecast model, by the prediction to second moment dynamic standard difference, is realized to traffic speed motion interval in short-term Prediction, further improves the reliability of target section traffic speed short-term prediction.
The present invention is described in further detail with reference to the accompanying drawings and detailed description.
Brief description of the drawings
Fig. 1 is the main flow chart of traffic speed motion interval Forecasting Approach for Short-term of the present invention.
Fig. 2 is that No. 1012016 sections of embodiment are based on standard GARCH (1,1), GJR-GARCH (1,1), fGARCH (1,1) The traffic speed first-order difference series error standard deviation of model predicts the outcome comparison diagram.
Fig. 3 is that No. 1004030 sections of embodiment are based on standard GARCH (1,1), GJR-GARCH (1,1), fGARCH (1,1) The traffic speed first-order difference series error standard deviation of model predicts the outcome comparison diagram.
Fig. 4 is that No. 1001010 sections of embodiment are based on standard GARCH (1,1), GJR-GARCH (1,1), fGARCH (1,1) The traffic speed first-order difference series error standard deviation of model predicts the outcome comparison diagram.
Fig. 5 is that No. 1003006 sections of embodiment are based on standard GARCH (1,1), GJR-GARCH (1,1), fGARCH (1,1) The traffic speed first-order difference series error standard deviation of model predicts the outcome comparison diagram.
Embodiment
As shown in figure 1, traffic speed motion interval Forecasting Approach for Short-term of the present invention, comprises the following steps:
(10) traffic speed time series is obtained:Obtain target section traffic speed time series observation on road;
The traffic speed data of target section collection are the continuous time series data with 5 minutes for constant duration, and And original time series is unstable.
(20) stationary time series is obtained:By first-order difference computing, when traffic speed time series is converted into steady Between sequence;
(30) first-order difference predictor calculation:According to traffic speed first-order difference time series predicting model, calculate respectively when Traffic speed first-order difference predicted value in preceding time interval t;
(30) the first-order difference predictor calculation step is specially:
According to traffic speed first-order difference time series predicting model, m=max (p, q) is made, historical time intervals are obtained (t-1), the first-order difference value of (t-2) ..., the target section traffic speed time series up to (t-m), when calculating current Between interval t in traffic speed first-order difference predicted value be,
In formula,The first-order difference predicted value for being target section traffic speed in current time interval t, Δ yt-iFor mesh First-order difference observation of the section traffic speed in previous interval (t-i) is marked, c is constant term;P is autoregressive process Lag order, q is the lag order of moving average process, φiAnd θjFor auto regressive moving average ARMA (p, q) model coefficient, εt The residual error for being traffic speed first-order difference sequence in current time interval t, εt-jIt is traffic speed first-order difference sequence preceding Residual error in one time interval (t-j), and assume series { εtIt is the white-noise process for obeying 0 average normal distribution.
The Autoregressive p and moving average order q of ARMA (p, q) model are determined by bayesian information criterion;Constant Item c and model coefficient φi、θjObtained using Least Square Method.
(40) residual error standard deviation predictor calculation:According to the comprehensive generilized auto regressive conditional heteroskedastic prediction mould of residual error Type, calculates residual error standard deviation predicted value in each current time interval t;
(40) residual error standard deviation predictor calculation step is specially:According to the comprehensive broad sense autoregressive conditions of residual error Singular variance forecast model, the predicted value for calculating the residual error in time interval t standard deviation is
I.e.
In formula,
f(εt-1)=| εt-1-b|-c(εt-1- b),
The comprehensive generilized auto regressive conditional heteroskedastic forecast model of residual error is, according to residual error sequence { εtGained, Synthesis generilized auto regressive conditional heteroskedastic fGARCH (1,1) model with the dynamic average item of first-order autoregression and single order,
Wherein, εtFor the residual values of traffic speed first-order difference time series in current time interval t, εt-1For it is previous when Between be spaced (t-1) interior traffic speed first-order difference time series residual values,For residual error standard deviation in current time interval t Predicted value,For the predicted value of the interior residual error standard deviation of previous interval (t-1), residual error sequence { εtEqual to obey Value is that 0, standard deviation is σtNormal distribution;ztTo obey the white noise mistake that average is the independent standard normal distribution that 0, variance is 1 Journey;ω, β, γ are regression parameter;λ is Box-Cox transfer ratios;B is displacement factor, for quantifying less traffic speed ripple It is dynamic;C is twiddle factor, for quantifying larger traffic speed fluctuation;ω, β, γ, λ, b, c are fGARCH (1,1) model Parameter to be estimated.
The parameter ω to be estimated of fGARCH (1,1) model, beta, gamma, λ, b, c estimates to obtain using maximum-likelihood method.
(50) target section traffic speed forecast interval is determined:According to the traffic speed in each previous interval (t-1) Traffic speed first-order difference predicted value and each current time interval residual error standard deviation are pre- in observation, each current time interval t Measured value, determines traffic speed forecast interval of the target section in each time interval t.
(50) the target section traffic speed forecast interval determines that step is specially:
The interval prediction value of traffic speed is in preceding time interval t
Wherein, the traffic speed prediction higher limit in current time interval t is
Traffic speed in current time interval t predicts that lower limit is
In formula, yt-1For the traffic speed observation in previous interval (t-1),To be handed in current time interval t Logical speed first-order difference predicted value,In current time interval t
Residual error standard deviation predicted value, zα/2For the upper α quantiles of standardized normal distribution.
Describe the use process of the present invention in detail with specific embodiment below.
In the present embodiment, the data used is each 2 on the city of Kunshan inner city major trunk roads and subsidiary road of actual acquisition The traffic speed time series of section.Wherein major trunk roads detection section numbering be respectively 1012016 (Bai Lulu), it is 1004030 (preceding Route), subsidiary road detection section numbering is respectively 1001010 (Xiao Linlu), 1003006 (Tong Fenglu).The collection of initial data Time range is on July 22,21 days to 2014 July in 2014, and the acquisition time of data was at intervals of 5 minutes.The data gathered In, the data on July 21st, 2014 are used for model construction and parameter Estimation, and the data on July 22nd, 2014 are used for estimated performance Assess.
Original (level) time series of the traffic speed of target section is labeled as { y by the present embodimentt}.Target section is handed over The leveled time sequence of logical speed carries out first-order difference computing, by original non-stationary { ytIt is converted into the stable time Sequence { Δ yt}.Autoregression movement is built to the section traffic flow speed first-order difference time series that on July 21st, 2014 obtains Average ARMA (p, q) model, expression formula is as follows:
In formula (1)The first-order difference predicted value for being target section traffic flow speed in time interval t;Δyt-iFor First-order difference observation of the target section traffic flow speed in time interval (t-i);εtFor traffic speed first-order difference sequence Residual error in time interval t, εt-jThe residual error for being traffic speed first-order difference sequence in time interval (t-j) c, φi、θjFor the parameter to be estimated of ARMA (p, q) model;P and q is that the respectively Autoregressive of ARMA (p, q) model and movement are flat Equal exponent number, is determined, its result is given in Table 1 by bayesian information criterion.
The autoregression of table 1 target section ARMA (p, q) model and moving average order
It is determined that on the basis of the autoregression of each target section ARMA (p, q) model and moving average order, further The parameter of ARMA (p, q) model, which is entered, to be predicted to each target section traffic speed first-order difference serial mean using least square method Row estimation, as a result as shown in table 2.
Table 2 target section ARMA (p, q) model parameter estimation
According to ARMA (p, q) model parameter estimation knot shown in formula (1) fruit and table 2, acquisition time interval t can be calculated The mean prediction result of each interior target section traffic speed first-order difference sequence is respectively:
Section 1012016
Section 1004030
Section 1001010
Section 1003006
On the basis of each target section ARMA (p, q) model construction and parameter Estimation is completed, ARMA (p, q) mould is extracted The residual sequence of type, and fGARCH (1,1) model of residual sequence is built, expression formula is as follows
FGARCH (1,1) model coefficient is estimated using Maximum Likelihood Estimation Method, as a result as shown in table 3.
Table 3 target section fGARCH (1,1) model parameter estimation
According to fGARCH (1,1) model parameter estimation result shown in formula (6) and table 3, acquisition time interval t can be calculated Each interior target section traffic speed first-order difference series error standard deviation predicts the outcome respectively:
Section 1012016
Wherein, f (εt-1)=| εt-1-0.93|+0.44(εt-1-0.93)
Section 1004030
Wherein, f (εt-1)=| εt-1-1.24|+0.82(εt-1-1.24)
Section 1001010
Wherein, f (εt-1)=| εt-1-1.32|+0.77(εt-1-1.32)
Section 1003006
Wherein, f (εt-1)=| εt-1-0.61|+0.64(εt-1-0.61)
Obtaining the mean prediction result of traffic speed first-order differenceAnd its residual sequence standard deviation predicted valueBar Under part, the present embodiment gives level of significance α=0.05, that is, when specifying 95% confidence level, can calculate acquisition time interval t The prediction higher limit of interior traffic speed first-order difference sequence is
Lower limit is
On this basis, further calculate that target section traffic speed prediction higher limit is in time interval t
Lower limit is
Target section traffic speed prediction average in time interval t, which can finally be obtained, isForecast interval For
The present embodiment is estimated using average confidence interval width ACL to the interval prediction performance of traffic speed, expression Formula is shown in formula (15).
In formula (15), n is sample number;CLtFor traffic speed forecast interval width in time interval t, and
In order to be compared with traditional Forecasting Methodology, present case give simultaneously based on standard GARCH (1,1) models and The interval prediction performance of GJR-GARCH (1,1) model.For standard GARCH (1,1) model, its model coefficient λ=2, b =c=0, method of estimation same fGARCH (1,1) model of other coefficients;For GJR-GARCH (1,1) model, its model Coefficient lambda=2, b=0, method of estimation same fGARCH (1,1) model of other coefficients.In addition, in order to compare the heavy traffic period and The estimated performance of the different models of the non-peak hours/period of traffic, present case by assess in the period 6:30AM~9:30PM. is divided into friendship Logical peak hours/period, other remaining Time segments divisions are non-peak hours/period.Table 4 gives standard GARCH (1,1) model, GJR-GARCH The interval prediction performance of (1,1) model and fGARCH (1,1) model.
The standard GARCH (1,1) of table 4, GJR-GARCH (1,1), fGARCH (1,1) model interval prediction performance comparison
The result provided from table 4 can be seen that under conditions of identical confidence level is given, fGARCH (1,1) model The interval width of consensus forecast is less than other two class models, is especially embodied in when the traffic that fluctuation is more difficult to quantization is non-busy Section.It can be seen that the motion interval short-term prediction of use fGARCH (1,1) model progress traffic speed is resulted in preferably, prediction can By property.In addition, present case give above-mentioned 4 targets section standard GARCH (1,1) model, GJR-GARCH (1,1) models and The prediction standard deviation of the traffic speed first-order difference series error of fGARCH (1,1) model, is shown in Fig. 2 to Fig. 5.Can from figure Intuitively to find out, the target section traffic speed first-order difference series error standard deviation based on fGARCH (1,1) model Predicted value generally has less fluctuation, is especially embodied in the non-peak hours/period of night traffic.

Claims (4)

1. a kind of traffic speed motion interval Forecasting Approach for Short-term, it is characterised in that comprise the following steps:
(10) traffic speed time series is obtained:Obtain target section traffic speed time series observation on road;
(20) stationary time series is obtained:By first-order difference computing, traffic speed time series is converted into stationary time sequence Row;
(30) first-order difference predictor calculation:According to traffic speed first-order difference time series predicting model, when calculating each current Between interval t in traffic speed first-order difference predicted value;
(40) residual error standard deviation predictor calculation:According to the comprehensive generilized auto regressive conditional heteroskedastic forecast model of residual error, meter Calculate residual error standard deviation predicted value in each current time interval;
(50) target section traffic speed forecast interval is determined:According to each previous interval) in traffic speed observation, each Traffic speed first-order difference predicted value and each current time interval residual error standard deviation predicted value, determine mesh in current time interval Mark traffic speed forecast interval of the section in each time interval.
2. Forecasting Methodology according to claim 1, it is characterised in that (30) the first-order difference predictor calculation step tool Body is:
According to traffic speed first-order difference time series predicting model, m=max (p, q) is made, historical time intervals (t-1) are obtained, (t-2) the first-order difference value of the target section traffic speed time series ..., up to (t-m), calculates current time interval t Interior traffic speed first-order difference predicted value is,
In formula,The first-order difference predicted value for being target section traffic speed in current time interval t, Δ yt-iIt is disconnected for target First-order difference observation of the face traffic speed in previous interval (t-i), c is constant term;P is delayed for autoregressive process Exponent number, q is the lag order of moving average process, φiAnd θjFor auto regressive moving average ARMA (p, q) model coefficient, εtTo hand over Residual error of the logical speed first-order difference sequence in current time interval t, εt-jIt is traffic speed first-order difference sequence when previous Between residual error in interval (t-j), and assume series { εtIt is the white-noise process for obeying 0 average normal distribution.
3. Forecasting Methodology according to claim 2, it is characterised in that (40) the residual error standard deviation predictor calculation step It is rapid to be specially:
According to the comprehensive generilized auto regressive conditional heteroskedastic forecast model of residual error, the residual error standard deviation in time interval t is calculated Predicted value be
<mrow> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mi>t</mi> <mi>&amp;lambda;</mi> </msubsup> <mo>=</mo> <mi>&amp;omega;</mi> <mo>+</mo> <mi>&amp;beta;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> <mi>f</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>&amp;epsiv;</mi> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> <mi>&amp;lambda;</mi> </msup> <mo>+</mo> <mi>&amp;gamma;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> <mo>,</mo> </mrow>
I.e.
<mrow> <msub> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mi>t</mi> </msub> <mo>=</mo> <msup> <mi>e</mi> <mfrac> <mrow> <mi>ln</mi> <mrow> <mo>(</mo> <mi>&amp;omega;</mi> <mo>+</mo> <mi>&amp;beta;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> <mi>f</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>&amp;epsiv;</mi> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> <mi>&amp;lambda;</mi> </msup> <mo>+</mo> <mi>&amp;gamma;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mi>&amp;lambda;</mi> </mfrac> </msup> <mo>,</mo> </mrow>
In formula,
f(εt-1)=| εt-1-b|-c(εt-1- b),
The comprehensive generilized auto regressive conditional heteroskedastic forecast model of residual error is, according to residual error sequence { εtGained, with one Rank autoregression and synthesis generilized auto regressive conditional heteroskedastic fGARCH (1,1) model of single order rolling average,
<mrow> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <msub> <mi>&amp;epsiv;</mi> <mi>t</mi> </msub> <mo>=</mo> <msub> <mi>z</mi> <mi>t</mi> </msub> <msub> <mi>&amp;sigma;</mi> <mi>t</mi> </msub> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msub> <mi>z</mi> <mi>t</mi> </msub> <mo>~</mo> <mi>I</mi> <mi>I</mi> <mi>N</mi> <mrow> <mo>(</mo> <mn>0</mn> <mo>,</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mi>t</mi> <mi>&amp;lambda;</mi> </msubsup> <mo>=</mo> <mi>&amp;omega;</mi> <mo>+</mo> <mi>&amp;beta;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> <mi>f</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>&amp;epsiv;</mi> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> <mi>&amp;lambda;</mi> </msup> <mo>+</mo> <mi>&amp;gamma;</mi> <msubsup> <mover> <mi>&amp;sigma;</mi> <mo>^</mo> </mover> <mrow> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>&amp;lambda;</mi> </msubsup> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mi>f</mi> <mrow> <mo>(</mo> <msub> <mi>&amp;epsiv;</mi> <mi>t</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mo>|</mo> <mrow> <msub> <mi>&amp;epsiv;</mi> <mi>t</mi> </msub> <mo>-</mo> <mi>b</mi> </mrow> <mo>|</mo> <mo>-</mo> <mi>c</mi> <mrow> <mo>(</mo> <msub> <mi>&amp;epsiv;</mi> <mi>t</mi> </msub> <mo>-</mo> <mi>b</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>,</mo> </mrow> 1
Wherein, εtFor the residual values of traffic speed first-order difference time series in current time interval t, εt-1For previous interval (t-1) residual values of interior traffic speed first-order difference time series,For the prediction of residual error standard deviation in current time interval t Value,For the predicted value of the interior residual error standard deviation of previous interval (t-1), residual error sequence { δtBe obey average be 0, Standard deviation is σtNormal distribution;ztTo obey the white-noise process that average is the independent standard normal distribution that 0, variance is 1;ω、 β, γ are regression parameter;λ is Box-Cox transfer ratios;B is displacement factor, for quantifying less traffic speed fluctuation;C is Twiddle factor, for quantifying larger traffic speed fluctuation.ω, β, γ, λ, b, c be fGARCH (1,1) model wait estimate ginseng Number.
4. Forecasting Methodology according to claim 3, it is characterised in that (50) the target section traffic speed forecast interval Determine that step is specially:
The interval prediction value of traffic speed is in current time interval t
Wherein, the traffic speed prediction higher limit in current time interval t is
Traffic speed in current time interval t predicts that lower limit is
In formula, yt-1For the traffic speed observation in previous interval (t-1),For traffic speed in current time interval t First-order difference predicted value is spent,Residual error standard deviation predicted value, z in current time interval tα/2For upper α points of standardized normal distribution Site.
CN201710584159.6A 2017-07-18 2017-07-18 Short-time prediction method for traffic speed dynamic interval Active CN107293118B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710584159.6A CN107293118B (en) 2017-07-18 2017-07-18 Short-time prediction method for traffic speed dynamic interval

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710584159.6A CN107293118B (en) 2017-07-18 2017-07-18 Short-time prediction method for traffic speed dynamic interval

Publications (2)

Publication Number Publication Date
CN107293118A true CN107293118A (en) 2017-10-24
CN107293118B CN107293118B (en) 2020-10-20

Family

ID=60101465

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710584159.6A Active CN107293118B (en) 2017-07-18 2017-07-18 Short-time prediction method for traffic speed dynamic interval

Country Status (1)

Country Link
CN (1) CN107293118B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108629977A (en) * 2018-06-06 2018-10-09 上海城市交通设计院有限公司 Trip characteristics analysis method based on vehicle electron identifying technology
CN109727455A (en) * 2019-03-05 2019-05-07 湖北汇程信息技术有限公司 A kind of processing method of traffic information
CN110517488A (en) * 2019-08-19 2019-11-29 南京理工大学 The Short-time Traffic Flow Forecasting Methods with Recognition with Recurrent Neural Network are decomposed based on timing
WO2020093357A1 (en) * 2018-11-07 2020-05-14 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for speed prediction

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080033630A1 (en) * 2006-07-26 2008-02-07 Eun-Mi Lee System and method of predicting traffic speed based on speed of neighboring link
CN103903452A (en) * 2014-03-11 2014-07-02 东南大学 Traffic flow short time predicting method
CN105118293A (en) * 2015-09-16 2015-12-02 东南大学 Road section traffic speed short-term prediction method in view of long-term equilibrium relationship
WO2016036024A1 (en) * 2014-09-01 2016-03-10 한국교통연구원 Traffic link speed prediction method and apparatus for same
CN106448151A (en) * 2016-07-07 2017-02-22 河南理工大学 Short-time traffic flow prediction method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080033630A1 (en) * 2006-07-26 2008-02-07 Eun-Mi Lee System and method of predicting traffic speed based on speed of neighboring link
CN103903452A (en) * 2014-03-11 2014-07-02 东南大学 Traffic flow short time predicting method
WO2016036024A1 (en) * 2014-09-01 2016-03-10 한국교통연구원 Traffic link speed prediction method and apparatus for same
CN105118293A (en) * 2015-09-16 2015-12-02 东南大学 Road section traffic speed short-term prediction method in view of long-term equilibrium relationship
CN106448151A (en) * 2016-07-07 2017-02-22 河南理工大学 Short-time traffic flow prediction method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
聂庆慧,夏井新,钱振东: "城市道路交通流短时预测及可靠性分析", 《西南交通大学学报》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108629977A (en) * 2018-06-06 2018-10-09 上海城市交通设计院有限公司 Trip characteristics analysis method based on vehicle electron identifying technology
WO2020093357A1 (en) * 2018-11-07 2020-05-14 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for speed prediction
US11004335B2 (en) 2018-11-07 2021-05-11 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for speed prediction
CN109727455A (en) * 2019-03-05 2019-05-07 湖北汇程信息技术有限公司 A kind of processing method of traffic information
CN110517488A (en) * 2019-08-19 2019-11-29 南京理工大学 The Short-time Traffic Flow Forecasting Methods with Recognition with Recurrent Neural Network are decomposed based on timing

Also Published As

Publication number Publication date
CN107293118B (en) 2020-10-20

Similar Documents

Publication Publication Date Title
CN107293118A (en) A kind of traffic speed motion interval Forecasting Approach for Short-term
CN104408913B (en) A kind of traffic flow three parameter real-time predicting method considering temporal correlation
CN102693633B (en) Short-term traffic flow weighted combination prediction method
CN102763048B (en) Using adaptability prediction algorithm and when determine the method and apparatus of using adaptability prediction algorithm in virtual measurement
CN102629418B (en) Fuzzy kalman filtering-based traffic flow parameter prediction method
CN103903430B (en) Dynamic fusion type travel time predicting method with multi-source and isomorphic data adopted
CN104699894A (en) JITL (just-in-time learning) based multi-model fusion modeling method adopting GPR (Gaussian process regression)
CN102521989B (en) Dynamic-data-driven highway-exit flow-quantity predicting method
CN105260607A (en) Serial connection and parallel connection coupling multi-model hydrological forecasting method
CN105741549B (en) A kind of traffic flow forecasting method theoretical based on space copula
CN103985000A (en) Medium-and-long term typical daily load curve prediction method based on function type nonparametric regression
CN108846517B (en) Integration method for predicating quantile probabilistic short-term power load
CN105303835B (en) A kind of Forecasting Approach for Short-term of road traffic stream mode
CN105118293B (en) A kind of road section traffic speed Forecasting Approach for Short-term for considering long-run equilibrium relation
Li et al. A k-nearest neighbor locally weighted regression method for short-term traffic flow forecasting
CN104537233B (en) A kind of power distribution network puppet based on Density Estimator measures generation method
Ozoh et al. A comparative analysis of techniques for forecasting electricity consumption
CN107481523A (en) A kind of traffic flow speed Forecasting Methodology and system
CN107591001A (en) Expressway Traffic Flow data filling method and system based on on-line proving
CN104008433A (en) Method for predicting medium-and-long-term power loads on basis of Bayes dynamic model
CN103870890A (en) Prediction method for traffic flow distribution of expressway network
CN103544362B (en) A kind of harmonic wave medium-and long-term forecasting method based on two-dimensional curve prediction
CN104463347A (en) Method for electronic product degenerate state trend prediction with singular signals
CN102930353B (en) The Forecasting Methodology of a kind of community power load and device
CN105046089B (en) A kind of method for predicting heavy showers and flood

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