CN106548625B - A kind of urban highway traffic situation combination forecasting method - Google Patents
A kind of urban highway traffic situation combination forecasting method Download PDFInfo
- Publication number
- CN106548625B CN106548625B CN201611114072.4A CN201611114072A CN106548625B CN 106548625 B CN106548625 B CN 106548625B CN 201611114072 A CN201611114072 A CN 201611114072A CN 106548625 B CN106548625 B CN 106548625B
- Authority
- CN
- China
- Prior art keywords
- traffic
- formula
- traffic condition
- average speed
- moment
- 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
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0133—Traffic data processing for classifying traffic situation
Landscapes
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of urban highway traffic situation combination forecasting methods.The present invention uses maximin technique estimation current time traffic condition sequence and neighbouring all similarities with the magnitude of traffic flow and average speed between Sunday traffic condition sequence in the same time first, if all similarities calculate the magnitude of traffic flow and average speed of subsequent time using all similarities in tolerance interval；If all similarities in tolerance interval, the magnitude of traffic flow and average speed of prediction time are not calculated using the prediction algorithm based on fractal theory；Finally, using the traffic congestion index of relative evaluation method estimation current road segment prediction time.The algorithm of fractal dimension is based on structure function method in the present invention, it is contemplated that the traffic condition situation of the different time intervals on the basis of prediction time, and repaired with special weighting coefficient.The present invention has the advantages that calculation amount elasticity is big, precision of prediction is high, the traffic state data of output is comprehensive.
Description
Technical field
The present invention relates to a kind of urban highway traffic situation combination forecasting methods, more specifically, more particularly to a kind of base
In the urban highway traffic situation combination forecasting method of urban road traffic flow Weekly similarity and fractal characteristic.
Background technique
In intelligent transportation system, the prediction of urban road traffic state is in highly important status.Accurate traffic
State is to formulate the basis of suitable traffic management measure, and the multiple function of intelligent transportation system is all the state with road traffic
It is predicted as center deployment, traffic status prediction method passes through to current road segment relevant historical traffic data and Real-time Traffic Information
Effective analysis can be handed over for city intelligent road thus the method for obtaining the overall operation state of current road segment subsequent time
Logical management system provides strong technical support.
Be divided into three categories generally about the method for traffic status prediction at present, can be divided into three classes: one kind is with mathematics
Prediction model based on the traditional mathematics such as statistics and calculus and physical method, such prediction model often have complexity
High, computationally intensive feature；One kind is to be with modern science and technology and method (such as analogue technique, neural network, fuzzy control)
Main research means and the prediction model formed, its main feature is that used model and method do not pursue proper mathematics
Derivation and specific physical significance, and more pay attention to the fitting effect to true traffic flow phenomenon；One kind is the group for combining the two
Prediction model is closed, 1969, J.N.Bates and C.W.J.Granger were put forward for the first time the theory and method of combined prediction, will not
Same prediction technique is combined, in the hope of generating preferable prediction effect.Due to the non-linear, complicated of road traffic system and not
Deterministic essential characteristic, combination forecasting, as having the combination of model algorithm and model-free algorithm, neural network theory and something lost
The combination of propagation algorithm, fuzzy theory, wavelet theory, spectrum analysis etc. obtains more and more extensive research and application.
Fractal theory is a branch in non-linear scientific theory, to describe the rule of complicated, chaos phenomenon behind
Property, disclose the relationship between part and entirety.Quantizating index --- fractal dimension can not only be determined a key in its theory
The complexity of amount ground description things, and its variation tends to illustrate the change of certain characteristic of things.Currently, dividing the general of shape
It reads and thought is abstracted as a kind of methodology by people, fractal prediction is intended to seem unordered, higher-dimension, dynamic from information world
Data in by fractal dimension find out in it regularity, and detect according to the situation of change of fractal dimension the dynamic of environment
Change to predict future.
There are Fractal Phenomenons under certain time scale for traffic flow and average speed, excavate traffic flow using fractal method
The inherent law of time series avoids some difficulties of problem analysis bring of starting with from the influence factor of traffic flow.But
The time that short-term prediction needs to predict is shorter, divides the self-similarity of shape weaker, this point needs to improve during use.
Summary of the invention
The present invention in order to overcome the shortcomings of the above technical problems, provides a kind of urban highway traffic situation combined prediction side
Method.
Urban highway traffic situation combination forecasting method of the invention, which is characterized in that realized by following steps:
A) sets current time as t moment, to predict the traffic condition of t+ time Δt, obtains α forward since t moment
The traffic state data at moment forms current t moment traffic condition sequence Zt={ zt-(α-1)Δt,zt-(α-2)Δt,...,zt-2Δt,
zt-Δt,zt, wherein the traffic state data z at each momenti=[qi,vi], t- α Δ t+ Δ t≤i≤t, qiIndicate the friendship at i moment
Through-current capacity, viIndicate the average speed at i moment；
B) chooses the traffic condition sequence of m neighbouring same Sundays in the same time from historical data base, and forms set A
={ Z1t,Z2t,...,Zmt, wherein Zjt={ zj(t-(α-1)Δt), zj(t-(α-2)Δt),...,zj(t-2Δt),zj(t-Δt),zj(t), 1≤j
≤ m indicates j-th of traffic condition sequence with Sunday in the same time in set A；zj(i)=[qj(i), vj(i)], t- α Δ t+ Δ t≤i
≤ t, qj(i)Indicate j-th of magnitude of traffic flow with the i moment in Sunday in the same time traffic condition sequence, vj(i)Indicate j-th of same Sunday
In the same time in traffic condition sequence the i moment average speed；
C) calculates current t moment traffic condition sequence and the m traffic conditions with Sunday in the same time with maximin technique
All similarities between sequence traffic condition pass through step c-1) to step c-2) Lai Shixian:
C-1) calculates all similarities of the magnitude of traffic flow, calculates current t moment traffic condition sequence and m according to formula (1)
With all similarity Rq of the magnitude of traffic flow between the traffic condition sequence of Sunday in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
C-2) calculates all similarities of average speed, calculates current t moment traffic condition sequence and m according to formula (2)
With all similarity Rv of average speed between the traffic condition sequence of Sunday in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
D) screens m with the traffic condition sequence of Sunday in the same time using formula (3),
Rqj≥Sq∩Rvj≥Sv (3)
In formula, SqIndicate the threshold value of magnitude of traffic flow week similarity, SvIndicate the threshold value of average speed week similarity；It will
All traffic condition Sequence composition set B={ Z for meeting formula (3)1t, Z2t..., Zht},Zxt={ zx(t-(n-1)Δt),
zx(t-(n-2)Δt)..., zx(t-2Δt),zx(t-Δt),zx(t)Indicate x-th of traffic condition sequence with Sunday in the same time in set B,
1≤x≤h；
When if there is meeting the sequence of formula (3), i.e. h > 0, then follow the steps e), to calculate moment t+ Δ t's to be predicted
The magnitude of traffic flow and average speed；If satisfactory traffic condition sequence is not present in set B, then follow the steps f)；
E) calculates separately the traffic flow forecasting value q ' of prediction time t+ Δ t according to formula (4) and formula (5)t+ΔtPeace
The predicted value v ' of equal speedt+Δt:
Wherein, Rqx、RvxRespectively indicate x-th of traffic condition sequence in the traffic condition sequence and set B of current t moment
Between the magnitude of traffic flow all similarities, all similarities of average speed；qx(t+Δt)、vx(t+Δt)X-th of traffic in respectively set B
The practical angle value of the actual value of the magnitude of traffic flow of situation sequence t+ time Δt, average speed；Then step g) is executed；
F) calculates the magnitude of traffic flow of prediction time t+ Δ t and the prediction of average speed using the algorithm based on fractal theory
Value；
G) then., the traffic density of t+ Δ t prediction time current road segment is calculated by formula (6):
k′t+Δt=q 't+Δt/v′t+Δt (6)
If the traffic congestion density of current road segment is Kstop, value range isKsmoothIt indicates
The optimum density of current road segment, value range areT+ Δ t is then calculated according to relative evaluation method
The traffic congestion index B ' of prediction time current road segmentt+Δt, formula is as follows:
H) exports t+ Δ t prediction time current road segment traffic condition predictions vector R 't+Δt=[q 't+Δt,v′t+Δt,k
′t+Δt,B′t+Δt]。
Urban highway traffic situation combination forecasting method of the invention, based on the algorithm of fractal theory described in step f)
The magnitude of traffic flow of prediction time t+ Δ t and the predicted value of average speed are calculated, is realized by following steps:
F-1) fractal prediction algorithm is on the basis of prediction time t+ Δ t, using the traffic data currently acquired, respectively with
Δt、2Δt、...、iΔt、...、λmaxΔ t is that time interval constructs λmaxA traffic condition sequence, and it is combined into collection
It closesThere is the traffic data at n+1 time point in each traffic condition sequence；Set Z institute
Shown in the traffic condition sequence such as expression formula (8) of expression:
Wherein, ZλΔtThe traffic condition of middle each time point includes the magnitude of traffic flow and average speed, λ=1,2 ..., λmax；
F-2) is the calculation formula of basic construction analysis dimension with structure function method, passes through step f-2-1) and step f-
2-2) find out the fractal dimension D of the magnitude of traffic flowqWith the fractal dimension D of average speedv:
F-2-1) is for sequence Z each in set ZλΔt, λ=1,2 ..., λmax, using formula (9) and formula (10), divide
The arithmetic mean of instantaneous value of the difference side for the traffic condition sequence that each time interval is λ Δ t is not found out:
In formula, Sq(λΔt)、Sv(λ Δ t) respectively indicates traffic condition sequence Z in set ZλΔtThe middle magnitude of traffic flow, average speed
The arithmetic mean of instantaneous value of the difference side of degree；
F-2-2) is by [λ Δ t, Sq(λ Δ t)] draw [In (λ Δ t), In (Sq(λ Δ t))] on double logarithmic chart, utilize
Least square method linear regression algorithm finds out the fractal dimension D of the magnitude of traffic flow in formula (11)qWith constant CqValue；
Equally, by [λ Δ t, Sv(λ Δ t)] draw [In (λ Δ t), In (Sv(λ Δ t))] on double logarithmic chart, using most
Small square law linear regression algorithm finds out the fractal dimension D of average speed in formula (12)vWith constant CvValue:
F-3) utilizes fractal dimension Dq, constant CqWith formula (9), formula (11), each traffic condition sequence in set of computations Z
Arrange ZλΔtIn the previous moment magnitude of traffic flow adjacent with t+ Δ t prediction time predicted value q 't-λΔt+Δt, λ=1,2 ..., λmax,
Seek the distance between its true value and predicted value dqλ, λ=1,2 ..., λmax；
Equally, fractal dimension D is utilizedv, constant CvWith formula (10), formula (12), each traffic condition in set of computations Z
Sequence ZλΔtIn the previous moment average speed adjacent with t+ Δ t prediction time predicted value v 't-λΔt+Δt, λ=1,2 ...,
λmax, seek the distance between its true value and predicted value dvλ, λ=1,2 ..., λmax；
F-4) is by fractal dimension DqWith constant CqValue bring each traffic condition sequence in formula (13) set of computations Z into
ZλΔtThe predicted value q ' of middle t+ time Δt traffic flowλ(t+Δt), λ=1,2 ..., λmax:
Equally, by fractal dimension DvWith constant CvValue bring each traffic condition sequence in formula (14) set of computations Z into
ZλΔtThe predicted value v ' of middle t+ time Δt average speedλ(t+Δt), λ=1,2 ..., λmax:
F-5) calculates separately out the pre- of the current road segment t+ Δ t prediction time magnitude of traffic flow according to formula (15), formula (16)
The predicted value of measured value and average speed:
In formula, dqλIndicate each traffic condition sequence Z in set ZλΔtIn the previous moment adjacent with t+ Δ t prediction time
The magnitude of traffic flow predicted value at a distance from actual value, dvλIndicate each traffic condition sequence Z in set ZλΔtIn it is pre- with t+ Δ t
The predicted value of the average speed of moment adjacent previous moment is surveyed at a distance from actual value.
Urban highway traffic situation combination forecasting method of the invention, the set ZtIn traffic condition sequence in
Element number α meets: α ∈ [8,10]；The same Sunday chosen from historical data base in the same time traffic condition time series
Number m meets: m ∈ [50,72]；The magnitude of traffic flow week similarity threshold value SqMeet: Sq∈ [0.85,1]；Average speed week is similar
The threshold value S of degreevMeet: Sv∈ [0.80,1]；The element number n+1 that each traffic condition sequence contains in the set Z is full
Foot: n ∈ [8,10]；The maximum value λ of time interval integral multiplemaxMeet: λmax∈[20,30]。
The beneficial effects of the present invention are: urban highway traffic situation combination forecasting method of the invention, using minimax
Method estimates the magnitude of traffic flow of current point and the neighbouring point with Sunday in the same time and all similarities of average speed, if week is similar
Degree in tolerance interval, then utilize same Sunday in the same time data weighting summation method calculate prediction time the magnitude of traffic flow peace
Equal speed.If all similarities are not in tolerance interval, using the prediction algorithm prediction subsequent time based on fractal theory
The magnitude of traffic flow and average speed, using structure function method as theoretical basis, the design of algorithm is considered with pre- for the calculating of fractal dimension
The traffic condition situation of the different time intervals on the basis of the moment is surveyed, and with each traffic sequence prediction moment adjacent to previous
Inverse of the predicted value at moment at a distance from actual value is weighting coefficient, is repaired to predicted value, ensure that prediction data
Accuracy.It uses relative evaluation method on the basis of the traffic jam density and optimum density of current road segment, estimates current road segment
The traffic congestion index of prediction time；The traffic condition state vector finally exported includes: the magnitude of traffic flow of prediction time, prediction
The average speed of moment vehicle, the roading density of prediction time, the traffic congestion index of prediction time.
The prediction algorithm that the present invention designs has the traffic state data that calculation amount elasticity is big, precision of prediction is high, exports complete
The advantages of face, be a kind of effective traffic movement prediction method in short-term, prediction result can for traffic management department into
Row traffic guidance and control service provide foundation.
Detailed description of the invention
Fig. 1 is the flow chart of urban highway traffic situation combination forecasting method of the invention；
Fig. 2 is the traffic condition predictions flow chart based on fractal theory in the present invention.
Specific embodiment
Below with reference to embodiment, the invention will be further described.
As shown in Figure 1, the flow chart of urban highway traffic situation combination forecasting method of the invention is given, it is specific logical
Following methods are crossed to realize:
A) sets current time as t moment, to predict the traffic condition of t+ time Δt, obtains α forward since t moment
The traffic state data at moment forms current t moment traffic condition sequence Zt={ zt-(α-1)Δt, zt-(α-2)Δt,...,zt-2Δt,
zt-Δt, zt, wherein the traffic state data z at each momenti=[qi, vi], t- α Δ t+ Δ t≤i≤t, qiIndicate the friendship at i moment
Through-current capacity, viIndicate the average speed at i moment；
In the step, average speed refers to that the corresponding moment passes through the calculation of the instantaneous speed of all vehicles of a certain section in section
Art average value；Set ZtIn traffic condition sequence in element number α can be in section [8,10] interior value.
B) chooses the traffic condition sequence of m neighbouring same Sundays in the same time from historical data base, and forms set A
={ Z1t,Z2t,...,Zmt, wherein Zjt={ zj(t-(α-1)Δt), zj(t-(α-2)Δt)..., zj(t-2Δt), zj(t-Δt), zj(t), 1≤j
≤ m indicates j-th of traffic condition sequence with Sunday in the same time in set A；zj(i)=[qj(i), vj(i)], t- α Δ t+ Δ t≤i
≤ t, qj(i)Indicate j-th of magnitude of traffic flow with the i moment in Sunday in the same time traffic condition sequence, vj(i)Indicate j-th of same Sunday
In the same time in traffic condition sequence the i moment average speed；
In the step, the number m of traffic condition time series in the same time of the same Sunday chosen in historical data base can be in area
Between value between [50,72]；For example, current time t is morning Tuesday 9:00, then Z in set A1tIt is opened for morning last Tuesday 9:00
Begin the traffic condition sequence recorded forward, Z2tStart the traffic condition sequence recorded forward by two morning of week before last 9:00
Column ..., and so on.
C) calculates current t moment traffic condition sequence and the m traffic conditions with Sunday in the same time with maximin technique
All similarities between sequence traffic condition pass through step c-1) to step c-2) Lai Shixian:
C-1) calculates all similarities of the magnitude of traffic flow, calculates current t moment traffic condition sequence and m according to formula (1)
With all similarity Rq of the magnitude of traffic flow between the traffic condition sequence of Sunday in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
C-2) calculates all similarities of average speed, calculates current t moment traffic condition sequence and m according to formula (2)
With all similarity Rv of average speed between the traffic condition sequence of Sunday in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
D) screens m with the traffic condition sequence of Sunday in the same time using formula (3),
Rqj≥Sq∩Rvj≥Sv (3)
In formula, SqIndicate the threshold value of magnitude of traffic flow week similarity, SvIndicate the threshold value of average speed week similarity；It will
All traffic condition Sequence composition set B={ Z for meeting formula (3)1t,Z2t..., Zht, Zxt={ zx(t-(n-1)Δt),
zx(t-(n-2)Δt)..., zx(t-2Δt),zx(t-Δt),zx(t)Indicate x-th of traffic condition sequence with Sunday in the same time in set B,
1≤x≤h；
In the step, the magnitude of traffic flow week similarity threshold value SqMeet: Sq∈[0.85,1]；Average speed week similarity
Threshold value SvMeet: Sv∈[0.80,1]。
When if there is meeting the sequence of formula (3), i.e. h > 0, then follow the steps e), to calculate moment t+ Δ t's to be predicted
The magnitude of traffic flow and average speed；If satisfactory traffic condition sequence is not present in set B, then follow the steps f)；
E) calculates separately the traffic flow forecasting value q ' of prediction time t+ Δ t according to formula (4) and formula (5)t+ΔtPeace
The predicted value v ' of equal speedt+Δt:
Wherein, Rqx、RvxRespectively indicate x-th of traffic condition sequence in the traffic condition sequence and set B of current t moment
Between the magnitude of traffic flow all similarities, all similarities of average speed；qx(t+Δt)、vx(t+Δt)X-th of traffic in respectively set B
The traffic flow magnitude of situation sequence t+ time Δt, average speed value；Then step g) is executed；
F) calculates the magnitude of traffic flow of prediction time t+ Δ t and the prediction of average speed using the algorithm based on fractal theory
Value；
As shown in Fig. 2, the traffic condition predictions flow chart in the present invention based on fractal theory is given, the benefit in the step
The magnitude of traffic flow of prediction time t+ Δ t and the predicted value of average speed are calculated with the algorithm based on fractal theory, pass through following step
It is rapid to realize:
F-1) fractal prediction algorithm is on the basis of prediction time t+ Δ t, using the traffic data currently acquired, respectively with
Δt、2Δt、...、iΔt、...、λmaxΔ t is that time interval constructs λmaxA traffic condition sequence, and it is combined into collection
It closesThere is the traffic data at n+1 time point in each traffic condition sequence；Set Z institute
Shown in the traffic condition sequence such as expression formula (8) of expression:
Wherein, ZλΔtThe traffic condition of middle each time point includes the magnitude of traffic flow and average speed, λ=1,2 ..., λmax；
In the step, the element number n+1 that each traffic condition sequence contains in set Z meets: n ∈ [8,10]；Time
The maximum value λ of spaced at integer timesmaxMeet: λmax∈ [20,30].
F-2) is the calculation formula of basic construction analysis dimension with structure function method, passes through step f-2-1) and step f-
2-2) find out the fractal dimension D of the magnitude of traffic flowqWith the fractal dimension D of average speedv:
F-2-1) is for sequence Z each in set ZλΔt, λ=1,2 ..., λmax, using formula (9) and formula (10), divide
The arithmetic mean of instantaneous value of the difference side for the traffic condition sequence that each time interval is λ Δ t is not found out:
In formula, Sq(λΔt)、Sv(λ Δ t) respectively indicates traffic condition sequence Z in set ZλΔtThe middle magnitude of traffic flow, average speed
The arithmetic mean of instantaneous value of the difference side of degree；
F-2-2) is by [λ Δ t, Sq(λ Δ t)] draw [In (λ Δ t), In (Sq(λ Δ t))] on double logarithmic chart, utilize
Least square method linear regression algorithm finds out the fractal dimension D of the magnitude of traffic flow in formula (11)qWith constant CqValue；
Equally, by [λ Δ t, Sv(λ Δ t)] draw [In (λ Δ t), In (Sv(λ Δ t))] on double logarithmic chart, using most
Small square law linear regression algorithm finds out the fractal dimension D of average speed in formula (12)vWith constant CvValue:
F-3) utilizes fractal dimension Dq, constant CqWith formula (9), formula (11), each traffic condition sequence in set of computations Z
Arrange ZλΔtIn the previous moment magnitude of traffic flow adjacent with t+ Δ t prediction time predicted value q 't-λΔt+Δt, λ=1,2 ..., λmax,
Seek the distance between its true value and predicted value dqλ, λ=1,2 ..., λmax；
Equally, fractal dimension D is utilizedv, constant CvWith formula (10), formula (12), each traffic condition in set of computations Z
Sequence ZλΔtIn previous moment average speed adjacent with t+ Δ t prediction time predicted value v 't-λΔt+Δt, λ=1,2 ..., λmax,
Seek the distance between its true value and predicted value dvλ, λ=1,2 ..., λmax；
F-4) is by fractal dimension DqWith constant CqValue bring each traffic condition sequence in formula (13) set of computations Z into
ZλΔtThe predicted value q ' of middle t+ time Δt traffic flowλ(t+Δt), λ=1,2 ..., λmax:
Equally, by fractal dimension DvWith constant CvValue bring each traffic condition sequence in formula (14) set of computations Z into
ZλΔtThe predicted value v ' of middle t+ time Δt average speedλ(t+Δt), λ=1,2 ..., λmax:
F-5) calculates separately out the pre- of the current road segment t+ Δ t prediction time magnitude of traffic flow according to formula (15), formula (16)
The predicted value of measured value and average speed:
In formula, dqλIndicate each traffic condition sequence Z in set ZλΔtIn the previous moment adjacent with t+ Δ t prediction time
The magnitude of traffic flow predicted value at a distance from actual value, dvλIndicate each traffic condition sequence Z in set ZλΔtIn it is pre- with t+ Δ t
The predicted value of the average speed of moment adjacent previous moment is surveyed at a distance from actual value.
G) passes through the traffic density that formula (6) calculates t+ Δ t prediction time current road segment first:
k′t+Δt=q 't+Δt/v′t+Δt (6)
If the traffic congestion density of current road segment is Kstop, value range isKsmoothIt indicates
The optimum density of current road segment, value range areThe then traffic congestion index of current road segment
B′t+ΔtCalculation formula it is as follows:
H) exports t+ Δ t prediction time current road segment traffic condition predictions vector R 't+Δt=[q 't+Δt,v′t+Δt,k
′t+Δt,B′t+Δt]。
Claims (3)
1. a kind of urban highway traffic situation combination forecasting method, which is characterized in that realized by following steps:
A) sets current time as t moment, to predict the traffic condition of t+ time Δt, obtains α moment forward since t moment
Traffic state data, form current t moment traffic condition sequence Zt={ zt-(α-1)Δt, zt-(α-2)Δt..., zt-2Δt, zt-Δt,
zt, wherein the traffic state data z at each momenti=[qi, vi], t- α Δ t+ Δ t≤i≤t, qiIndicate the traffic flow at i moment
Amount, viIndicate the average speed at i moment；
B) chooses the traffic condition sequence of m neighbouring same Sundays in the same time from historical data base, and forms set A=
{Z1t, Z2t..., Zmt, wherein Zjt={ zj(t-(α-1)Δt), zj(t-(α-2)Δt),...,zj(t-2Δt),zj(t-Δt),zj(t), 1≤j≤
M indicates j-th of traffic condition sequence with Sunday in the same time in set A；zj(i)=[qj(i), vj(i)], t- α Δ t+ Δ t≤i≤
T, qj(i)Indicate j-th of magnitude of traffic flow with the i moment in Sunday in the same time traffic condition sequence, vj(i)J-th of expression same with Sunday
The average speed at i moment in moment traffic condition sequence；
C) calculates current t moment traffic condition sequence and the m traffic condition sequences with Sunday in the same time with maximin technique
All similarities between traffic condition pass through step c-1) to step c-2) Lai Shixian:
C-1) calculates all similarities of the magnitude of traffic flow, calculates current t moment traffic condition sequence and m same weeks according to formula (1)
All similarity Rq of the magnitude of traffic flow between day traffic condition sequence in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
C-2) calculates all similarities of average speed, calculates current t moment traffic condition sequence and m same weeks according to formula (2)
All similarity Rv of average speed between day traffic condition sequence in the same timej:
In formula, 1≤j≤m, 0≤i≤α -1；
D) screens m with the traffic condition sequence of Sunday in the same time using formula (3),
Rqj≥Sq∩Rvj≥Sv (3)
In formula, SqIndicate the threshold value of magnitude of traffic flow week similarity, SvIndicate the threshold value of average speed week similarity；To own
Meet the traffic condition Sequence composition set B={ Z of formula (3)1t, Z2t..., Zht, Zxt={ zx(t-(n-1)Δt),
zx(t-(n-2)Δt),...,zx(t-2Δt),zx(t-Δt),zx(t)Indicate x-th of traffic condition sequence with Sunday in the same time in set B,
1≤x≤h；
When if there is meeting the sequence of formula (3), i.e. h > 0, then follow the steps e), to calculate the traffic of moment t+ Δ t to be predicted
Flow and average speed；If satisfactory traffic condition sequence is not present in set B, then follow the steps f)；
E) calculates separately the traffic flow forecasting value q ' of prediction time t+ Δ t according to formula (4) and formula (5)t+ΔtPeaceful average rate
The predicted value v ' of degreet+Δt:
Wherein, Rqx、RvxIt respectively indicates in the traffic condition sequence and set B of current t moment between x-th of traffic condition sequence
All similarities, all similarities of average speed of the magnitude of traffic flow；qx(t+Δt)、vx(t+Δt)X-th of traffic condition in respectively set B
Actual value, the actual value of average speed of the magnitude of traffic flow of sequence t+ time Δt；Then step g) is executed；
F) calculates the magnitude of traffic flow of prediction time t+ Δ t and the predicted value of average speed using the algorithm based on fractal theory；
G) then., the traffic density of t+ Δ t prediction time current road segment is calculated by formula (6):
k′t+Δt=qt+Δt/v′t+Δt (6)
If the traffic congestion density of current road segment is Kstop, value range isKsmoothIndicate current
The optimum density in section, value range areT+ Δ t prediction is then calculated according to relative evaluation method
The traffic congestion index B ' of moment current road segmentt+Δt, formula is as follows:
H) exports t+ Δ t prediction time current road segment traffic condition predictions vector R 't+Δt=[q 't+Δt,v′t+Δt,k′t+Δt,
B′t+Δt]。
2. urban highway traffic situation combination forecasting method according to claim 1, which is characterized in that described in step f)
Algorithm based on fractal theory calculate the magnitude of traffic flow of prediction time t+ Δ t and the predicted value of average speed, pass through following step
It is rapid to realize:
F-1) fractal prediction algorithm is on the basis of prediction time t+ Δ t, using the traffic data currently acquired, respectively with Δ t, 2
Δt、...、iΔt、...、λmaxΔ t is that time interval constructs λmaxA traffic condition sequence, and it is combined into setThere is the traffic data at n+1 time point in each traffic condition sequence；Set Z institute table
Shown in the traffic condition sequence such as expression formula (8) shown:
Wherein, ZλΔtThe traffic condition of middle each time point includes the magnitude of traffic flow and average speed, λ=1,2 ..., λmax；
F-2) is the calculation formula of basic construction analysis dimension with structure function method, passes through step f-2-1) and step f-2-2)
Find out the fractal dimension D of the magnitude of traffic flowqWith the fractal dimension D of average speedv:
F-2-1) is for sequence Z each in set ZλΔt, λ=1,2 ..., λmax, using formula (9) and formula (10), ask respectively
Each time interval is the arithmetic mean of instantaneous value of the difference side of the traffic condition sequence of λ Δ t out:
In formula, Sq(λΔt)、Sv(λ Δ t) respectively indicates traffic condition sequence Z in set ZλΔtThe middle magnitude of traffic flow, average speed
The arithmetic mean of instantaneous value of difference side；
F-2-2) is by [λ Δ t, Sq(λ Δ t)] draw [In (λ Δ t), In (Sq(λ Δ t))] on double logarithmic chart, utilize minimum
Square law linear regression algorithm finds out the fractal dimension D of the magnitude of traffic flow in formula (11)qWith constant CqValue；
Equally, by [λ Δ t, Sv(λ Δ t)] draw [In (λ Δ t), In (Sv(λ Δ t))] on double logarithmic chart, utilize minimum two
Multiplication linear regression algorithm finds out the fractal dimension D of average speed in formula (12)vWith constant CvValue:
F-3) utilizes fractal dimension Dq, constant CqWith formula (9), formula (11), each traffic condition sequence in set of computations Z
ZλΔtIn the previous moment magnitude of traffic flow adjacent with t+ Δ t prediction time predicted value q 't-λΔt+Δt, λ=1,2 ..., λmax, ask
The distance between its true value and predicted value dqλ, λ=1,2 ..., λmax；
Equally, fractal dimension D is utilizedv, constant CvWith formula (10), formula (12), each traffic condition sequence in set of computations Z
ZλΔtIn the previous moment average speed adjacent with t+ Δ t prediction time predicted value v 't-λΔt+Δt, λ=1,2 ..., λmax, ask
The distance between its true value and predicted value dvλ, λ=1,2 ..., λmax；
F-4) is by fractal dimension DqWith constant CqValue bring each traffic condition sequence Z in formula (13) set of computations Z intoλΔtMiddle t
The predicted value q ' of+time Δt traffic flowλ(t+Δt), λ=1,2 ..., λmax:
Equally, by fractal dimension DvWith constant CvValue bring each traffic condition sequence Z in formula (14) set of computations Z intoλΔtMiddle t
The predicted value v ' of+time Δt average speedλ(t+Δt), λ=1,2 ..., λmax:
F-5) calculates separately out the predicted value of the current road segment t+ Δ t prediction time magnitude of traffic flow according to formula (15), formula (16)
With the predicted value of average speed:
In formula, dqλIndicate each traffic condition sequence Z in set ZλΔtIn the previous moment adjacent with t+ Δ t prediction time friendship
The predicted value of through-current capacity is at a distance from actual value, dvλIndicate each traffic condition sequence Z in set ZλΔtIn with t+ Δ t predict when
The predicted value of the average speed of adjacent previous moment is carved at a distance from actual value.
3. urban highway traffic situation combination forecasting method according to claim 1 or 2, it is characterised in that: the collection
Close ZtIn traffic condition sequence middle element number α meet: α ∈ [8,10]；The same Sunday chosen from historical data base is same
The number m of moment traffic condition time series meets: m ∈ [50,72]；The magnitude of traffic flow week similarity threshold value SqMeet: Sq∈
[0.85,1]；Average speed week similarity threshold value SvMeet: Sv∈[0.80,1]；Each traffic condition sequence in the set Z
It arranges the element number n+1 contained to meet: n ∈ [8,10]；The maximum value λ of time interval integral multiplemaxMeet: λmax∈[20,30]。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611114072.4A CN106548625B (en) | 2016-12-07 | 2016-12-07 | A kind of urban highway traffic situation combination forecasting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611114072.4A CN106548625B (en) | 2016-12-07 | 2016-12-07 | A kind of urban highway traffic situation combination forecasting method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106548625A CN106548625A (en) | 2017-03-29 |
| CN106548625B true CN106548625B (en) | 2019-02-26 |
Family
ID=58396265
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611114072.4A Active CN106548625B (en) | 2016-12-07 | 2016-12-07 | A kind of urban highway traffic situation combination forecasting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106548625B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967803A (en) * | 2017-11-17 | 2018-04-27 | 东南大学 | Traffic congestion Forecasting Methodology based on multi-source data and variable-weight combined forecasting model |
| CN108305459B (en) * | 2018-02-05 | 2021-02-02 | 北京交通大学 | Road network operation index prediction method based on data driving |
| CN110288824B (en) * | 2019-05-20 | 2020-12-01 | 浙江工业大学 | Early-late peak congestion condition and propagation mechanism analysis method based on Granger cautuality road network |
| CN112907969B (en) * | 2021-02-02 | 2022-04-22 | 中国科学院计算技术研究所 | Method and system for predicting road traffic flow |
| CN114463972B (en) * | 2022-01-26 | 2024-02-27 | 成都和乐信软件有限公司 | Road section interval traffic analysis prediction method based on ETC portal communication data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003242589A (en) * | 2002-02-18 | 2003-08-29 | Hiroshi Tanaka | Road running time predicting system |
| CN101438334A (en) * | 2006-03-03 | 2009-05-20 | 因瑞克斯有限公司 | Dynamic time series prediction of future traffic conditions |
| CN101694743A (en) * | 2009-08-25 | 2010-04-14 | 北京世纪高通科技有限公司 | Method and device for predicting road conditions |
| CN105139656A (en) * | 2015-09-28 | 2015-12-09 | 百度在线网络技术（北京）有限公司 | Road state prediction method and device |
| CN105389987A (en) * | 2015-12-03 | 2016-03-09 | 青岛海信网络科技股份有限公司 | Road traffic condition prediction method and device |
-
2016
- 2016-12-07 CN CN201611114072.4A patent/CN106548625B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003242589A (en) * | 2002-02-18 | 2003-08-29 | Hiroshi Tanaka | Road running time predicting system |
| CN101438334A (en) * | 2006-03-03 | 2009-05-20 | 因瑞克斯有限公司 | Dynamic time series prediction of future traffic conditions |
| CN101694743A (en) * | 2009-08-25 | 2010-04-14 | 北京世纪高通科技有限公司 | Method and device for predicting road conditions |
| CN105139656A (en) * | 2015-09-28 | 2015-12-09 | 百度在线网络技术（北京）有限公司 | Road state prediction method and device |
| CN105389987A (en) * | 2015-12-03 | 2016-03-09 | 青岛海信网络科技股份有限公司 | Road traffic condition prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106548625A (en) | 2017-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106548625B (en) | A kind of urban highway traffic situation combination forecasting method | |
| CN109658695B (en) | Multi-factor short-term traffic flow prediction method | |
| CN109035761B (en) | Travel time estimation method based on auxiliary supervised learning | |
| CN111080029B (en) | Urban traffic road speed prediction method and system based on multi-path segment space-time correlation | |
| CN105741549B (en) | A kind of traffic flow forecasting method theoretical based on space copula | |
| CN104183119B (en) | Based on the anti-arithmetic for real-time traffic flow distribution forecasting method pushed away of section OD | |
| Kalayathankal et al. | A fuzzy soft flood alarm model | |
| CN103280098A (en) | Traffic congestion index calculation method | |
| CN102629418A (en) | Fuzzy kalman filtering-based traffic flow parameter prediction method | |
| CN103295414A (en) | Bus arrival time forecasting method based on mass historical GPS (global position system) trajectory data | |
| CN103177570A (en) | Method for predicting traffic jam indexes for rush hours in morning and evening | |
| CN105512741A (en) | Bus passenger traffic combined prediction method | |
| Kumar et al. | Sensitivity analysis of the GIUH based Clark model for a catchment | |
| CN102033892A (en) | Method and system for generating historical standard data of traffic flow | |
| CN110969312A (en) | Short-term runoff prediction coupling method based on variational modal decomposition and extreme learning machine | |
| CN104182807A (en) | Reservoir dispatching risk evaluation method by considering runoff forecast uncertainty | |
| US20100036793A1 (en) | Method and System for Geospatial Forecasting of Events Incorporating Data Error and Uncertainty | |
| CN113192325B (en) | Traffic incident correlation analysis method and system | |
| Qiao et al. | Short-term traffic flow forecast based on parallel long short-term memory neural network | |
| Lababede et al. | Mosques smart domes system using machine learning algorithms | |
| Drobne et al. | Evaluating functional regions | |
| Yu et al. | Application of Grey Model toward Runoff Forecasting 1 | |
| CN102708283A (en) | Modeling and simulating method for travelling times of residents based on travelling chain | |
| Sharma et al. | Adoptive neuro-fuzzy inference system for traffic noise prediction | |
| Freire et al. | Predicting building's corners hygrothermal behavior by using a Fuzzy inference system combined with clustering and Kalman filter |
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 |