CN104183119A - Real-time traffic flow distribution prediction system based on road section OD backstepping - Google Patents

Abstract

The invention discloses a real-time traffic flow distribution prediction system based on road section OD backstepping. The system is used for predicting real-time travel and absorbing traffic volume of a road section and carrying out real-time traffic flow distribution. The system has the advantages that division of current scattered traffic zones, prediction of traffic volume of the traffic zones, real-time prediction of OD traffic volume and flow distribution of the road section are systematically integrated, a traffic zone division method based on a road network and a prediction system of real-time traffic flow distribution are constructed, knowledge and technologies about travel origin-destination and traffic volume distribution are organically combined, and organic combination is achieved in the perspective of software.

Description

Based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD

Technical field

The present invention relates to intelligent transport technology field, especially relate to a kind of generation of road grid traffic and the real-time estimate system of traffic attraction.

Background technology

Be accompanied by widespread use and real time information system growing of intelligent transportation system in wisdom urban construction and management, that traffic administration forms is gradually intelligent, the feature of data, real-time.Vehicle supervision department and research institution obtain a large amount of real time traffic datas, and are applied to the field of traffic control such as real-time road.A large amount of traffic information data builds historical data base, for traffic administration person and researchist provide important Data support.In order to carry out effective traffic administration and for traveler provides Traffic Information accurately, need to carry out traffic generation accurately and traffic attraction prediction based on Real-time Traffic Information.Traffic trip and the traffic attraction prediction based on real time traffic data, carried out in short-term contribute to traffic forecast of distribution, and this is the pith that carries out traffic administration and planning.

The prediction of traffic trip and traffic attraction at present, how by traffic study, is calculated travel amount and the traffic attraction of traffic zone according to factors such as the personnel component of each traffic zone, motor vehicle retention.First, the traffic study of large sample amount needs a large amount of time and fund input; Secondly, the scope of dividing traffic zone with according to different, can cause the inaccurate of prediction, and the prediction of traffic zone cannot carry out Proper Match with traffic network, predicts the outcome and disagrees with actual conditions; The 3rd, the investigation based on static and prediction are difficult to reflect real-time traffic stream mode and carry out short-time traffic flow forecast more accurately.

Summary of the invention

In order to overcome above-mentioned the deficiencies in the prior art, the invention provides a kind of based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, to divide the traffic trip of the real-time OD distribution of carrying out and the prognoses system of traffic attraction based on traffic zone, for improving the accuracy of trip forecast of distribution.

To achieve these goals, the technical solution adopted in the present invention is:

The present invention has carried out the integration of system to " the traffic zone division based on section, the volume of traffic short-term prediction based on road section traffic volume community, OD Prediction of Traffic Flow in Real-time, link flow distribution ", the traffic zone division methods of structure based on section and the prognoses system of arithmetic for real-time traffic flow distribution, realization to trip OD between the prediction of real-time traffic amount, and the method for further distributing according to shortest path, by OD between the real-time estimate volume of traffic be assigned in each section of road network, obtain real-time estimate traffic flow distribution situation.

Based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, comprise the following steps:

1.1, every section in transportation network is considered as to traffic zone, the transport information (Floating Car gps data) gathering is carried out to OD extraction and section coupling;

1.2, according to prediction constantly, choose this prediction in a few days in the historical Floating Car OD traffic data storehouse OD data of front and back each half an hours constantly, and averaged calculating, using result of calculation as basic OD matrix, in substitution OD matrix;

1.3, according to prediction constantly, choose prediction in this day the first two OD traffic data constantly constantly, the Time Series Method that it is once moved, obtain the prediction traffic volume forecast value of each traffic zone (section) O, D constantly, and in the aggregation of forecasts value substitution OD matrix using it as each O, D;

1.4, according to balanced growth Y-factor method Y to trip OD matrix counter pushing away, obtain OD under predicted time between the volume of traffic;

1.5, according to the algorithm of shortest path, the prediction OD volume of traffic is carried out to traffic flow distribution, obtain real-time road traffic delay prediction.

Compared with prior art, the invention has the beneficial effects as follows: section is considered as to traffic zone, can solve the trip of current traffic zone, attract the volume of traffic to mate bad problem with road network, simultaneously, can reduce the cost of traffic zone investigation, the volume of traffic that is obtained trip between traffic zone, attracted by real time traffic data, solves real-time traffic distribution forecast problem.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the open method of the present invention.

Fig. 2 is for adopting traffic zone of the present invention division methods to interrupt with OD and extract schematic diagram the road network of Panyu District of Guangzhou City.

Fig. 3 is 0-1 apportion design process flow diagram.

Fig. 4 is increment apportion design process flow diagram schematic diagram.

Fig. 5 is straight average method process flow diagram.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

Techniqueflow of the present invention as shown in Figure 1, comprises the division of the traffic zone based on section, the structure of database, in real time OD traffic volume forecast, arithmetic for real-time traffic flow distribute four steps.

1.1, the traffic zone based on section is divided:

The unit of traditional traffic four-phase model is traffic zone, and the present invention replaces section traffic zone as the unit of four-phase model, and interrupt the road network in region into the section of junior unit more according to the principle that has crossing just to interrupt in the section to whole region.Traffic zone is changed into based on section, and counter the pushing away with traffic of OD of serving below with this distributed.Specific algorithm flow process is as follows:

Choose the basic road network of research, according to crossing, interrupt and the section road axis angle of turning is less than the principle that interrupt the corner of 150 °, Regional Road Network is broken into the more section of small scale, section after interrupting will replace traffic zone, and it is numbered and adds up it as four-phase model the minimum unit of OD travel amount.Traditional OD matrix is all to take traffic analysis community as analytic unit, does when traffic flow distributes and the traffic flow of each traffic zone need to be matched on corresponding road network carrying out routing like this, and this process can be introduced error.Directly adopt in the present embodiment section as the analytic unit of OD matrix, can avoid drawing-in system error in this course.

1.2, the structure of database:

The structure of database is based on gps data being carried out to real-time OD extraction, section coupling and OD tabulate statistics three parts based on section.

1. O, D extract in real time: first gps data was sorted according to car number and date, time, then according to the variation of passenger carrying status field in gps data, judge O point or D point, for example: in passenger carrying status, 1 is load-carrying, 0 is unloaded,, when the passenger carrying status of same car is when 1 becomes 0, be D; When the passenger carrying status of same car is when 0 becomes 1, be O; When finding O point, down find corresponding D point; When finding D point, recall and find corresponding O point, with this, mate OD point right.

2. mate in section: the O extracting, D point is converted to shapefile dot format, adopts the method coupling section in nearest section, search out the nearest section of GPS point, and assignment is numbered to GPS point in section.

3. the OD tabulate statistics based on section: to matching the GPS OD point in section, the segment number of satisfying the need is carried out tabulate statistics, and passenger carrying status is carried out to counting statistics, obtains O, D matrix based on section.

1.3, real-time OD traffic volume forecast:

OD traffic volume forecast is divided into three steps in real time: the first, according to historical data, determine basic OD matrix; The second, according to real time data, determine OD Matrix prediction value; The 3rd, counter the estimating of OD calculated the prediction OD volume of traffic constantly.

1. according to historical data, determine basic OD matrix

Historical data based on predicted time point and the proportionate relationship of historical Road network traffic, predict.As to predict each OD in 8:00 road network in morning between the volume of traffic, by extracting each OD of road network to the historical volume of traffic at 8:00, and using its mean value as historical data, bring in OD matrix.

2. according to real time data, determine OD Matrix prediction value

Forecasting Methodology based on lineary system theory mainly comprises: linear statistical regression, historical trend method, Time Series Forecasting Methods, Kalman prediction method, Smoothing Forecasting and adaptive weighting forecast model method etc.The present embodiment be take Time Series Forecasting Methods---and Single moving average method is carried out short-time traffic flow forecast as example.

If x ₁, x ₂, x ₃..., x _nfor specific O point, take t as the time interval, the generation volume of traffic in every period, or specific D point, take t as the time interval, the attraction volume of traffic in every period.

Therefore, it is a time series, and sample size is n.

The formula of the method for moving average is as follows:

$M_T^{\left(1\right)}=\frac{x_T+x_{T-1}+...+x_{T-N+1}}{N},T\≥N$

Wherein, N is the data amount check that each moving average comprises; it is the Single moving average value of T phase.

The effect of moving average is smoothing data, eliminates some random disturbance, and secular trend are revealed, thereby can be used for trend analysis and prediction.

Generally, if time series does not have the obvious cycle to change and trend changes, the Single moving average value of available T phase is as the predicted value of T+1 phase, that is:

${\hat{x}}_{T+1}=M_T^{\left(1\right)}=\frac{x_T+x_{T-1}+...+x_{T-N+1}}{N}$

Abbreviation obtains recursion formula:

${\hat{x}}_{T+1}=M_T^{\left(1\right)}=M_{T-1}^{\left(1\right)}+\frac{x_T-x_{T-N}}{N}={\hat{x}}_T+\frac{x_T-x_{T-N}}{N}$

When N is larger, with recursion formula, can greatly reduce calculated amount.

Therefore, according to OD between historical data, the volume of traffic in next time interval is predicted.

Finally can add up and obtain the generation traffic volume forecast value of each O in OD matrix and the attraction traffic volume forecast value of each D, by in historical data base, traffic flow data around prediction constantly within the scope of certain hour is assigned in OD matrix according to origin and destination, as the basis matrix of OD matrix.Meanwhile, according to the traffic flow data in several moment before Time Series Method and prediction constantly, extrapolate prediction origin and destination traffic volume forecast value constantly, as the predicted value in OD matrix.

3. OD is counter estimates the real-time OD volume of traffic

This research adopts the distribution of growth factor methed prognosis traffic volume.

The algorithm steps of growth factor methed is as follows:

The first step, makes calculation times m=0

Second step, in given present situation OD table t ^mand the U in OD table in the future _i, V _j;

Wherein while representing to calculate for the m time, traffic zone i is to the volume of traffic between the j of traffic zone;

while representing to calculate for the m time, there is the volume of traffic in i traffic zone;

while representing to calculate for the m time, the attraction volume of traffic of j traffic zone;

U _irepresent the generation volume of traffic in OD table in future;

V _jrepresent the attraction volume of traffic in OD table in future;

The 3rd step, obtains the generation and the rate of growth that attracts the volume of traffic of each community

$F_{\mathrm{Oi}}^m=U_i/O_i^m$

$F_{\mathrm{Dj}}^m=V_j/D_j^m$

The 4th step, asks the approximate value of the m+1 time distribution volume of traffic

$q_{\mathrm{ij}}^{m+1}=q_{\mathrm{ij}}^m\×f(F_{\mathrm{Oi}}^m,F_{\mathrm{Dj}}^m)$

The 5th step, convergence is differentiated, and discrimination formula is:

$O_i^{m+1}={\mathrm{\Σ}}_j{q}_{\mathrm{ij}}^{m+1}$

$D_j^{m+1}={\mathrm{\Σ}}_j{q}_{\mathrm{ij}}^{m+1}$

$1-\mathrm{\&epsiv;}<F_{\mathrm{Oi}}^{m+1}=U_i/O_i^m<1+\mathrm{\&epsiv;}$

$1-\mathrm{\&epsiv;}<F_{\mathrm{Dj}}^{m+1}=V_j/D_i^{m+1}<1+\mathrm{\&epsiv;}$

In formula, U _irepresent the generation volume of traffic in OD table in future;

V _jrepresent the attraction volume of traffic in OD table in future;

represent that growth factor occurs calculating for the m time of i community;

represent that calculating for the m time of j community attracts growth factor;

the iterative manner of growth factor, relevant with every kind of method;

ε represents any given error constant, generally gets 3%.

According to function kind different, growth factor methed can be divided into normal growth factor methed (Unique Growth Factor Method), balanced growth Y-factor method Y (Average Growth Factor Method), Detroit method (Detroit Method), Follett, Mary Parker method (Fratar Method) and Buddhist Nice method (Furness Method).

The present invention adopts balanced growth Y-factor method Y, and the method formula is concisely in calculating, and shortcoming is that convergence is slow, and iterations is many, and computational accuracy is low.Balanced growth coefficient is expressed as follows:

$F_{\mathrm{Oi}}^0=U_i/O_i^0$

$F_{\mathrm{Dj}}^0=V_j/D_j^0$

First approximation: $q_{\mathrm{ij}}^1=q_{\mathrm{ij}}^0\&times;(F_{\mathrm{Oi}}^0+F_{\mathrm{Dj}}^0)\&divide;2$

(4) arithmetic for real-time traffic flow distributes:

Conventionally traffic flow distribution method being divided in the world to balance distributes and the large class of non-equilibrium distribution two.Equilibrium state for meeting the definition of Wardrop principle completely, is called balance apportion design; Apportion model for adopting heuristic or other approximation methods, is called non-equilibrium distribution method.

After obtaining real-time OD matrix by OD estimation model, adopt respectively again three kinds of non-equilibrium allocation algorithm exploitation dynamic traffic assignment models such as 0-1 apportion design, increment apportion design, straight average method, the new real-time OD volume of traffic is assigned to road network, obtains the arithmetic for real-time traffic flow distribution situation of road network.

1. 0-1 apportion design

0-1 apportion design also claims entirely have completely without apportion design, is the most basic the simplest volume of traffic apportion design, in the traffic programme of Chicago,U.S city, obtains first application.This algorithm has 2 features: the one, think that the time out of shape in section is only relevant with the length in section, and be a definite value, irrelevant with the size of the volume of traffic on section, with the degree of crowding of road, do not change to some extent; The 2nd, think that all drivers of OD of same group select identical route.The main processes of calculation of this distribution method is to find shortest path.

0-1 apportion design is simple and efficiency is high.Be applicable in the remote districts of road rareness, and the volume of traffic of general city road network is not suitable for adopting this distribution method in distributing.Calculation procedure can be summarized as follows:

The first step, in computational grid each OD between shortest path;

Second step, is all assigned to the volume of traffic between OD on corresponding shortest path.

Because 0-1 apportion design can not reflect crowded effect, be mainly used for some non-crowded road network, this apportion design is not for there is no the situation of the network of traffic capacity restriction.Therefore, suggestion usable range is: the unrestricted area of road passage capability can adopt between intercity; The traffic flow of general urban road network distributes should not adopt the method.In practice due to its simple and practical characteristic, generally as the basis of other various distribution techniques, Reusability in the methods such as increment apportion design and balance apportion design.

2. increment apportion design

Increment apportion design is a kind of approximate balance apportion design.It is that the OD volume of traffic is divided equally into some equal portions, cyclically the OD volume of traffic of each equal portions is assigned in network.The OD volume of traffic of cycle assignment one equal portions is to corresponding shortest path each time, and the travel time in each section is once just recalculated and upgrades in every distribution, then by the travel time after upgrading, recalculates the shortest path between each OD of network.In next circulation, by the shortest path after upgrading, distribute the OD volume of traffic of next equal portions.Its calculation procedure is as follows:

The first step, initialization.Every group of OD volume of traffic is divided equally into N equal portions, order n=1, $x_a^n=0,\&ForAll;a.$

Second step, the travel time in renewal section. calculate shortest path between OD.

The 3rd step, increment distributes.Press second step and calculate gained by 0-1 apportion design by the OD volume of traffic of 1/N be assigned in network and go.Obtain like this one group of additional volume of traffic

The 4th step, the volume of traffic is cumulative.Order $x_a^n=x_a^{n-1}+w_a^n,\&ForAll;a.$

The 5th step, judges.If n=N, stops calculating.Current road section traffic volume amount is final solution; If n < is N, make n=n+1, return to second step.

Increment apportion design obtains complexity and solves accuracy all between 0-1 apportion design and balance apportion design.Consistent with the result of 0-1 apportion design when N=1; When N → ∞, its solution is consistent with the solution of balance apportion design.The method is simple, in the business software of many maturations, provides use, and degree of accuracy can regulate according to the size of N, in the actual road net volume of traffic distributes, is often applied.But this distribution method remains a kind of approximation method, sometimes the too much volume of traffic is assigned on the section that some capacity is very little.

3. straight average method

Straight average method is a kind of round-robin technique between increment apportion design and balance apportion design.It is that continuous adjustment has been assigned to the volume of traffic on each section and has moved closer to or arrive balance distribution.In every step circulation, according to the volume of traffic being assigned on each section, carry out the additional volume of traffic that a 0-1 distribution obtains one group of each section.Then with the additional volume of traffic obtaining in the distribute traffic amount in each section in last circulation and this circulation, be weighted average and obtain the section distribute traffic amount in this circulation.When the distribute traffic amount in adjacent two circulations equals definite value, can stop calculating, this equivalence is a very little value, approximates 1, can think that the distribute traffic amount in adjacent two circulations is very approaching.The distribute traffic amount obtaining in last circulation is the final volume of traffic.Its main algorithm step is as follows:

The first step, initialization.According to the free travel time in each section, carry out a 0-1 and distribute, obtain the distribute traffic amount in each section

Second step, according to the distribute traffic amount in current each section calculate each Link Travel Time, and calculate shortest path between OD.

The 3rd step, the Link Travel Time calculating according to second step and the OD volume of traffic carry out a 0-1 and distribute, and obtain the additional volume of traffic F in each section _a.

The 4th step, calculates the current volume of traffic in each section by the method for method of weighted mean as follows:

$x_a^n=(1-\mathrm{\&theta;})x_a^{n-1}+{\mathrm{\&theta;F}}_a,0\&le;\mathrm{\&theta;}\&le;1$

The 5th step, if loop termination, distributes otherwise re-start 0-1, carries out circulation.

In the 4th step and the 5th step, weight coefficient θ and parameter K need be defined by reckoner oneself.Arranging of parameter can determine the number of times of circulation and the operation efficiency of circulation.

Straight average method is to be both simply suitable for, again close to a kind of distribution method of balance apportion design; In every step circulation, the value of weight coefficient θ, in strict accordance with mathematical programming model value, can be balanced the solution of distribution.

By above method, the service level of Dynamic Acquisition road network in real time, for dredging decision support be provided traffic flow under special traffic events (as waterlogging, traffic hazard etc.) situation.

Above-mentioned based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, combine the aspects such as traffic zone division, short-term traffic flow prediction, OD Prediction of Traffic Flow in Real-time, link flow distribution, realized to trip OD between the prediction of real-time traffic amount, and the method for further distributing according to shortest path, by OD between the real-time estimate volume of traffic be assigned in each section of road network, obtain real-time estimate traffic flow distribution situation.

Compare with method in the past, the invention has the beneficial effects as follows: one, the present invention is considered as traffic zone by section, can solve the trip of current traffic zone, attract the volume of traffic to mate bad problem with road network; Two, the present invention can reduce the cost of traffic zone investigation, and the volume of traffic that is obtained trip between traffic zone, attracted by real time traffic data solves real-time traffic distribution forecast problem.

Based on above feature, what the present invention announced can play a great role at aspects such as traffic flow real-time estimate, road conditions assessments based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, and the development that also can be digital city and wisdom city increases power-assisted.

The above embodiment has only expressed the possible embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.

Claims (9)

1. based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, comprise the following steps:

S1. every section in transportation network is considered as to traffic zone, i.e. O and D in OD matrix;

S2. according to the GPS data from taxi passenger carrying status field obtained, the OD couple of judgement trip, the OD point of going on a journey extraction, and the OD point of extraction is carried out to section with traffic zone mate, so gather obtain each traffic zone OD between the volume of traffic;

S3. the traffic zone OD obtaining according to step S2 between the gathering of the volume of traffic, build historical OD traffic data storehouse;

S4. according to prediction constantly, in historical OD traffic data storehouse, choose the OD data in certain limit around this prediction constantly in a few days, and averaged calculating, using result of calculation as basic OD matrix, in substitution OD matrix;

S5. according to prediction constantly, choose the OD traffic data in this day constantly front several moment of prediction, according to Time Series Method, it is moved, obtain the prediction traffic volume forecast value of each traffic zone O, D constantly, and in the aggregation of forecasts value substitution OD matrix using it as each O, D;

S6. the basic OD matrix calculating according to step S4 and S5 and O, D aggregation of forecasts value, adopt balanced growth Y-factor method Y to push away OD matrix is counter, obtain prediction each OD constantly between the volume of traffic distribute;

S7. the algorithm that adopts shortest path to distribute, by each OD between the real-time estimate volume of traffic be assigned on each section of road network, obtain the real-time traffic amount distribution forecast in each section in this prediction moment road network.

2. according to claim 1 based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, step S1 is less than 150 ° by each section in road network according to crossing and section angle to interrupt, and each section after interrupting is as traffic zone.

3. according to claim 2 based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, the right method of OD of step S2 judgement trip is: gps data was sorted according to car number and date, time, then according to the variation of passenger carrying status field in gps data, judge O point or D point, when the passenger carrying status of same car is when load-carrying becomes zero load, be D; When the passenger carrying status of same car is when zero load becomes load-carrying, be O; When finding O point, down find corresponding D point; When finding D point, recall and find corresponding O point, with this, mate OD point right.

4. according to claim 3 based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, in step S2, the OD point of extraction being carried out to section with traffic zone mates, gather obtain each traffic zone OD between the detailed process of the volume of traffic be: by the O extracting, D point is converted to shapefile dot format, adopt the method coupling section in nearest section, search out the nearest section of GPS point, and assignment is numbered to GPS point in section: to matching the GPS OD point in section, the segment number of satisfying the need is carried out tabulate statistics, passenger carrying status is carried out to counting statistics, obtain the O based on section, D matrix.

5. according to claim 4 based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, in step S5 according to prediction constantly, choose prediction in this day the first two OD traffic data constantly constantly, the Time Series Method that it is once moved, obtains the prediction traffic volume forecast value of each traffic zone O, D constantly.

6. according to claim 5ly based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, the Time Series Method that it is once moved, specifically adopts Single moving average method, is specially:

If x ₁, x ₂, x ₃..., x _nfor specific O point, take t as the time interval, the generation volume of traffic in every period, or x ₁, x ₂, x ₃..., x _nfor specific D point, take t as the time interval, the attraction volume of traffic in every period;

Therefore, it is a time series, and sample size is n;

The formula of the method for moving average is as follows:

$M_T^{\left(1\right)}=\frac{x_T+x_{T-1}+...+x_{T-N+1}}{N},T\&GreaterEqual;N$

Wherein, N is the data amount check that each moving average comprises; it is the Single moving average value of T phase;

When time series has obvious cycle to change and trend variation, the predicted value by the Single moving average value of T phase as the T+1 phase that is:

${\hat{x}}_{T+1}=M_T^{\left(1\right)}=\frac{x_T+x_{T-1}+...+x_{T-N+1}}{N}$

Abbreviation obtains recursion formula:

${\hat{x}}_{T+1}=M_T^{\left(1\right)}=M_{T-1}^{\left(1\right)}+\frac{x_T-x_{T-N}}{N}={\hat{x}}_T+\frac{x_T-x_{T-N}}{N}.$

7. according to claim 6ly based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, in step S6, adopt balanced growth Y-factor method Y to the anti-detailed process pushing away of OD matrix to be:

The first step, makes calculation times m=0;

Second step, in given present situation OD table and the U in OD table in the future _i, V _j;

Wherein while representing to calculate for the m time, traffic zone i is to the volume of traffic between the j of traffic zone;

while representing to calculate for the m time, there is the volume of traffic in i traffic zone;

while representing to calculate for the m time, the attraction volume of traffic of j traffic zone;

U _irepresent the generation volume of traffic in OD table in future;

V _jrepresent the attraction volume of traffic in OD table in future;

The 3rd step, obtains the rate of growth of the generation volume of traffic of each community with the rate of growth that attracts the volume of traffic

$F_{\mathrm{Oi}}^m=U_i/O_i^m$

$F_{\mathrm{Dj}}^m=V_j/D_j^m$

The 4th step, asks the approximate value of the m+1 time distribution volume of traffic

$q_{\mathrm{ij}}^{m+1}=q_{\mathrm{ij}}^m\&times;f(F_{\mathrm{Oi}}^m,F_{\mathrm{Dj}}^m)$

The 5th step, convergence is differentiated, and discrimination formula is:

$O_i^{m+1}={\mathrm{\&Sigma;}}_j{q}_{\mathrm{ij}}^{m+1}$

$D_j^{m+1}={\mathrm{\&Sigma;}}_j{q}_{\mathrm{ij}}^{m+1}$

$1-\mathrm{\&epsiv;}<F_{\mathrm{Oi}}^{m+1}=U_i/O_i^{m+1}<1+\mathrm{\&epsiv;}$

$1-\mathrm{\&epsiv;}<F_{\mathrm{Dj}}^{m+1}=V_j/D_i^{m+1}<1+\mathrm{\&epsiv;}$

In formula, U _irepresent the generation volume of traffic in OD table in future;

V _jrepresent the attraction volume of traffic in OD table in future;

represent that growth factor occurs calculating for the m time of i community;

represent that calculating for the m time of j community attracts growth factor;

the iterative manner of growth factor, relevant with every kind of method;

ε represents any given error constant;

According to function kind different, growth factor methed is divided into normal growth factor methed, balanced growth Y-factor method Y, Detroit method, Follett, Mary Parker method or Buddhist Nice method.

8. according to claim 7ly based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, described growth factor methed adopts balanced growth Y-factor method Y, its balanced growth coefficient be expressed as follows:

$F_{\mathrm{Oi}}^0=U_i/O_i^0$

$F_{\mathrm{Dj}}^0=V_j/D_j^0$

First approximation: $q_{\mathrm{ij}}^1=q_{\mathrm{ij}}^0\&times;(F_{\mathrm{Oi}}^0+F_{\mathrm{Dj}}^0)\&divide;2.$

9. according to claim 8 based on the anti-arithmetic for real-time traffic flow forecast of distribution system pushing away of section OD, it is characterized in that, step S7 adopts 0-1 apportion design, increment apportion design, three kinds of non-equilibrium allocation algorithm exploitation dynamic traffic assignment models of straight average method, the new real-time OD volume of traffic is assigned to road network, and its detailed process is:

0-1 apportion design wherein, be in computational grid each OD between shortest path, the volume of traffic between OD is all assigned on corresponding shortest path;

Increment apportion design, is that the OD volume of traffic is divided equally into some equal portions, cyclically the OD volume of traffic of each equal portions is assigned in network; The OD volume of traffic of each cycle assignment one equal portions is to corresponding shortest path, and the travel time in each section is once just recalculated and upgrades in every distribution, then by the travel time after upgrading, recalculates the shortest path between each OD of network; In next circulation, by the shortest path after upgrading, distribute the OD volume of traffic of next equal portions;

Straight average method, constantly adjustment has been assigned to the volume of traffic on each section and has moved closer to or arrived balance distribution, in every step circulation, according to the volume of traffic being assigned on each section, carries out the additional volume of traffic that a 0-1 distribution obtains one group of each section; Then with the additional volume of traffic obtaining in the distribute traffic amount in each section in last circulation and this circulation, be weighted average and obtain the section distribute traffic amount in this circulation; When the distribute traffic amount in adjacent two circulations equals definite value, can stop calculating; The distribute traffic amount obtaining in last circulation is the final volume of traffic.