CN106504535A - A Traffic Distribution Prediction Method Combining Gravity Model and Fratar Model - Google Patents

Abstract

The invention discloses a kind of combination Gravity Models and the trip distribution modeling method of Fratar models, comprise the steps：The generation for gathering each cell first attracts the volume of traffic and the distribution of present situation OD；Secondly demarcate the generation without constraint Gravity Models parameter and each cell of the prediction non-coming year and attract the volume of traffic；Then apply that both demarcated to calculate non-coming year OD distributions without constraint Gravity Models；Finally application Fratar model runnings once obtain new non-coming year prediction OD distributions；Convergence judgement is carried out with last time circulation result to operation result, the OD distributions for being met convergence criterion are each minizone OD forecast of distribution final results of the non-coming year.The distribution forecasting method combines the advantage of Gravity Models and Fratar models, both present situation trip distributed intelligence had been taken full advantage of, the impact that the change and Land_use change that road network can be considered simultaneously again is produced to people's trip, improves the applicability of the accuracy and forecast model for predicting the outcome.

Description

A Traffic Distribution Prediction Method Combining Gravity Model and Fratar Model

technical field

The invention relates to a traffic distribution prediction method combining a gravity model and a Fratar model, and belongs to the technical field of traffic demand and traffic distribution prediction.

Background technique

The rational planning of transportation projects is inseparable from accurate transportation demand analysis and forecasting. As the key technology of traffic planning, traffic demand analysis and forecasting determine the accuracy and reliability of grasping the future traffic development trend, thus affecting the decision-making of traffic departments and planners. Today, as the level of urbanization is getting higher and higher, a large number of new cities or new urban planning areas (hereinafter referred to as new urban areas) have emerged, which has brought new problems and challenges to urban transportation planning, especially in the analysis of traffic demand. and the prediction stage. At present, the traditional procedural traffic demand analysis model is mainly used in traffic demand analysis and forecasting in traffic planning, that is, the four-stage forecasting model of traffic generation, traffic distribution, traffic mode division and traffic allocation. For the second stage of the four-stage method traffic distribution prediction, many scholars have proposed different prediction models and methods.

At present, the current state method (also known as the growth coefficient method) is mostly used in the traffic distribution prediction of the highway feasibility study, and the Fratar method is widely used by planning managers because of its fast convergence speed. The basic assumption of the Fratar method is: the travel volume between traffic areas has nothing to do with the change of the road network structure, or there is no major change in the road network in the forecast year. Therefore, the Fratar method has an unavoidable obvious defect, that is, it only uses the growth rate as the only indicator to realize the future traffic volume, without taking into account many factors that affect the traffic distribution in the network, so in new traffic modes, new roads, etc. , new toll policies, or new cell generation cannot describe changes in traffic distribution. In addition, the current state method is highly dependent on the accuracy of the travel distribution in the base year, and the reliability of the travel distribution in the future year cannot exceed the base year, and any errors in the travel distribution in the base year will be eliminated in the calculation process. enlarge. In contrast, the "gravity model" method, or "comprehensive model" method, considers that the traffic distribution between districts is affected by all traffic impedances such as distance between districts, running time, and cost, that is, the distance between districts The travel distribution among the districts is directly proportional to the travel attraction of each district, while the traffic resistance between the same districts is inversely proportional. It has higher applicability and accuracy for the traffic distribution prediction of some new urban areas. However, the gravity model method is completely based on the consideration of factors affecting travel distribution, lacks the analysis of people's travel behavior, and does not make full use of the current travel distribution data, and the prediction results may have certain deviations from the actual situation.

Contents of the invention

Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a traffic distribution prediction method combining the gravity model and the Fratar model. And the impact of land use on travel distribution, fully combined with the actual situation of the current travel distribution, and at the same time solve the problems of the single growth model of the Fratar method and the difficulty of obtaining relevant influencing factors of the gravity model method, it can make a reasonable prediction of the traffic distribution in the future year.

Technical scheme: in order to achieve the above object, the technical scheme adopted in the present invention is:

A traffic distribution prediction method combining gravity model and Fratar model, comprising the following steps:

Step 10) collect the current situation of each community to attract traffic volume, current OD distribution and related basic data, wherein, the current situation of each community to attract traffic volume includes the trip generation volume Oi of the community _i and the travel attraction volume D of the community j _j , i=1, 2...n, j=1, 2...n, n represents the number of cells.

Step 20) Calibrate the unconstrained gravity model, including step 201) to step 204):

Step 201) determine the unconstrained gravity model form: q _ij represents the traffic volume between cell i and j, c _ij represents the impedance between cell i and j, α, β, γ are unconstrained gravity model parameters to be calibrated.

Step 202) to Take the logarithm on both sides, and get ln(q _ij )=lnα+βln(O _i D _j )-γln(c _ij ).

Step 203) Let Y=ln(q _ij ), a ₀ =lnα, a ₁ =β, a ₂ =-γ, X ₁ =ln(O _i D _j ), X ₂ =ln(c _ij ), then Y =a ₀ +a ₁ X ₁ +a ₂ X ₂ , where a ₀ , a ₁ , a ₂ are coefficients to be calibrated, and X ₁ , X ₂ , Y are vectors containing sample data sets.

Step 204) Determine the sample data sets X ₁ , X ₂ , Y according to the current situation of attracting traffic volume and current OD distribution in each district, use the least square method to calibrate the sample data, and determine the parameters of the unconstrained gravity model.

Step 30) Based on the current situation of attracting traffic volume, predict the occurrence and attracting traffic volume of each district in the future year, and the occurrence and attracting traffic volume of each district in the future year includes the trip generation volume P _i and the i-th district's future year The travel attraction A _i of the community in the coming year.

Step 40) Substitute the occurrence and attraction traffic volume of each district in the future year obtained in step 30) into the unconstrained gravity model calibrated in step 20), and calculate the future year OD distribution q _ij .

Step 50): Using the OD distribution q _ij obtained by applying the unconstrained gravity model in step 40) as the initial OD, run the Fratar model once to obtain a new predicted OD distribution q _ij ¹ for the future year, including steps 501) to 503).

Step 501): According to the unconstrained gravity model, the OD distribution q _ij between the cells i and j is obtained, and the traffic volume generated and attracted by each cell is obtained by summarizing:

Step 502): Calculate Fratar model correlation coefficient Among them, F _Oi represents the travel occurrence convergence coefficient of the i-th cell when the Fratar model is applied, F _Dj represents the travel attraction convergence coefficient of the j-th cell, L _ij represents the travel occurrence adjustment coefficient of the i-th cell, and L _ji represents the j-th cell The travel attraction adjustment coefficient of a district.

Step 503): Solve the new future year OD distribution prediction q _ij ¹ =q _ij *F _Oi *F _Dj *(L _ii +L _jj )/2.

Step 60): Convergence Judgment: According to the OD distribution q _ij ¹ obtained in one cycle of the Fratar model in Step 50), the new future annual forecasted traffic attraction of each district is obtained as

If the traffic volume errors of each district are within the acceptable range, that is:

Go to step 70), otherwise let Go to step 40).

Step 70) The prediction result satisfies the convergence condition, and the cycle ends, and the OD distribution at this time is the prediction result of the OD distribution among the cells in the future year.

Preferably: the relevant basic data in the step 10) includes data of community population, area, land use, and location factors.

Preferably: in the step 30), the methods for predicting the traffic volume of each subdistrict in the future based on the current volume of traffic volume include the original unit method, the growth rate method, the cross classification method, and the function method.

Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects:

1) Inheriting the advantages of the gravity model, it can comprehensively consider the impact of various factors such as road network changes and land use on the generation and attraction of traffic in the community. At the same time, it has high accuracy and applicability for traffic demand prediction of new cities or urban planning new areas.

2) Absorbing and merging the advantages of the Fratar growth model, making full use of the available travel distribution information of the current community, to a certain extent, controlling the deviation between the predicted results of the travel distribution in the future and the actual travel distribution, and ensuring the credibility of the prediction results.

3) When the current OD distribution is difficult to obtain or the base year distribution information is missing, the gravity model can still be used to initialize the predicted year OD, and then the Fratar model can be used to gradually adjust the predicted results.

Description of drawings

Fig. 1 is a flowchart of the present invention.

detailed description

Below in conjunction with accompanying drawing and specific embodiment, further illustrate the present invention, should be understood that these examples are only for illustrating the present invention and are not intended to limit the scope of the present invention, after having read the present invention, those skilled in the art will understand various aspects of the present invention All modifications of the valence form fall within the scope defined by the appended claims of the present application.

A traffic distribution prediction method combining the gravity model and the Fratar model, as shown in Figure 1, includes the following steps:

Step 10) Collect the current traffic volume O _i , D _i , current situation OD distribution and related basic data of each district, wherein the collection of relevant basic data includes: the districts that have a direct or indirect impact on the traffic traffic volume generated by the traffic district Collection of data related to population, area, land use, and location.

Step 20) Calibrate the unconstrained gravity model, including step 201) to step 204):

Step 201) determine the unconstrained gravity model form: Among them, i=1, 2...n, j=1, 2...n, n represents the number of cells, q _ij represents the traffic volume between cells i and j, c _ij represents the impedance between cells i and j, O _i represents the amount of trips in cell i, D _j represents the amount of travel attraction in cell j, and α, β, γ are the parameters to be calibrated in the unconstrained gravity model.

Step 202) to Take the logarithm on both sides, get ln(q _ij )=lnα+βln(O _i D _j )-γln(c _ij )

Step 203) Let Y=ln(q _ij ), a ₀ =lnα, a ₁ =β, a ₂ =-γ, X ₁ =ln(O _i D _j ), X ₂ =ln(c _ij ), then Y =a ₀ +a ₁ X ₁ +a ₂ X ₂ , where a ₀ , a ₁ , a ₂ are coefficients to be calibrated, and X ₁ , X ₂ , Y are vectors containing sample data sets.

Step 204) Determine the sample data sets X ₁ , X ₂ , Y according to the current situation of attracting traffic volume and current OD distribution in each district, use the least square method to calibrate the sample data, and determine the parameters of the unconstrained gravity model.

Step 30) Based on the current generated and attracted traffic volume, predict the generated and attracted traffic volume of each district in the future year, and the generated and attracted traffic volume of the i-th district is denoted as P _i and A _i respectively. Among them, the predictive methods for the generation and attraction traffic volume of the community in the future include the original unit method, the growth rate method, the cross classification method, and the function method.

Step 40) Apply the unconstrained gravity model calibrated in step 20), substitute the traffic volumes P _i and A _i that will be attracted in the future year, and calculate the OD distribution in the future year.

Step 50): Use the OD distribution obtained by applying the unconstrained gravity model in step 40) as the initial OD, and apply the Fratar model to perform a convergence to obtain a new forecasted OD distribution in the future, including steps 501) to 503):

Step 501): According to the unconstrained gravity model, the OD distribution q _ij between the cells i and j is obtained, and the traffic volume generated and attracted by each cell is summarized according to the OD table:

Step 502): Calculate Fratar model related parameters:

Step 503): Solve the new future year OD distribution forecast

Step 60): Convergence Judgment: According to the OD distribution q _ij ¹ obtained in one cycle of the Fratar model in Step 50), the new future annual forecasted traffic attraction of each district is obtained as

If the traffic volume errors of each district are within the acceptable range, that is:

Go to step 70), otherwise let Go to step 40).

Step 70) The prediction result satisfies the convergence condition, and the cycle ends, and the OD distribution at this time is the prediction result of the OD distribution among the cells in the future year.

The invention provides a traffic distribution prediction method combining the gravity model and the Fratar model. The prediction method can integrate the advantages of the gravity model method and the Fratar model, and not only considers the influence of road network changes and land use on travel distribution, but also fully combines the advantages of the gravity model and the Fratar model. The actual situation of the current travel distribution, at the same time solves the problems of the single growth mode of the Fratar method and the accuracy of the relevant influencing factors of the gravity model method, and makes a reasonable prediction of the traffic distribution in the next year.

The present invention fully considers the actual situation that may occur in the traffic distribution prediction process, combines the gravity model method and the Fratar growth rate method, and reasonably predicts the travel distribution in the future on the premise of utilizing the current travel distribution to obtain information to the greatest extent. The method of the present invention first uses the gravity model to initialize and predict the traffic distribution in the future, and then converges the initial prediction results successively through the Fratar model, and repeats the above two steps until the final convergence condition is met. The method of the present invention does not substantially change the basic principle of the existing distribution prediction model, but it combines the advantages of the gravity model and the Fratar model, which not only makes full use of the current travel distribution information, but also considers the impact of changes in the road network and land use. The impact of people's travel improves the accuracy of forecast results and the applicability of forecast models. Therefore, the present invention has strong practicability and can be applied to traffic distribution and traffic demand forecasting in new and old urban areas, and has important practical significance.

A specific embodiment is given below.

Taking the travel distribution prediction of three traffic districts in a certain city in Jiangsu Province as an example, the practicability and advantages of the inventive method are illustrated.

Step 10) Collect the current generation and attraction traffic volume P and A of each district, the current OD distribution and related basic data, count the current travel time between each district and within the district, and predict the future travel time.

Table 1 Status OD distribution table

Table 2 Current driving time

Table 3 Future travel time

Step 20) Calibrate the unconstrained gravity model:

Step 201) determine the unconstrained gravity model form:

Step 202) to Take the logarithm on both sides, and get ln(q _ij )=lnα+βln(O _i D _j )-γln(c _ij ).

Step 203) Let Y=ln(q _ij ), a ₀ =lnα, a ₁ =β, a ₂ =-γ, X ₁ =ln(O _i D _j ), X ₂ =ln(c _ij ), then Y ＝a ₀ +a ₁ X ₁ +a ₂ X ₂ , where a ₀ , a ₁ , a ₂ are coefficients to be calibrated, c _ij is the travel time between plots and inside plots, X ₁ , X ₂ , Y are included samples A vector of datasets.

Step 204) Determine the sample data sets X ₁ , X ₂ , Y according to the current situation of attracting traffic volume and current OD distribution in each district, use the least square method to calibrate the sample data, and determine the parameters of the unconstrained gravity model.

Table 4 Gravity model calibration sample data

Solve the binary linear equation parameters a ₀ =-2.084, a ₁ =1.173, a ₂ =-1.455, further solve the gravity model parameters α=0.124, β=1.173, γ=1.455, that is, the calibrated gravity model is:

Step 30) Based on the current generated and attracted traffic volume, predict the generated and attracted traffic volume of each district in the future year, and the generated and attracted traffic volume of the i-th district is denoted as P _i and A _i respectively.

Table 5 Attracting traffic volume in the coming years

Step 40) Apply the unconstrained gravity model calibrated in step 20), substitute the traffic volumes P _i and A _i that will be attracted in the future year, and calculate the OD distribution in the future year.

Table 6 OD distribution table in the coming years

Step 50): Use the OD distribution table obtained by applying the unconstrained gravity model in step 40) as the initial OD, and apply the Fratar model to perform a convergence to obtain a new forecasted OD distribution in the future, including steps 501) to 502):

Step 501): According to the unconstrained gravity model, the OD distribution q _ij between the cells i and j is obtained, and the traffic volume generated and attracted by each cell is summarized according to the OD table:

Table 7. Gravity model and actual forecasts to attract traffic in future years

Step 502): seek

Table 7 Fratar model related parameters

Step 503): Calculate the new future year OD distribution prediction q _ij ¹ =q _ij *F _Oi *F _Dj *(L _ii +L _jj )/2.

Table 8 OD distribution table in the coming years

Step 60): Convergence judgment:

According to the OD distribution q _ij ¹ obtained in one cycle of the Fratar model in step 50), the new forecasted traffic volume of each district in the future is obtained as

Table 9 Forecasting traffic volume statistics in future years

Judging the ratio of the amount of traffic attracted by each community after this cycle to the result of the previous cycle Whether it meets the acceptable range:

Table 10 two cycle error statistics

The convergence condition is not satisfied, so let Go to step 40).

Step 40) to step 60) after carrying out 8 cycles, obtain:

Table 11 OD distribution table in the coming years

Table 12 Statistics of two cycle errors

The convergence condition is satisfied, namely: Go to step 70),.

Step 70) ends the cycle, and the OD distribution at this time is the final result of OD distribution prediction among the cells in the future year.

Table 13 OD Distribution Table for Future Years

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made. It should be regarded as the protection scope of the present invention.

Claims (3)

1. a traffic distribution prediction method combining gravity model and Fratar model, is characterized in that, comprises the following steps: Step 10) collect the current situation of each community to attract traffic volume, current OD distribution and related basic data, wherein, the current situation of each community to attract traffic volume includes the trip generation volume Oi of the community _i and the travel attraction volume D of the community j _j , i=1, 2...n, j=1, 2...n, n represents the number of cells; Step 20) Calibrate the unconstrained gravity model, including step 201) to step 204): Step 201) determine the unconstrained gravity model form: q _ij represents the traffic volume between cell i and j, c _ij represents the impedance between cell i and j, α, β, γ are unconstrained gravity model parameters to be calibrated; Step 202) to Take the logarithm on both sides, get ln(q _ij )=lnα+βln(O _i D _j )-γln(c _ij ); Step 203) Let Y=ln(q _ij ), a ₀ =lnα, a ₁ =β, a ₂ =-γ, X ₁ =ln(O _i D _j ), X ₂ =ln(c _ij ), then Y ＝a ₀ +a ₁ X ₁ +a ₂ X ₂ , where a ₀ , a ₁ , a ₂ are coefficients to be calibrated, X ₁ , X ₂ , Y are vectors containing sample data sets; Step 204) Determine the sample data sets X ₁ , X ₂ , Y according to the current situation of each district to attract traffic and the current OD distribution, use the least square method to calibrate the sample data, and determine the parameters of the unconstrained gravity model; Step 30) Based on the current situation of attracting traffic volume, predict the occurrence and attracting traffic volume of each district in the future year, and the occurrence and attracting traffic volume of each district in the future year includes the trip generation volume P _i and the i-th district's future year The travel attraction A _i of the community in the coming year; Step 40) substituting the occurrence and attraction traffic volume of each subdistrict obtained in step 30) into the unconstrained gravity model calibrated in step 20), and calculating the future year OD distribution q _ij ; Step 50): Using the OD distribution q _ij obtained by applying the unconstrained gravity model in step 40) as the initial OD, run the Fratar model once to obtain a new forecasted OD distribution q _ij ¹ for the future year, including steps 501) to 503); Step 501): According to the unconstrained gravity model, the OD distribution q _ij between the cells i and j is obtained, and the traffic volume generated and attracted by each cell is obtained by summarizing: Step 502): Calculate Fratar model correlation coefficient Among them, F _Oi represents the travel occurrence convergence coefficient of the i-th cell when the Fratar model is applied, F _Dj represents the travel attraction convergence coefficient of the j-th cell, L _ij represents the travel occurrence adjustment coefficient of the i-th cell, and L _ji represents the j-th cell The travel attraction adjustment coefficient of a community; Step 503): Solve the new future year OD distribution prediction q _ij ¹ =q _ij *F _Oi *F _Dj *(L _ii +L _jj )/2; Step 60): Convergence Judgment: According to the OD distribution q _ij ¹ obtained in one cycle of the Fratar model in Step 50), the new future annual forecasted traffic attraction of each district is obtained as ${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{{\mathrm{<span\; class="notranslate">\; q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{{\mathrm{<span\; class="notranslate">\; q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}$ If the traffic volume errors of each district are within the acceptable range, that is: $\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; \&lsqb;</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.99</span>}<span\; class="notranslate">\; <</span>\frac{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; 1.01</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; \&lsqb;</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.99</span>}<span\; class="notranslate">\; <</span>\frac{{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; 1.01</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; no</span>}$ Go to step 70), otherwise let Go to step 40); Step 70) The prediction result satisfies the convergence condition, and the cycle ends, and the OD distribution at this time is the prediction result of the OD distribution among the cells in the future year. 2. the traffic distribution forecasting method combining gravity model and Fratar model according to claim 1, is characterized in that: in described step 10), relevant basic data comprises community population, area, land utilization, location factor data. 3. the traffic distribution forecasting method combining gravity model and Fratar model according to claim 1, it is characterized in that: in described step 30), attract the traffic volume that takes place in the present situation as the basis and predict the occurrence of attracting traffic volume in the future year of each subdistrict Methods include original unit method, growth rate method, cross classification method, and function method.