CN103854479A - Method and device for measuring traffic flow data of road segment without detector - Google Patents

Abstract

The invention discloses a method and device for measuring traffic flow data of a road segment without a detector. The method includes the first step of collecting traffic flow data of road segments on the periphery of the road segment without the detector through a data collection module, the second step of selecting the peripheral road segments having the largest influence on measurement of the traffic flow data of a target road segment through a key point selection module and setting the peripheral road segments as key nodes, the third step of constructing a time-space model through a model construction model according to the correlation between the traffic flow data of the peripheral road segments and the traffic flow data of the road segment without the detector and the time correlation between the traffic flow data of the peripheral road segments and the traffic flow data of the road segment without the detector, and the fourth step of calculating the traffic flow data of the road segment without the detector through a data calculation module by using the time-space model according to the traffic flow data of the key nodes. Through the method, the time-space model is constructed according to the time correlation and the space correlation between the traffic flow data of the road segment without the detector and the traffic flow data of the peripheral road segments, and the traffic flow data of the road segment without the detector are calculated by using the time-space model according to the traffic flow data of the key nodes. The method has the advantages of being less in calculation amount, and high in efficiency and practicality.

Description

A kind of measuring method and device of sensorless road section traffic volume flow data

Technical field

The present invention relates to urban road traffic flow monitoring field, specifically, is a kind of measuring method and device of sensorless road section traffic volume flow data.

Background technology

It is many traffic application that traffic flow data obtains technology, no matter is traffic programme, or traffic control, traffic guidance etc. provide basic, the abundantest Data Source, are indispensable parts in traffic behavior evaluation and prediction theory system.

Urban road traffic flow data acquisition refers to the process of utilizing various device, technological means that the Static and dynamic data in traffic flow operational process are gathered, processed.At present, obtaining of traffic flow data mainly contained to two kinds of means: non-automatic acquisition technique and automatic acquisition technology.The principal feature of non-automatic acquisition technique is to need people's intervention just can complete the means of collection, as artificial acquisition method, instruction carriage investigation method etc.And automatic acquisition technology generally refers to that relying on traffic flow checkout equipment monitors moving vehicle, thereby realize the means that traffic flow parameter gathers.Automatic acquisition technology can upload to data center by the data of collection in real time by communication facilities, is main traffic flow data sampling means.According to the installation position of detecting device, automatic acquisition technology can be divided into fixed acquisition technique and mobile model acquisition technique.

Fixed acquisition technique refers to that the checkout equipment by being fixed on a certain place realizes the technology to traffic flow data sampling, corresponding checkout equipment is called stationarity detecting device, as common Data mining device, microwave detector, ultrasonic detector, infrared detector, magnetic detector, video detector etc.The data of fixed detecting device collection are mainly the basic traffic flow parameters of flow, average velocity and the time occupancy etc. of place section.Especially, the stationarity of fixed detecting device, makes the space of detecting device in road network arrange whether density can have significant impact by Overall Acquisition to traffic flow data.

Mobile model acquisition technique refers to that the constant mark thing using on the moving vehicle detection road that particular device is installed gathers the general name of the method for traffic flow data.Mobile model acquisition technique mainly for be the traffic flow data in stroke section, the data of collection are mainly journey time and travel speed.

Common mobile collection technology mainly comprises the acquisition technique based on Floating Car, the acquisition technique based on mobile phone location, the acquisition technique based on electronic tag and the acquisition technique based on car plate identification at present, is wherein most widely used with the acquisition technique based on Floating Car.

Acquisition technique based on Floating Car is to utilize the GPS equipment installed on vehicle position coordinates and the time data with certain sampling interval registration of vehicle, through lap over analysis, calculate the instantaneous velocity of vehicle and pass through journey time and the travel speed of specific road section.If have many cars through specific road section within the given period, can obtain average travel time and the travel speed in this section.Similarly, the key distinction is to determine the position coordinates of vehicle by the communication base station of mobile phone to acquisition technique based on mobile phone location., at the ad-hoc location in every section, sign is set based on electronic tag and the acquisition technique based on car plate identification, by the time of adjacent two signs, and then determine travel speed and the journey time of this vehicle on this section by more same label or car plate.If have many cars through specific road section within the given period, can obtain average travel time and the travel speed in this section.Although may there is larger difference in the principle of work of various mobile model acquisition techniques and system architecture, detect precision and the fiduciary level of data if will ensure, need to ensure to have in road network the existence of enough detection vehicles.

In the road traffic system of actual cities, fixed acquisition technique and mobile model acquisition technique are simultaneous.If Beijing's road real-time traffic flow data is mainly from two aspects: be distributed in the fixed such as toroid winding, the microwave traffic flow checkout equipment on through street and major trunk roads; Spread all over the taxi that more than 10000 of Urban Streets is provided with GPS equipment.These two kinds of acquisition techniques mutually supplement, mutually support, have enriched the content of traffic flow data sampling, have improved the coverage of traffic flow data sampling simultaneously.

At present, along with the development of urban economy, for the transport solution problem of blocking up, city road network construction project is more and more, and city road network density is increasing, and section and crossing are more and more, due to the reason of cost, traffic flow detecting device is only deployed in main roads and crossing, and this narrow layout causes existing in urban road network certain " vacuum " area.In addition,, due to traffic flow checkout equipment fault or other reasons, cause the traffic flow data sequence gathering may occur disappearance.Therefore, the traffic flow data under these two kinds of situations is obtained to technology research and there is important theory and realistic meaning.

Summary of the invention

The technical problem to be solved in the present invention is to provide a kind of measuring method and device of the sensorless road section traffic volume flow data that can obtain the traffic flow data that there is no the section of traffic flow checkout equipment and have the section of equipment failure.

In order to solve the problems of the technologies described above, the invention provides a kind of measuring method of sensorless road section traffic volume flow data, comprising:

The traffic flow data in the periphery section in A, collection sensorless section;

The periphery section that B, selection have the greatest impact to target road traffic delay DATA REASONING is made as key node;

C, according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, build space-time model;

D, utilize described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

Further, described step B specifically comprises:

Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in B1, calculating sensorless section;

B2, employing forward stepwire regression analytic approach are selected the periphery section that target road traffic delay DATA REASONING is had the greatest impact and are made as key node.

Further, described Pearson correlation coefficient is calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2{\mathrm{\Σ}(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m),

the arithmetic mean of X, it is the arithmetic mean of Y.

The present invention also provides a kind of measurement mechanism of sensorless road section traffic volume flow data, comprising:

Data collection module, for collecting the traffic flow data in periphery section in sensorless section;

Key point is selected module, for selecting the periphery section that target road traffic delay DATA REASONING is had the greatest impact to be made as key node;

Model construction module, for according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, builds space-time model;

Data computation module, for utilizing described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

Further, described key point selects module to comprise:

Pearson correlation coefficient unit, for calculating the Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in sensorless section;

Key point selected cell, is made as key node for adopting forward stepwire regression analytic approach to select the periphery section that target road traffic delay DATA REASONING is had the greatest impact.

Further, described Pearson correlation coefficient is calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2\mathrm{\Σ}{(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m),

the arithmetic mean of X,

it is the arithmetic mean of Y.

Sensorless road traffic delay data measuring method of the present invention and device, build space-time model according to correlativity spatially of the traffic flow data in sensorless section and periphery road section traffic volume flow data and correlativity in time, and select to have the greatest impact periphery section as key node, utilize space-time model to calculate the traffic flow data in sensorless section according to the interchange data of key node.In obtaining traffic flow data, its calculated amount is little, and efficiency is high, practical.

Brief description of the drawings

Fig. 1 is the process flow diagram of the measuring method of sensorless road section traffic volume flow data of the present invention.

Fig. 2 is the theory diagram of the measurement mechanism of sensorless road section traffic volume flow data of the present invention.

Embodiment

Below in conjunction with the drawings and specific embodiments, the invention will be further described, can be implemented, but illustrated embodiment is not as a limitation of the invention so that those skilled in the art can better understand the present invention also.

As shown in Figure 1, the measuring method of sensorless road section traffic volume flow data of the present invention, comprising:

Step 101, collects the traffic flow data in the periphery section in sensorless section;

Step 102, selects the periphery section that target road traffic delay DATA REASONING is had the greatest impact to be made as key node;

Step 103, according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, builds space-time model;

Step 104, utilizes described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

Wherein, described step B specifically comprises:

Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in B1, calculating sensorless section;

B2, employing forward stepwire regression analytic approach are selected the periphery section that target road traffic delay DATA REASONING is had the greatest impact and are made as key node.

Wherein, described Pearson correlation coefficient is calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2\mathrm{\Σ}{(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m),

the arithmetic mean of X, it is the arithmetic mean of Y.

As shown in Figure 2, the measurement mechanism of sensorless road section traffic volume flow data of the present invention, comprising:

Data collection module 201, for collecting the traffic flow data in periphery section in sensorless section;

Key point is selected module 202, for selecting the periphery section that target road traffic delay DATA REASONING is had the greatest impact to be made as key node;

Model construction module 203, for according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, builds space-time model;

Data computation module 204, for utilizing described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

Wherein, described key point selects module 202 to comprise:

Pearson correlation coefficient unit, for calculating the Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in sensorless section;

Key point selected cell, is made as key node for adopting forward stepwire regression analytic approach to select the periphery section that target road traffic delay DATA REASONING is had the greatest impact.

Certainly, Pearson correlation coefficient wherein is also calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2{\mathrm{\Σ}(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m),

the arithmetic mean of X,

it is the arithmetic mean of Y.

Using the road network between North 3rd Ring Road, Beijing and North 4th Ring Road as survey region, validity of the present invention is described below.In survey region, have 18 detecting devices and 17 sections, and these sections all belong to through street.

Data of the present invention are mainly derived from Beijing Expressway Traffic Flow detection system, this system has covered 15 interconnections between two rings, three rings, Fourth Ring, five rings and loop, the data that gather comprise flow, speed, time occupancy and large vehicle flowrate, and acquisition interval is 2 minutes.Interval was shorter due to 2 minutes, for the needs of traffic behavior parameter estimation, these data was processed to storage with 5 minutes intervals.

Parameter estimation model has the statistical indicator of a lot of tolerance validity, as root-mean-square error (RMSE), mean absolute deviation (MAD), and absolute percent error (APE) etc.Absolute percent error can be the sensitiveest to the situation of change of data, and because the variation of traffic flow between peak and non-peak period is obvious, and MAPE has reflected the average amplitude of data variation.Therefore, select the measure of effectiveness index of mean absolute percentage error (MAPE) as this paper model herein.

$\mathrm{MAPE}=\frac{1}{n}\underset{t=1}{\overset{n}{\mathrm{\Σ}}}\left|\frac{y_t-{\tilde{y}}_t}{y_t}\right|\×100\%,$ Accuracy＝1-MAPE (0-1)

Here y, _trepresent the observed reading in t moment,

show the estimated value in t moment.

For the magnitude of traffic flow of survey region road network, to experiment Analysis of the present invention.In order fully to reflect validity of the present invention, embodiment estimates respectively the traffic behavior parameter of peak and off-peak period.Definition only refers to the morning peak period (7:00～9:00) peak period, and off-peak period refers to 11:00 to 13:00.In addition, data are divided into two classes: training set and test set.From on August 4,28 days to 2009 July in 2009, one-week complete data are as the training set of model.And using the data on August 5th, 2009 as test set.The three loop sections of experiment section from River of Ten Thousand Springs bridge to Haidian bridge, are laid with detecting device S5, and length is about 620 meters.

The first step, the detector data of all upstream and downstreams in the distance range of object detector S5 constraint will be used as input data set and select key node.After the average overall travel speed statistics of different periods of through street, Beijing speed, can obtain, the average velocity of morning peak period is about 35 kilometers/hour, and the average velocity of off-peak period is about 55 kilometers/hour.For the sake of simplicity, we allow the radius of restriction range equal interval and average velocity long-pending estimated time, and all detecting devices in about beam radius are all using the candidate as key node.Because the distance between detecting device is known, therefore, the Candidate Set s that we can obtain in peak period _peak{ 1,2,7,8,10,11,12} and off-peak period Candidate Set s _off-peak{ 1,2,3,4,7,8,9,10,11,12,13,14,15,16}.

Second step, the calculating of Pearson correlation coefficient.In the preliminary election process of model, Candidate Set s _peakand s _off-peakinternal detector different time lag as (t-1), (t-1) ..., (t-d) } should take into full account.In the present embodiment, definition d=5.Table 1 has been listed respectively maximally related 15 the flow sequences in peak and off-peak period and detecting device S5 and corresponding related coefficient.Be very easy to find, flow sequence and the target detection data of hysteresis have larger related coefficient, and this absolutely proves considers that this factor is necessary.

The 3rd step, is used forward stepwire regression selection and the maximally related data of object detector as key node collection sd, thereby build estimation model, object detector S5 traffic behavior parameter is estimated.

Test performance of the present invention, select other three estimators to compare: first is the space-time arma modeling that there is no preliminary election process; Another is arest neighbors regression model, this model considered with restricted area in neighbours' detector data, and adopt be that maximum likelihood method is trained.The 3rd is linear regression model (LRM) progressively, and this model has only been considered current traffic behavior, but there is no the restriction of region distance.

Maximally related 15 the flow sequences of peak and off-peak period and detecting device S5:

SPSS is business advanced in the world, and the statistical software of government and academic institution, for solving the problem of enterprise and research.Due to its good statistical function, and easily use.Mainly utilize " time series " and " recurrence " statistical module in SPSS (version 16.0.0) to test our model herein.

Experimental result statistics:

Compare the present invention and there is no the space-time arma modeling of process in advance, aspect model accuracy, the former will be higher than the latter, in the non-peak hours, but not as the latter in the precision of peak period; But aspect computing time, no matter be peak period or off-peak period, the latter is obviously not as the former.The computing time of arest neighbors recurrence and successive Regression model is shorter than the present invention, but precision (being all starkly lower than 75%) on the low side is difficult to meet actual needs.The present invention is respectively 83.2%, 86.4% in the precision of peak period and off-peak period, not only has higher precision, and stability is also better, can meet the demand of practical application.

The above embodiment is only the preferred embodiment for absolutely proving that the present invention lifts, and protection scope of the present invention is not limited to this.What those skilled in the art did on basis of the present invention is equal to alternative or conversion, all within protection scope of the present invention.Protection scope of the present invention is as the criterion with claims.

Claims (6)

1. a measuring method for sensorless road section traffic volume flow data, is characterized in that, comprising:

The traffic flow data in the periphery section in A, collection sensorless section;

The periphery section that B, selection have the greatest impact to target road traffic delay DATA REASONING is made as key node;

C, according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, build space-time model;

D, utilize described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

2. the measuring method of sensorless road section traffic volume flow data according to claim 1, is characterized in that, described step B specifically comprises:

Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in B1, calculating sensorless section;

B2, employing forward stepwire regression analytic approach are selected the periphery section that target road traffic delay DATA REASONING is had the greatest impact and are made as key node.

3. the measuring method of sensorless road section traffic volume flow data according to claim 2, is characterized in that, described Pearson correlation coefficient is calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2{\mathrm{\Σ}(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m), the arithmetic mean of X,

it is the arithmetic mean of Y.

4. a measurement mechanism for sensorless road section traffic volume flow data, is characterized in that, comprising:

Data collection module, for collecting the traffic flow data in periphery section in sensorless section;

Key point is selected module, for selecting the periphery section that target road traffic delay DATA REASONING is had the greatest impact to be made as key node;

Model construction module, for according to the correlativity on correlativity and sensorless road traffic delay data time between periphery section and sensorless road section traffic volume flow data, builds space-time model;

Data computation module, for utilizing described space-time model to calculate the traffic flow data in sensorless section according to the traffic flow data of described key node.

5. the measurement mechanism of sensorless road section traffic volume flow data according to claim 4, is characterized in that, described key point selects module to comprise:

Pearson correlation coefficient unit, for calculating the Pearson correlation coefficient between the traffic flow data in periphery section and the traffic flow data in sensorless section in sensorless section;

Key point selected cell, is made as key node for adopting forward stepwire regression analytic approach to select the periphery section that target road traffic delay DATA REASONING is had the greatest impact.

6. the measurement mechanism of the sensorless road section traffic volume flow data described in a sharp claim 5, is characterized in that, described Pearson correlation coefficient is calculated by following formula:

$r=\frac{\mathrm{\Σ}(x_i-\stackrel{\‾}{x})(y_i-\stackrel{\‾}{y})}{\sqrt{\mathrm{\Σ}{(x_i-\stackrel{\‾}{x})}^2{\mathrm{\Σ}(y_i-\stackrel{\‾}{y})}^2}},$

Wherein, r is Pearson correlation coefficient, and X represents n sample { x _i(i=1 ..., set n), Y represents m sample { y _i(i=1 ..., set m),

the arithmetic mean of X,

it is the arithmetic mean of Y.

Images