CN104408915B - A kind of method of estimation of traffic state data and system - Google Patents
A kind of method of estimation of traffic state data and system Download PDFInfo
- Publication number
- CN104408915B CN104408915B CN201410617762.6A CN201410617762A CN104408915B CN 104408915 B CN104408915 B CN 104408915B CN 201410617762 A CN201410617762 A CN 201410617762A CN 104408915 B CN104408915 B CN 104408915B
- Authority
- CN
- China
- Prior art keywords
- data
- traffic
- sample
- data sample
- traffic data
- 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
- G08G1/0133—Traffic data processing for classifying traffic situation
Abstract
The invention discloses a kind of method of estimation of traffic state data and system.Method therein includes：Traffic data sample is obtained in real time；If the quantity of the Current traffic data sample gathered in current slot gathers smallest sample amount more than default traffic, data fitting is carried out to Current traffic data sample according to default training pattern then, it is modified based on data of the error ratio criterion after correction matrix is to fitting, and the estimated value according to revised data acquisition traffic state data；Otherwise, it was that Current traffic data sample, the traffic data sample of a upper time period and history quantized data are respectively provided with corresponding weight coefficient；And according to Current traffic data sample, the traffic data sample of a upper time period, history quantized data and respective weight coefficient, it was calculated the estimated value of traffic state data.By using method and system provided by the present invention, accuracy and the reliability of traffic state data estimation can be effectively improved.
Description
Technical field
The present invention relates to technical field of transportation, more particularly to a kind of method of estimation of traffic state data and system.
Background technology
At present, as China's transportation modernization process is quickly propelled, urban traffic jam is on the rise, and has a strong impact on
The convenience of resident trip and economic development.In order to efficiently solve urban traffic blocking present situation, vehicle supervision department has must
Traffic real-time running state is monitored, to find congested link carry out effectively, dredge in time and scheduling.
The real-time traffic dynamic state information for obtaining urban road is the important foundation for solving the problems, such as urban congestion, hands at present
Logical dynamic information collection mode mainly includes：Fixed point traffic information collection and mobile traffic information gathering.Fixed point transport information is adopted
Mode set is mainly using testing equipments such as earth magnetism, microwave, video, bayonet sockets.By taking microwave monitor as an example, it is that one kind is supervised with radar
The detector of microwave transmission traffic data is surveyed, by launching cline frequency modulated microwave on detection ground, when vehicle is through throwing
When penetrating area, microwave monitor can receive a microwave signal, draw vehicle by calculating receives frequency and the running parameter of time
Speed and length.The acquisition mode is mainly used in the transport information for pinpointing detection urban road emphasis section, but, the collection
There is the problems such as detection range is little, and installation and maintenance cost are higher in mode.The mainly floating of mobile traffic information gathering mode
Car data is gathered, using Floating Car its operational factor of real time record in the process of moving for loading location equipment, by positional information
Associate on room and time, obtain the vehicle running state of target road section and predict traffic.But, in the collection
In mode, can there is " urban canyon effect " in gps system signal, cause gps system obtain corresponding period city vehicle
Transport condition；Additionally, the reflection of gps system also has appreciable impact to GPS location precision.
Relevant traffic state data (for example, speed) method of estimation is a lot.For example, based on speed weighting algorithm, main
There is estimated below method：One be based on simple majority according to source data fusion method, the method using multiple heterogeneous data sources and
Each data source weight coefficient carries out speed fusion.Two is based on data anastomosing algorithm, mainly melts including neutral net in the method
Matched moulds type, Kalman filtering, D-S evidential reasonings and fuzzy logic scheduling algorithm.Wherein, first method is needed according to concrete traffic
Environment determines the weight coefficient of various data sources, and needs the similar temporal characteristics (for example, peak, Ping Feng, night etc.) of consideration
Weight coefficient with factors such as target roads (for example, highway, through street, surface road etc.)；Second method is mainly sharp
Traffic data fusion and prediction are carried out with data fusion framework and algorithm, the algorithm of this mode is complex, biases toward theory and grinds
Study carefully, practical application is less.
In the prior art, it was also proposed that the method for estimation of some traffic state datas.
For example, in only one China patent application (application number 201410084202.9, denomination of invention：" based on multiple positioning numbers
Method according to road fusion travel speed is calculated ") in, it is proposed that a kind of based on multiple location datas calculating road fusion range of driving car
The method of speed.The method by carrying out speed fusion treatment to mobile phone terminal positioning data and Floating Car GPS location data, to obtain
Average range of driving speed by way of in the target road section of road, and then obtain the traffic behavior of urban road.This method solve single floating
Motor-car data source collection mode of transportation does not collect friendship in city proper main roads partial period and city suburbs highway
The situation of logical state, and single gps data source collection mode of transportation is effectively improved in the less road section of sample size
Upper traffic state quality problem.But, the method needs to consider temporal characteristics (for example, peak, Ping Feng, night etc.) and target track
Road (for example, highway, through street, surface road etc.) coefficient, although choose two kinds of signal sources and merged, but above-mentioned time
The value of characteristic coefficient and target road coefficient is difficult to determine there is a problem of unstable.
In another Chinese patent application (application number 200910199819.4, denomination of invention：" put based on Floating Car speed
The traffic state data method of estimation of reliability ") in, based on the analysis to floating car data, confidence level factor is introduced, and is resit an exam
Consider floating car data sample size, speed and traffic behavior continuity in time, it is proposed that a kind of excellent based on confidence level
The method of estimation of the traffic state data (speed) of change.By confidence majority vehicle, rapid vehicle and merged a upper period and go through
History data, it is achieved that the estimation to Floating Car average speed, effectively reduce impact of the abnormal data to result of calculation, make calculating
As a result closer to true road conditions, while traffic state data estimation problem when solving sample size deficiency, effectively increases traffic
The accuracy of state estimation and stationarity.But, the method in the case where sample size is less still with reference on a period
Sample data, fusion treatment estimates the period traffic state data with this, when sample data is more than minimum sample flow rate, does not examine
Consider impact of the abnormal data to result, directly using speed weighting algorithm, there is certain error；In addition, the method is not with synchronized
Degree gear arranges speed weight coefficient, and the determination of the weight coefficient size of high speed is not very clear and definite.
Additionally, also proposed a kind of many car speed blending algorithms based on floating car data in prior art.In the algorithm
In, construct a kind of new many car speed blending algorithms based on floating car data.The algorithm is from floating vehicle travelling feature, many power
Weight coefficient and various road conditions state angularly, consider characterize real-time road when many car samples of Floating Car between general character with individual
Sex differernce goes to merge many car speeds, improves the accuracy of real-time road, and can quickly adjust phase according to actual traffic environment
Related parameter, has carried out assessment checking finally by proof analysis to its accuracy, as a result shows to effectively improve dynamic traffic letter
The accuracy of breath, with good practicality.But, the algorithm needs to determine parameter and weight system according to actual traffic environment
, therefore there is parameter and coefficient estimation error in number.
In summary, there are some above-mentioned to the method for estimation of traffic state data in prior art.Therefore,
Need badly and propose a kind of new method of estimation to traffic state data, such that it is able to effectively improve traffic state data estimation
Accuracy and reliability.
Content of the invention
In view of this, the present invention provides a kind of method of estimation of traffic state data and system, such that it is able to effectively carry
Accuracy and reliability that high traffic state data is estimated.
Technical scheme is specifically realized in：
A kind of method of estimation of traffic state data, the method include：
Traffic data sample is obtained in real time；
When the quantity of the Current traffic data sample gathered in current slot gathers most sample more than default traffic
During this amount, data fitting is carried out to Current traffic data sample according to default training pattern, the data after being fitted；It is based on
Data of the error ratio criterion after correction matrix is to fitting are modified, and according to revised data acquisition traffic shape
The estimated value of state parameter；
When the quantity of the Current traffic data sample gathered in current slot is gathered less than or equal to default traffic
During smallest sample amount, it was that Current traffic data sample, the traffic data sample of a upper time period and history quantized data are set respectively
Put corresponding weight coefficient；And according to Current traffic data sample, the traffic data sample of a upper time period, history quantized data
And respective weight coefficient, it is calculated the estimated value of traffic state data.
Preferably, the traffic state data is speed.
Preferably, the real-time acquisition traffic data sample includes：
Real-time Collection original traffic data；
According to the original traffic data and road section information that are gathered, positional information is generated by map-matching algorithm and associates road
The gps data of section, and the gps data is quantified, the speed probability data after being quantified, by the speed after the quantization
Degree probability data is used as traffic data sample.
Preferably, the original traffic data for being gathered include：
Car number, taxi company, reception time, longitude, latitude, azimuth, instantaneous velocity and operation state.
Preferably, before data fitting is carried out according to default training pattern to Current traffic data sample, the method
Still further comprise：
According to the history quantized data of traffic state data, and corresponding training data is calculated using curve fitting algorithm
The parameters of equation, obtain training data equation, and pre-set corresponding training pattern according to the training data equation.
Preferably, the data based on error ratio criterion after correction matrix is to fitting are modified, and root
Include according to the estimated value of revised data acquisition traffic state data：
Using the default training pattern, the traffic shape under current road segment current time is obtained according to history quantized data
The probability history matched curve of state parameter；
According to the probability function that acquired traffic data sample determines traffic state data；
Probability function according to probability history matched curve and traffic state data obtains the error ratio matrix K of sample；
Complementary matrix I is obtained according to the error ratio matrix K；
According to the correction matrix Λ that complementary matrix I obtains each traffic data sample；
According to the correction matrix Λ, correction value Z' of revised each traffic data sample is obtained；
According to the correction value of each traffic data sample, traffic state data is calculated by Weighted Average Algorithm
Estimated value.
Preferably, obtaining error ratio matrix K by formula as described below：
Wherein, K is expressed as [ε1,ε2,…εn], Ψ be probability history matched curve, vi(i=1,2,3 ..., n) it is
Traffic state data, probability functions of the Z for traffic state data.
Preferably, obtaining complementary matrix I by formula as described below：
Preferably, the correction matrix Λ is expressed as：
Λ=[w1,w2,……wn]；
Wherein, wiFor traffic state data viModifying factor, and
Preferably, obtaining correction value Z' of traffic data sample by formula as described below：
Preferably, described for Current traffic data sample, the traffic data sample of a upper time period and history quantized data
Being respectively provided with corresponding weight coefficient includes：
The weight coefficient of Current traffic data sample is set to n/Nmin；Wherein, numbers of the n for Current traffic data sample
Amount, NminSmallest sample amount is gathered for default traffic；
Quantity n when the traffic data sample of a upper time periodpIt is less than NminWhen, by the traffic data sample of a upper time period
This weight coefficient is set to np/Nmin, and the weight coefficient of history quantized data is set to (1-np/Nmin)；
Work as npIt is more than or equal to NminWhen, the weight coefficient of the traffic data sample of a upper time period is set to 1, and will
The weight coefficient of history quantized data is set to 0；
Work as npWhen being equal to 0, the weight coefficient of the traffic data sample of a upper time period was set to 0, and history is quantified
The weight coefficient of data is set to 1.
Preferably, calculating the estimated value of traffic state data by formula as described below
Wherein,For the weighted value of Current traffic data sample,For a upper time period traffic data sample plus
Weights,Weighted value for history quantized data；DescribedWithNegotiation speed weighting algorithm is obtained.
Present invention also offers a kind of estimating system of traffic state data, the system includes：Data acquisition module, judgement
Module, training data module, correcting module and Fusion Module；
Wherein, the data acquisition module, for obtaining in real time traffic data sample, and by acquired traffic data sample
Originally the judge module is sent to；
Whether the judge module, the quantity of the Current traffic data sample for judging to be gathered in current slot are big
Smallest sample amount is gathered in default traffic；If it is, by the Current traffic data sample gathered in the current slot
Originally the training data module is sent to；Otherwise, the Current traffic data sample gathered in the current slot is sent
Give the Fusion Module；
The training data module, for carrying out to the Current traffic data sample for being received according to default training pattern
Data are fitted, the data after be fitted and by fitting after data is activation to the correcting module；
The correcting module, repaiies for the data based on error ratio criterion after correction matrix is to fitting
Just, estimated value and according to revised data acquisition traffic state data is simultaneously exported；
The Fusion Module, for the Current traffic data sample by receiving, the traffic data sample of a upper time period
Corresponding weight coefficient is respectively provided with history quantized data；And according to the Current traffic data sample, upper time period
Traffic data sample, history quantized data and respective weight coefficient, are calculated the estimated value of traffic state data defeated
Go out.
Preferably, the data acquisition module also includes：Collecting unit and converting unit；
The collecting unit, for Real-time Collection original traffic data, and by the original traffic data is activation for collecting to
The converting unit；
The converting unit, for according to the original traffic data and road section information for being received, by map-matching algorithm
The gps data that positional information associates section is generated, and the gps data will be quantified, the speed probability after being quantified
Data, the speed probability data after the quantization is exported as traffic data sample.
As above visible, the method for estimation and system of traffic state data provided by the present invention, due in the present invention
In the method for estimation and system of traffic state data, when the quantity ratio of the Current traffic data sample gathered in current slot
Less, sample data be not enough to reflect traffic state data variation tendency when, can merge Current traffic data sample, upper one
The traffic data sample and history quantized data of time period, to be calculated the estimated value of traffic state data, is examined due to comprehensive
Consider real time data, upper time segment data and history quantized data, therefore can effectively improve the less situation of sample data volume
The accuracy of lower parameter estimation and reliability；And the quantity of working as gathered Current traffic data sample is relatively more, it is possible to deposit
In abnormal data, then data fitting is carried out to Current traffic data sample using default training pattern first, be then based on again
Data of the error ratio criterion after correction matrix is to fitting are modified, and according to revised data acquisition traffic shape
The estimated value of state parameter, the scope due to setting weight coefficient using correction matrix are wider, and data-measuring yardstick is little, more can be comprehensive
Impact of each data to result is assessed, such that it is able to reduce adverse effect of the abnormal data to estimated value, effectively to judge
The traffic information on road.And, method and system provided by the present invention, be not rely on temporal characteristics (for example, peak,
Flat peak, night etc.) and target road (for example, highway, through street, surface road etc.) coefficient setting, but based on big
Amount history quantized data and error ratio criterion are modified to data, therefore without the need for additionally arranging nuisance parameter.
Description of the drawings
Fig. 1 is the schematic flow sheet of the method for estimation of the traffic state data in the embodiment of the present invention；
Fig. 2 is the structural representation of the estimating system of the traffic state data in the embodiment of the present invention.
Specific embodiment
For making the objects, technical solutions and advantages of the present invention become more apparent, develop simultaneously embodiment referring to the drawings, right
The present invention is further described.
Present embodiments provide a kind of method of estimation of traffic state data.
Fig. 1 is the schematic flow sheet of the method for estimation of the traffic state data in the embodiment of the present invention.As shown in figure 1, this
The method of estimation of the traffic state data in inventive embodiments mainly includes step as described below：
Step 11, obtains traffic data sample in real time.
In the inventive solutions, above-mentioned steps 11 can be realized in several ways.Below will be with therein
As a example by a kind of implementation, technical scheme is introduced.
For example, in the preferred embodiment, the step 11 can include：
Step 110, Real-time Collection original traffic data.
Preferably, in a particular embodiment of the present invention, the original traffic data for being gathered can be included but is not limited to：Car
Numbering, taxi company, receive the data such as time, longitude, latitude, azimuth, instantaneous velocity and operation state.
Step 111, according to the original traffic data and road section information that are gathered, generates position letter by map-matching algorithm
The gps data in breath association section, and the gps data is quantified, the speed probability data after being quantified, by the amount
Speed probability data after change is used as traffic data sample.
Preferably, in a particular embodiment of the present invention, the quantization unit that the gps data is quantified can be
1.After quantifying to the gps data, you can the speed probability data after being quantified, so as to speed that can be after the quantization
Probability data is used as traffic data sample.
Whether step 12, judge the quantity of the Current traffic data sample gathered in current slot more than default friendship
Logical collection smallest sample amount；If it is, execution step 13；Otherwise, execution step 15；
In the inventive solutions, traffic collection smallest sample amount N can be pre-setmin, it is then possible to
Judge quantity n of the Current traffic data sample gathered in current slot whether more than Nmin.Wherein, the current time
The length of section can preset, and will not be described here.
If n>Nmin, the data volume of the gathered traffic data sample of expression is than larger, it is possible to can there is abnormal number
According to.In order to reduce error estimation, need to be modified the traffic data sample for being gathered, to reduce abnormal data to estimating knot
The adverse effect of fruit, so execution step 13；
If n≤Nmin, then it represents that the traffic data sample data for now being gathered cannot reflect that traffic behavior is joined exactly
The variation tendency of number (for example, speed), needed the traffic data sample of Current traffic data sample and a upper time period and went through
History quantized data carries out fusion treatment, to obtain the estimated value of traffic state data, therefore execution step 15.
In addition, in the preferred embodiment, the traffic gathers smallest sample amount NminSize can set in advance
Put.For example, traffic can be pre-set by method as described below and gathers smallest sample amount NminSize：
The sample value of the traffic state data collected from road, for example, the sample value (v of speed1,v2,…,vn)
Submit to N (μ, σ2) normal distribution, its basic ASSOCIATE STATISTICS amount isSample variance is
Can be obtained according to theory of probability theory：
WhereinFor degree of freedom it isT-distribution, a is significance level, therefore, according to above-mentioned formula：Current friendship
The quantity of logical data sampleTherefore, described can set：
Step 13, carries out data fitting according to default training pattern to Current traffic data sample, after being fitted
Data.
Preferably, in a particular embodiment of the present invention, before above-mentioned steps 13, can be with according to history quantized data
Pre-set corresponding training pattern.
Data fitting is carried out for example, it is possible to be based on substantial amounts of history and quantify gps data, instruction is formed using curve fitting algorithm
Practice data equation.
Specifically, according to law of great number, in fixed section and under the premise of the time, traffic state data (for example, speed)
Data (v1,p1,v2,p2,…,vn,pn) meet normal distribution model, wherein piFor speed viThe probability of generation.
Assume that training data equation isUsing curve fitting algorithm (for example, least square
Method), can try to achieve parameter μ, σ, wherein μ are that training data equation is expected, σ is training data equation variance, and its calculating process is as follows：
To equationLogarithmic transformation is made on both sides, obtains：
Order：
Then ln (p (v))=av2+bv+c
Thus, it can be known that the history quantized data (v according to traffic state data (for example, speed)1,p1,v2,p2,…,vn,
pn), and each ginseng of corresponding training data equation p (v) can be calculated using curve fitting algorithm (for example, method of least square)
Number a, b, c, then can obtain training data equation p (v), and pre-set corresponding training pattern according to the training data equation.
Therefore, in this step, data can be carried out to Current traffic data sample according to above-mentioned default training pattern
Fitting, so as to the data after being fitted, for example, it is possible to obtain a matched curve formed by the data after being fitted.
Step 14, is modified based on data of the error ratio criterion after correction matrix is to fitting, and according to repairing
The estimated value of the data acquisition traffic state data after just, terminates flow process.
In the inventive solutions, above-mentioned steps 14 can be realized in several ways.Below will be with therein
As a example by a kind of implementation, technical scheme is introduced.
For example, in the preferred embodiment, the step 14 can include：
Step 141, using the default training pattern, obtains under current road segment current time according to history quantized data
Traffic state data (for example, speed) probability history matched curve Ψ (v1,v2,......vM), wherein viFor traffic behavior
Parameter (for example, speed).
According to acquired traffic data sample, step 142, determines that the probability function of traffic state data is Z (v1,
v2......vn).
Step 143, the probability function according to probability history matched curve and traffic state data obtain the error ratio of sample
Matrix.
For example, in the preferred embodiment, error ratio matrix K can be obtained by formula as described below：
Step 144, obtains complementary matrix I according to the error ratio matrix.
For example, in the preferred embodiment, complementary matrix I can be obtained by formula as described below：
Step 145, according to the correction matrix Λ that complementary matrix I obtains each traffic data sample.
For example, in the preferred embodiment, the correction matrix Λ can be expressed as：Λ=[w1,w2,……
wn].
Wherein wiFor traffic state data (for example, speed) viModifying factor,
Step 146, according to the correction matrix Λ, obtains correction value Z' of revised each traffic data sample.
For example, in the preferred embodiment, correction value Z' of the traffic data sample can pass through following institute
State formula to obtain：
Step 147, according to the correction value of each traffic data sample, is calculated traffic shape by Weighted Average Algorithm
The estimated value of state parameter.
For example, in the preferred embodiment, the estimated value of the traffic state data can pass through as described below
Formula is obtained：
Step 15, was Current traffic data sample, the traffic data sample of a upper time period and history quantized data difference
Corresponding weight coefficient is set.
In the inventive solutions, above-mentioned steps 15 can be realized in several ways.Below will be with therein
As a example by a kind of implementation, technical scheme is introduced.
For example, in the preferred embodiment, the step 15 can include：
The weight coefficient of Current traffic data sample is set to n/Nmin；Wherein, numbers of the n for Current traffic data sample
Amount, NminSmallest sample amount is gathered for default traffic；
Quantity n when the traffic data sample of a upper time periodpIt is less than NminWhen, by the traffic data sample of a upper time period
This weight coefficient is set to np/Nmin, and the weight coefficient of history quantized data is set to (1-np/Nmin)；
Work as npIt is more than or equal to NminWhen, the weight coefficient of the traffic data sample of a upper time period is set to 1, and will
The weight coefficient of history quantized data is set to 0；
Work as npWhen being equal to 0, the weight coefficient of the traffic data sample of a upper time period was set to 0, and history is quantified
The weight coefficient of data is set to 1.
Step 16, according to Current traffic data sample, the traffic data sample of a upper time period, history quantized data and
Respective weight coefficient, is calculated the estimated value of traffic state data.
In the inventive solutions, above-mentioned steps 16 can be realized in several ways.Below will be with therein
As a example by a kind of implementation, technical scheme is introduced.
For example, in the preferred embodiment, traffic state data can be calculated by formula as described below
Estimated value
Wherein,For the weighted value of Current traffic data sample,For a upper time period traffic data sample plus
Weights,Weighted value for history quantized data.Preferably, describedWithCan Negotiation speed weighting algorithm
Obtain.
In addition, in the inventive solutions, the method for estimation of the traffic state data provided according to the present invention, this
The bright estimating system for additionally providing corresponding traffic state data.
Fig. 2 is the structural representation of the estimating system of the traffic state data in the embodiment of the present invention.As shown in Fig. 2 this
The estimating system of the traffic state data in inventive embodiments can include：The system includes：Data acquisition module 21, judge mould
Block 22, training data module 23, correcting module 24 and Fusion Module 25；
Wherein, the data acquisition module 21, for obtaining in real time traffic data sample, and by acquired traffic data
Sample is sent to the judge module 22；
The judge module 22, for the quantity of Current traffic data sample that judges to be gathered in current slot whether
Smallest sample amount is gathered more than default traffic；If it is, the Current traffic data that will be gathered in the current slot
Sample is sent to the training data module 23；Otherwise, the Current traffic data sample that will be gathered in the current slot
It is sent to the Fusion Module 25；
The training data module 23, for entering to the Current traffic data sample for being received according to default training pattern
Row data are fitted, the data after be fitted and by fitting after data is activation to the correcting module 24；
The correcting module 24, repaiies for the data based on error ratio criterion after correction matrix is to fitting
Just, estimated value and according to revised data acquisition traffic state data is simultaneously exported；
The Fusion Module 25, for the Current traffic data sample by receiving, the traffic data sample of a upper time period
This and history quantized data are respectively provided with corresponding weight coefficient；And according to the Current traffic data sample, a upper time period
Traffic data sample, history quantized data and respective weight coefficient, be calculated the estimated value of traffic state data simultaneously
Output.
Preferably, in a particular embodiment of the present invention, the data acquisition module 21 can further include：Collection
Unit and converting unit；
The collecting unit, for Real-time Collection original traffic data, and by the original traffic data is activation for collecting to
The converting unit；
The converting unit, for according to the original traffic data and road section information for being received, by map-matching algorithm
The gps data that positional information associates section is generated, and the gps data is quantified, the speed probability number after being quantified
According to using the speed probability data after the quantization as the output of traffic data sample.
In summary, in due to the method for estimation and system of traffic state data in the present invention, work as current slot
The quantity of interior gathered Current traffic data sample is fewer, and sample data is not enough to reflect that the change of traffic state data becomes
During gesture, Current traffic data sample, the traffic data sample of a upper time period and history quantized data can be merged, to calculate
To the estimated value of traffic state data, due to having considered real time data, upper time segment data and history quantized data, because
This can effectively improve sample data volume less in the case of parameter estimation accuracy and reliability；And work as the current friendship for being gathered
The quantity of logical data sample is relatively more, it is possible to when there is abnormal data, then first using default training pattern to Current traffic
Data sample carries out data fitting, and then the data again based on error ratio criterion after correction matrix is to fitting are repaiied
Just, the estimated value and according to revised data acquisition traffic state data, due to setting weight coefficient using correction matrix
Scope is wider, and data-measuring yardstick is little, more can impact of each data of comprehensive assessment to result, such that it is able to reduce abnormal data pair
The adverse effect of estimated value, effectively to judge the traffic information of road.And, method provided by the present invention and it is
System, be not rely on temporal characteristics (for example, peak, Ping Feng, night etc.) and target road (for example, highway, through street,
Surface road etc.) coefficient setting, but data are modified based on a large amount of history quantized datas and error ratio criterion, need not
Extra setting nuisance parameter.
Presently preferred embodiments of the present invention is the foregoing is only, not in order to limit the present invention, all in essence of the invention
Within god and principle, any modification, equivalent substitution and improvements that is done etc. are should be included within the scope of protection of the invention.
Claims (13)
1. a kind of method of estimation of traffic state data, it is characterised in that the method includes：
Traffic data sample is obtained in real time；
When the quantity of the Current traffic data sample gathered in current slot gathers smallest sample amount more than default traffic
When, data fitting is carried out to Current traffic data sample according to default training pattern, the data after being fitted；
Traffic data sample is modified by correction matrix based on error ratio criterion, and is obtained according to revised data
The estimated value of traffic state data is taken, including：Using the default training pattern, current road is obtained according to history quantized data
The probability history matched curve of the traffic state data under section current time；Traffic is determined according to acquired traffic data sample
The probability function of state parameter；Probability function according to probability history matched curve and traffic state data obtains the error of sample
Compare matrix K；Complementary matrix I is obtained according to the error ratio matrix K；Repairing for each traffic data sample is obtained according to complementary matrix I
Positive matrices Λ；According to the correction matrix Λ, obtain revised each traffic data sample correction value Ζ '；According to described each
The correction value of traffic data sample, the estimated value that traffic state data is calculated by Weighted Average Algorithm；Work as current time
When the quantity of the Current traffic data sample gathered in section gathers smallest sample amount less than or equal to default traffic, it is current
Traffic data sample, the traffic data sample of a upper time period and history quantized data are respectively provided with corresponding weight coefficient；And
According to Current traffic data sample, the traffic data sample of a upper time period, history quantized data and respective weight coefficient,
It is calculated the estimated value of traffic state data.
2. method according to claim 1, it is characterised in that：
The traffic state data is speed.
3. method according to claim 1, it is characterised in that the real-time acquisition traffic data sample includes：
Real-time Collection original traffic data；
According to the original traffic data and road section information that are gathered, positional information is generated by map-matching algorithm and associates section
Gps data, and the gps data is quantified, the speed probability data after being quantified will be general for the speed after the quantization
Rate data are used as traffic data sample.
4. method according to claim 3, it is characterised in that the original traffic data for being gathered include：
Car number, taxi company, reception time, longitude, latitude, azimuth, instantaneous velocity and operation state.
5. method according to claim 1, it is characterised in that according to default training pattern to Current traffic data sample
Originally, before carrying out data fitting, the method is still further comprised：
According to the history quantized data of traffic state data, and corresponding training data equation is calculated using curve fitting algorithm
Parameters, obtain training data equation, and corresponding training pattern pre-set according to the training data equation.
6. method according to claim 5, it is characterised in that error ratio matrix K is obtained by formula as described below：
Wherein, K is expressed as [ε1,ε2,…,εi,…εn]；Ψ is probability history matched curve；vi, i=1,2,3 ... ..., n are friendship
Logical state parameter；Probability functions of the Z for traffic state data.
7. method according to claim 6, it is characterised in that complementary matrix I is obtained by formula as described below：
8. method according to claim 7, it is characterised in that the correction matrix Λ is expressed as：
Λ=[w1,w2,……wn]；
Wherein, wiFor traffic state data viModifying factor, and
9. method according to claim 8, it is characterised in that traffic data sample is obtained by formula as described below
Correction value Ζ '：
10. method according to claim 1, it is characterised in that described for Current traffic data sample, a upper time period
Traffic data sample and history quantized data are respectively provided with corresponding weight coefficient to be included：
The weight coefficient of Current traffic data sample is set to n/Nmin；Wherein, quantity of the n for Current traffic data sample,
NminSmallest sample amount is gathered for default traffic；
Quantity n when the traffic data sample of a upper time periodpIt is less than NminWhen, by the power of the traffic data sample of a upper time period
Weight coefficient is set to np/Nmin, and the weight coefficient of history quantized data is set to (1-np/Nmin)；
Work as npIt is more than or equal to NminWhen, the weight coefficient of the traffic data sample of a upper time period is set to 1, and by history amount
The weight coefficient for changing data is set to 0；
Work as npWhen being equal to 0, the weight coefficient of the traffic data sample of a upper time period was set to 0, and by history quantized data
Weight coefficient is set to 1.
11. methods according to claim 10, it is characterised in that calculate traffic behavior ginseng by formula as described below
Several estimated values
Wherein,For the weighted value of Current traffic data sample,For the weighted value of the traffic data sample of a upper time period,Weighted value for history quantized data；DescribedWithNegotiation speed weighting algorithm is obtained.
12. a kind of estimating systems of traffic state data, it is characterised in that the system includes：Data acquisition module, judge mould
Block, training data module, correcting module and Fusion Module；
Wherein, the data acquisition module, for obtaining traffic data sample in real time, and acquired traffic data sample is sent out
Give the judge module；
Whether the judge module, the quantity of the Current traffic data sample for judging to be gathered in current slot are more than in advance
If traffic collection smallest sample amount；If it is, the Current traffic data sample gathered in the current slot is sent out
Give the training data module；Otherwise, the Current traffic data sample gathered in the current slot is sent to institute
State Fusion Module；
The training data module, for carrying out data according to default training pattern to the Current traffic data sample for being received
Fitting, the data after be fitted and by fitting after data is activation to the correcting module；
The correcting module, for being modified to traffic data sample by correction matrix based on error ratio criterion, and
According to the estimated value of revised data acquisition traffic state data and export, including：Using the default training pattern, root
According to the probability history matched curve that history quantized data obtains the traffic state data under current road segment current time；According to being obtained
The traffic data sample for taking determines the probability function of traffic state data；According to probability history matched curve and traffic state data
Probability function obtain the error ratio matrix K of sample；Complementary matrix I is obtained according to the error ratio matrix K；According to complementary matrix
I obtains the correction matrix Λ of each traffic data sample；According to the correction matrix Λ, revised each traffic data sample is obtained
Correction value Ζ '；According to the correction value of each traffic data sample, traffic behavior ginseng is calculated by Weighted Average Algorithm
Several estimated values；
The Fusion Module, for the Current traffic data sample by receiving, the traffic data sample of a upper time period and went through
History quantized data is respectively provided with corresponding weight coefficient；And according to the Current traffic data sample, the traffic of a upper time period
Data sample, history quantized data and respective weight coefficient, are calculated the estimated value of traffic state data and export.
13. systems according to claim 12, it is characterised in that the data acquisition module also includes：Collecting unit and
Converting unit；
The collecting unit, for Real-time Collection original traffic data, and by the original traffic data is activation for collecting to described
Converting unit；
The converting unit, for according to the original traffic data and road section information for being received, being generated by map-matching algorithm
Positional information associates the gps data in section, and the gps data is quantified, the speed probability data after being quantified, will
Speed probability data after the quantization is exported as traffic data sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410617762.6A CN104408915B (en) | 2014-11-05 | 2014-11-05 | A kind of method of estimation of traffic state data and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410617762.6A CN104408915B (en) | 2014-11-05 | 2014-11-05 | A kind of method of estimation of traffic state data and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104408915A CN104408915A (en) | 2015-03-11 |
| CN104408915B true CN104408915B (en) | 2017-03-15 |
Family
ID=52646541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410617762.6A Active CN104408915B (en) | 2014-11-05 | 2014-11-05 | A kind of method of estimation of traffic state data and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104408915B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104809877B (en) * | 2015-05-14 | 2017-09-08 | 重庆大学 | The highway place traffic state estimation method of feature based parameter weighting GEFCM algorithms |
| CN105139656B (en) * | 2015-09-28 | 2017-09-29 | 百度在线网络技术（北京）有限公司 | A kind of road condition Forecasting Methodology and device |
| CN107798864A (en) * | 2016-09-06 | 2018-03-13 | 高德信息技术有限公司 | A kind of computational methods and device of road speed |
| CN106526339A (en) * | 2016-11-07 | 2017-03-22 | 中国电力科学研究院 | Method and system for determining smallest sampling sample capacity on basis of radio interference value sample data |
| JP6874540B2 (en) | 2017-06-02 | 2021-05-19 | トヨタ自動車株式会社 | Vehicle control method, information processing device, and vehicle control system |
| CN109270927B (en) * | 2017-07-17 | 2022-03-11 | 阿里巴巴（中国）有限公司 | Road data generation method and device |
| CN110108486B (en) * | 2018-01-31 | 2022-07-05 | 阿里巴巴集团控股有限公司 | Bearing fault prediction method, device and system |
| CN111723831B (en) * | 2019-03-20 | 2023-11-14 | 北京嘀嘀无限科技发展有限公司 | Data fusion method and device |
| CN111754772A (en) * | 2020-06-28 | 2020-10-09 | 蘑菇车联信息科技有限公司 | Traffic road condition determining method and electronic equipment |
| CN114580792B (en) * | 2022-04-28 | 2022-08-30 | 深圳丰尚智慧农牧科技有限公司 | Method and device for identifying working state of bulking machine, computer equipment and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7610145B2 (en) * | 2003-07-25 | 2009-10-27 | Triangle Software Llc | System and method for determining recommended departure time |
| GB2423154A (en) * | 2005-02-02 | 2006-08-16 | E Lead Electronic Co Ltd | Method for detecting vehicle speed |
| CN101533094B (en) * | 2009-04-23 | 2011-05-04 | 民航数据通信有限责任公司 | Calculating method and system for estimating position error based on TIS-B |
| CN102610092A (en) * | 2012-03-23 | 2012-07-25 | 天津大学 | Urban road speed predication method based on RBF (radial basis function) neural network |
| CN103794061B (en) * | 2014-03-10 | 2016-06-22 | 上海云砥信息科技有限公司 | The method that road merges travel speed is calculated based on multiple location data |
-
2014
- 2014-11-05 CN CN201410617762.6A patent/CN104408915B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104408915A (en) | 2015-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104408915B (en) | A kind of method of estimation of traffic state data and system | |
| WO2018122803A1 (en) | Smart road traffic anomaly detection method | |
| CN106205114B (en) | A kind of Freeway Conditions information real time acquiring method based on data fusion | |
| Debnath et al. | Navigational traffic conflict technique: a proactive approach to quantitative measurement of collision risks in port waters | |
| CN104751642B (en) | A kind of advanced road real-time predictor method of traffic flow operation risk | |
| CN104134349A (en) | Bus road condition processing system and method based on traffic multi-source data fusion | |
| CN112669594B (en) | Method, device, equipment and storage medium for predicting traffic road conditions | |
| CN111415533B (en) | Bend safety early warning monitoring method, device and system | |
| CN110176139A (en) | A kind of congestion in road identification method for visualizing based on DBSCAN+ | |
| CN107331191A (en) | Abnormal driving vehicle localization method, Cloud Server and system | |
| CN102087788A (en) | Method for estimating traffic state parameter based on confidence of speed of float car | |
| CN113160593A (en) | Mountain road driving safety early warning method based on edge cloud cooperation | |
| Stipancic et al. | Impact of congestion and traffic flow on crash frequency and severity: application of smartphone-collected GPS travel data | |
| CN104282151A (en) | Real-time floating vehicle path matching method based on high-frequency satellite positioning data | |
| CN106408944A (en) | Congestion level analysis platform based on double communication data | |
| CN104794895A (en) | Multisource traffic information fusion method for expressways | |
| CN107507417A (en) | A kind of smartway partitioning method and device based on microwave radar echo-signal | |
| CN104900057A (en) | City expressway main and auxiliary road floating vehicle map matching method | |
| CN104200082A (en) | Typhoon landing prediction method | |
| CN108460685A (en) | Method and apparatus for excavating the correlation rule between vehicle insurance risks and assumptions | |
| CN116017488B (en) | Boundary search-based large-area scattering communication coverage prediction method and system | |
| Yuan et al. | A roadway safety sustainable approach: modeling for real-time traffic crash with limited data and its reliability verification | |
| CN110310461A (en) | A kind of icy road real-time early warning system based on big data | |
| CN104268415A (en) | Travel congestion probability estimation method | |
| US20070233359A1 (en) | System and method for aggregating probe vehicle data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |