CN106469503B - A kind of method and apparatus for predicting traffic events coverage - Google Patents
A kind of method and apparatus for predicting traffic events coverage Download PDFInfo
- Publication number
- CN106469503B CN106469503B CN201510502663.8A CN201510502663A CN106469503B CN 106469503 B CN106469503 B CN 106469503B CN 201510502663 A CN201510502663 A CN 201510502663A CN 106469503 B CN106469503 B CN 106469503B
- Authority
- CN
- China
- Prior art keywords
- road chain
- speed
- vector
- road
- chain
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention provides a kind of method and apparatus for predicting traffic events coverage, there is no judged using the unique characteristics of traffic events as parameter this prediction technique, but utilize the pairs of existing matching characteristic vector sum predicted characteristics vector having built up in advance, the speed and chain speed in road associated there composition current signature vector of the road chain of traffic events will occur, similarity mode is carried out with matching characteristic vector, the corresponding predicted characteristics vector of the matching characteristic vector matched is the speed of each road chain in next period of prediction.It can predict that traffic events occur on outlet chain m, and the speed of other road chains to extend influence.Unique characteristics of this method of the application independent of traffic events, therefore, prediction result is more accurate.
Description
Technical field
The present invention relates to field of intelligent transportation technology, in particular to a kind of method and dress for predicting traffic events coverage
It sets.
Background technique
In intelligent transportation system (ITS, Intelligent Transportation System), it is generally based on floating
The coverage of motor-car (FCD, Float Car Data) data analysis traffic events.The coverage of traffic events can be friendship
The manager of interpreter's part provides decision-making foundation, or traveler provides accurate induction information, to reduce the travel time.Together
When, it is also used as the input of navigation engine Route Guidance System, car assisted driver preferably judges.Therefore, it predicts
Traffic events coverage it is significant.
Currently, the prediction to traffic events coverage, based on the data of fixed detector detection, such as traffic data
Or traffic density data etc., using in traffic flow theory queueing theory and traffic shock wave theory calculate coverage.But it is fixed
There are errors for the data such as the volume of traffic of detector detection and traffic density, and the result predicted in this way using the data of error is inaccurate
Really.
The prediction technique that the traffic events based on classification, linear regression and neural network influence in the prior art is to hand over
The unique characteristics of interpreter's part are as input quantity, for example, traffic events are to knock into the back, then knocking into the back is the unique characteristics of the traffic events.
But there is no directly connections for the coverage of the unique characteristics of traffic events and traffic events, so as to cause traffic thing is based on
There is very big gap with actual conditions as the coverage that criterion is predicted in the unique characteristics of part.
Therefore, those skilled in the art need to provide a kind of method and apparatus for predicting traffic events coverage, disobey
By the unique characteristics of traffic events, it is capable of the influence of Accurate Prediction traffic events.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of method and apparatus for predicting traffic events coverage, disobey
By the unique characteristics of traffic events, it is capable of the influence of Accurate Prediction traffic events.
The embodiment of the present invention provides a kind of method for predicting traffic events coverage, comprising the following steps:
Determine road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n that traffic events occur；The phase
Associated road chain m-1 ... ..., road chain m-n are that the road chains that involve of coverage of traffic events occurs for the chain m that shows the way；The n is pre-
If value；
The road chain m and road chain m-1 ... ... associated with road chain m when traffic events occur is obtained, road chain m-n's
Floating car data；
According to the speed of floating car data acquisition road chain m and road chain m-1 ... ... associated with road chain m, road chain
The speed of m-n generates current signature vector by the speed of all of above road chain；
It is searched and the matched matching characteristic vector of the current signature vector similarity from historical pattern library；
According to the speed in the corresponding predicted characteristics vector of the matching characteristic vector found, obtain road chain m and with road chain m
The speed of associated road chain m-1 ... ..., road chain m-n prediction saves existing matching characteristic in pairs in the historical pattern library
Vector sum predicted characteristics vector；The speed of road chain in the matching characteristic vector sum predicted characteristics vector is in advance according to non-friendship
Interpreter's part causes the speed of road chain congestion Shi Lulian to obtain；Wherein matching characteristic vector be period t when road chain m and with road chain m
Associated road chain m-1 ... ..., the speed of road chain m-n, prediction characteristic vector be period t+1 when road chain m and with road chain m phase
Associated road chain m-1 ... ..., road chain m-n speed.
Preferably, the road chain speed in the matching characteristic vector sum predicted characteristics vector in advance according to non-traffic events when
The speed of congestion road chain obtains, specifically:
When road chain is respectively less than in pre-set velocity threshold value and the road chain in the speed of period t and period t+1 without traffic events
When generation, determine that the road chain is congestion road chain caused by non-traffic events, using the speed of congestion road chain period t as described
Speed with the road eigen vector Zhong Gai chain, using the speed of the period t+1 of the congestion road chain as with the matching characteristic vector at
To the speed of the existing road predicted characteristics vector Zhong Gai chain.
Preferably, it is described from historical pattern library search with the matched matching characteristic of current signature vector similarity to
Amount, the speed in the corresponding predicted characteristics vector of the matching characteristic vector found be the next period predicted road chain m and with
The associated road chain m-1 ... ... of road chain m, the speed of road chain m-n, specifically:
By linear weighted function calculate in the historical pattern library with the current signature vector Euclidean distance nearest k
With feature vector；K is the integer greater than 1；The k matching characteristic vector corresponds to k predicted characteristics vector；It is pre- by the k
It surveys feature vector and obtains k speed of each road chain, the speed as road chain prediction is averaged to k speed of each road chain
Degree.
Preferably, from historical pattern library search with the matched matching characteristic of current signature vector similarity to
Amount, before further include:
Matching characteristic vector in the historical pattern library is normalized；
The lookup from historical pattern library and the matched matching characteristic vector of the current signature vector similarity, specifically
Are as follows:
It is calculated in the historical pattern library by linear weighted function and is returned with the current signature vector Euclidean distance nearest k
Matching characteristic vector after one change.
The embodiment of the invention provides a kind of devices that prediction traffic events influence, comprising: determining module, data obtain mould
Block, current signature vector generation module, matching module and the first prediction module；
The determining module, the road chain m and road chain m- associated with road chain m occurred for determining traffic events
1 ... ..., road chain m-n；The associated road chain m-1 ... ..., road chain m-n be show the way chain m occur traffic events influence model
Enclose the road chain involved；The n is preset value；
The data obtaining module, for obtaining the road chain m and road associated with road chain m when traffic events occur
Chain m-1 ... ..., the floating car data of road chain m-n；
The current signature vector generation module, for according to the floating car data obtain road chain m speed and with
The associated road chain m-1 ... ... of road chain m, the speed of road chain m-n generate current signature vector by the speed of all of above road chain；
The matching module is searched and the matched matching characteristic of current signature vector similarity from historical pattern library
Vector；
First prediction module, for according to the speed in the corresponding predicted characteristics vector of matching characteristic vector found
Degree obtains the speed of road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n prediction, the historical pattern library
It is middle to save existing matching characteristic vector sum predicted characteristics vector in pairs；In the matching characteristic vector sum predicted characteristics vector
The speed of road chain is to cause the speed of road chain congestion Shi Lulian to obtain according to non-traffic events in advance；Wherein matching characteristic vector is
Road chain m and road chain m-1 ... ... associated with road chain m when period t, the speed of road chain m-n, prediction characteristic vector are period t
No. 1 chain m and road chain m-1 ... ... associated with road chain m, road chain m-n speed when+.
Preferably, further includes: unit is established in historical pattern library；
Unit is established in the historical pattern library, for being respectively less than pre-set velocity when speed of the road chain in period t and period t+1
When occurring in threshold value and the road chain without traffic events, determines that the road chain is congestion road chain caused by non-traffic events, this is gathered around
Speed of the speed of stifled road chain period t as the road matching properties vector Zhong Gai chain, by the speed of the period t+1 of the congestion road chain
It spends as the speed with the matching characteristic vector existing road predicted characteristics vector Zhong Gai chain in pairs.
Preferably, first prediction module includes: computational submodule and prediction submodule；
Computational submodule, it is European with the current signature vector in the historical pattern library for being calculated by linear weighted function
Apart from k nearest matching characteristic vector；K is the integer greater than 1；The k matching characteristic vector correspond to k predicted characteristics to
Amount；；
The prediction submodule, for obtaining k speed of each road chain by the k predicted characteristics vector, to each
K speed of road chain is averaged the speed as road chain prediction.
Preferably, further includes: normalization submodule；
The normalization submodule, for matching with the current signature vector similarity being searched from historical pattern library
Matching characteristic vector, the matching characteristic vector in the historical pattern library is normalized；
The computational submodule, for by linear weighted function calculate in the historical pattern library with the current signature vector
Matching characteristic vector after k nearest normalization of Euclidean distance；K is the integer greater than 1；The k matching characteristic vector pair
Answer k predicted characteristics vector.
Compared with prior art, the invention has the following advantages that
This prediction technique is not judged using the unique characteristics of traffic events as parameter, but is utilized in advance
The history feature vector of foundation, speed and chain speed in the road associated there composition that the road chain of traffic events will occur are current
Matching characteristic vector in feature vector, with history feature vector carries out similarity mode, the matching characteristic vector pair matched
The predicted characteristics vector answered is the speed of each road chain in next period of prediction.It can predict to hand on outlet chain m
Interpreter's part, and the speed of other road chains to extend influence.Unique characteristics of this method of the application independent of traffic events,
Therefore, prediction result is more accurate.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is one flow chart of embodiment of the method that prediction traffic events provided by the invention influence；
Fig. 2 is two flow chart of embodiment of the method that prediction traffic events provided by the invention influence；
Fig. 3 is three flow chart of embodiment of the method that prediction traffic events provided by the invention influence；
Fig. 4 is one schematic diagram of Installation practice that prediction traffic events provided by the invention influence；
Fig. 5 is two schematic diagram of Installation practice that prediction traffic events provided by the invention influence；
Fig. 6 is three schematic diagram of Installation practice that prediction traffic events provided by the invention influence.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing to the present invention
Specific embodiment be described in detail.
Embodiment of the method one:
Referring to Fig. 1, which is one flow chart of embodiment of the method that prediction traffic events provided by the invention influence.
S101: road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n that traffic events occur are determined；
The associated road chain m-1 ... ..., road chain m-n are that the road chains that involve of coverage of traffic events occurs for the chain m that shows the way；Institute
Stating n is preset value；
It is understood that traffic events occur on the chain m of road, other associated road chains can be involved, therefore, it is necessary to judge
For the coverage of the neighbouring road road Lian Shou chain m for example, every road chain length is 100 meters, congestion caused by traffic events is usually big
In 1000 meters or more, then n at least takes 10.The value of n depends on the coverage that traffic events need to predict, range is longer, then n
It is bigger.When generally default n, the maximum magnitude that congestion caused by traffic events involves is considered, such as do not exceed 3000 meters, it is assumed that
Every road chain length is 100 meters, then presetting n is 30.
S102: the road chain m and road chain m-1 ... ... associated with road chain m when traffic events occur, road chain are obtained
The floating car data of m-n；
It should be noted that floating car data includes longitude and latitude and speed.
S103: according to the speed of floating car data acquisition road chain m and road chain m- associated with road chain m
1 ... ..., the speed of road chain m-n generates current signature vector by the speed of all of above road chain；
Road chain m, and n road chain associated with road chain m have n+1 road chain altogether, are corresponding with n+1 speed；This n+1
Speed forms a vector, i.e. current signature vector.
For example, the speed v of current period road chain m can be indicated are as follows: v (tnow,m)；The speed of current period road chain m-n can
To indicate are as follows: v (tnow,m-n)；Wherein, v indicates speed, tnowIndicate that current period, m indicate road chain m.
Then current signature vector s (t when road chain m generation traffic eventsnow, m) and it indicates are as follows:
s(tnow, m) and={ v (tnow,m),v(tnow,m-1),v(tnow,m-2),......v(tnow,m-n)} (1)
Current signature vector is a vector of speed it can be seen from formula (1), that is, includes the speed on n+1 road chain
Degree.
S104: it is searched and the matched matching characteristic vector of the current signature vector similarity from historical pattern library；
S105: according to the speed in the corresponding predicted characteristics vector of the matching characteristic vector found, obtain road chain m and with
The speed of road chain m associated road chain m-1 ... ..., road chain m-n prediction saves existing in pairs in the historical pattern library
With feature vector and predicted characteristics vector；The speed of road chain in the matching characteristic vector sum predicted characteristics vector be in advance according to
The speed of road chain congestion Shi Lulian is caused to obtain according to non-traffic events；Wherein matching characteristic vector be period t when road chain m and with
The associated road chain m-1 ... ... of road chain m, the speed of road chain m-n, road chain m and and road when prediction characteristic vector is period t+1
The associated road chain m-1 ... ... of chain m, road chain m-n speed.
It is understood that including the speed of multiple road chains in predicted characteristics vector.
It should be noted that the speed of the road chain in matching characteristic vector sum predicted characteristics vector in the historical pattern library
Degree is to cause the speed of road chain congestion Shi Lulian to obtain according to non-traffic events in advance；, establish the data of historical pattern library foundation
The speed of congestion road chain, the i.e. congestion of road chain are not as caused by traffic events when being non-traffic events.For example, can use
The speed of congestion road chain establishes historical pattern library when non-traffic events in one month.Because of congestion caused by traffic events
It influences to be really the problem of congestion is spread, congestion diffusion essence caused by conventional congestion (non-traffic events) diffusion and traffic events
On be because traffic bottlenecks caused by.Therefore, it can use conventional congestion data and establish historical pattern library.
Wherein, existing vector in pairs is saved in history feature library to be indicated with h (t, m)={ s (t, m), s (t+1, m) }.
Wherein, s (t, m) is matching characteristic vector, and s (t+1, m) is predicted characteristics vector.
By current signature vector s (tnow, m) and it is matched with matching characteristic vector s (t, m), if similarity mode,
The corresponding predicted characteristics vector s (t+1, m) of the matching characteristic vector is just the speed composition of all of above road chain in next period
Vector.For example, a cycle is 5 minutes.
That is, s (t, m) is used for and s (tnow, m) and it is matched, s (t+1, m) is corresponding predicted characteristics vector, s (t+1, m)
In each road chain speed be exactly each road chain of next period predicted speed.
Prediction technique provided in this embodiment, not judged using the unique characteristics of traffic events as parameter, and
It is that the road chain of traffic events will occur using the pairs of existing matching characteristic vector sum predicted characteristics vector having built up in advance
Speed and chain speed in road associated there composition current signature vector, with history feature vector in matching characteristic vector
The matching of similarity is carried out, the corresponding predicted characteristics vector of the matching characteristic vector matched is each of next period of prediction
The speed of a road chain.It can predict that traffic events occur on outlet chain m, and the speed of other road chains to extend influence.This
Apply for unique characteristics of this method independent of traffic events, therefore, prediction result is more accurate.
Embodiment of the method two:
Referring to fig. 2, which is two flow chart of embodiment of the method that prediction traffic events provided by the invention influence.
In the present embodiment, introduces how to establish historical pattern library first.
Road chain speed in the matching characteristic vector sum predicted characteristics vector in advance according to non-traffic events when congestion road
The speed of chain obtains, specifically:
When road chain is respectively less than in pre-set velocity threshold value and the road chain in the speed of period t and period t+1 without traffic events
When generation, determine that the road chain is congestion road chain caused by non-traffic events, using the speed of congestion road chain period t as described
Speed with the road eigen vector Zhong Gai chain, using the speed of the period t+1 of the congestion road chain as with the matching characteristic vector at
To the speed of the existing road predicted characteristics vector Zhong Gai chain.
For example, it is necessary to meet following condition for the speed sample in historical pattern library, i.e.,And
WhereinFor pre-set velocity threshold value, then it represents that the road chain m in next period of period t and t is congestion road chain.In historical pattern library
Speed sample standard deviation is the speed of congestion road chain.
For example, pre-set velocity threshold value is 20km/h.The speed of current period is lower than 20km/h, and the speed in next period
When also below 20km/h, then illustrate that the road chain gets congestion.
It should be noted that the method provided in embodiment of the method one is introduced with predicting the road chain speed in next period
, it is to be understood that after general generation traffic events, it will not only predict the road chain speed of next cycle, it generally can be pre-
The road chain speed in subsequent multiple periods is surveyed, for example, may predict that generation traffic events half are small by taking 5 minutes a cycles as an example
1 hour of traffic events later road chain speed occurs for the road chain speed of Shi Yihou, or prediction.
The method that the road chain speed of prediction phase week after next and phase week after next subsequent cycle is described below, needs to illustrate
It is that the prediction technique of the road chain speed of the subsequent cycle of phase week after next is identical with the prediction technique of phase week after next.Here, only with pre-
It is illustrated for the method for the road chain speed of survey phase week after next.
It further include S206 in the present embodiment: using the predicted characteristics vector of current period as the current signature vector, weight
It is new to carry out similarity mode, predict the speed of all of above road chain of subsequent cycle.
Wherein, S201-S205 is identical as S101-S105 respectively, and details are not described herein.
It is understood that in addition to predicting that next period is to establish current spy with the speed that the data of current Floating Car obtain
It levies other than vector, predicts the speed of phase week after next, and predict that the speed of the subsequent cycle of phase week after next is the previous period
The road chain speed of prediction re-starts highest similarity matching as current signature vector.
For example, the prediction technique provided through this embodiment, can issue traffic below after traffic events generation
Tutorial message: western 400 meter of four vehicle of North 4th Ring Road institute bridge knocks into the back, and current vehicle is lined up 500 meters, it is contemplated that vehicle queue after 10 minutes
3000 meters are up to, complete bridge to the estimated transit time of institute's bridge is 30 minutes, it is proposed that driver avoids this road.
Method provided in this embodiment can be predicted when traffic events occur, the speed that related road chain is got on the car, in turn
The jam situation that can predict automobile in subsequent predetermined time section, using the jam situation as Traffic information demonstration to driver, department
Machine can avoid the road chain of these congestions.Method provided in this embodiment does not contact directly with traffic events itself, therefore, fits
For all kinds of traffic events, applicability is more extensive, and prediction result is more quasi- than the method dependent on traffic events unique characteristics
Really.
Due to inherently a kind of anomalous event of traffic events, the generation of anomalous event is a small amount of, the prior art certainly
In based on traffic events unique characteristics prediction traffic events influence, need many traffic events as training sample, so
This method lack of training samples.In conventional method, a kind of mode is to use all traffic events as the training of certain stretch
Sample, but the complexity of urban road reduces the universality of this method, in turn results in prediction result inaccuracy.
Traffic events usually will cause congestion (having little significance to the traffic events predicted impact for not causing congestion), this hair
The bright middle influence by traffic events to traffic is interpreted as congestion diffusion (or low speed diffusion) problem, builds by history congestion information
Vertical historical pattern library, what which was reflected is exactly the diffusion of congestion at any time.Since history congestion is a kind of routine
Event, urban road are all getting congestion daily, therefore the sample data of this method provided by the invention is sufficient.
Embodiment of the method three:
Referring to Fig. 3, which is three flow chart of embodiment of the method that prediction traffic events provided by the invention influence.
S301-S303 is identical as S201-S203 in the present embodiment, and details are not described herein；S306 is identical as S206, herein not
It repeats again.
It should be noted that size of the similarity to measure inter-individual difference, including Euclidean distance, Minkowski away from
From, manhatton distance, vector space cosine similarity, Pearson correlation coefficients, Jaccard similarity factor.In engineer application,
Euclidean distance is most popular phase knowledge and magnanimity measure.
It should be noted that when carrying out similarity calculation, in order to avoid fast road chain occupies biggish calculating ratio
Weight, the low road chain of speed occupy lesser specific gravity, can be special to the matching in all history feature vectors in historical pattern library
Sign vector is standardized, and is then weighted Euclidean distance similarity calculation again.
For example, standardization each matching characteristic vector can be normalized, that is, it is transformed into 0 to 1 section.Than
Such as, it is assumed that the speed that chain m in road occurs for traffic events is 5km/h, and road chain m-5 speed is 70kn/h, a certain spy in historical pattern library
The speed for levying vector m is 10km/h, and the speed of m-5 is 50km/h.Matching of the matching compared to road chain m of road chain m-5 is in phase at this time
When calculating like degree, bigger specific gravity is occupied, such situation will influence the accuracy of final prediction result.In standardization
Afterwards, all in the range of 0-1, this phenomenon will not exist all values, so as to improve the standard of similarity calculation
Exactness, and then effectively improve the accuracy of prediction result.
In the lookup from historical pattern library and the matched matching characteristic vector of the current signature vector similarity, before also
Include:
Matching characteristic vector in the historical pattern library is normalized；
S304: it is described from historical pattern library search with the matched matching characteristic of current signature vector similarity to
Amount, specifically:
It is calculated in the historical pattern library by linear weighted function and is returned with the current signature vector Euclidean distance nearest k
Matching characteristic vector after one change；K is the integer greater than 1；The k matching characteristic vector corresponds to k predicted characteristics vector；
Similarity mode is matched so that Euclidean distance is nearest as an example in the present embodiment.Calculates current signature vector and go through
The Euclidean distance between matching characteristic vector in logotype library obtains the k matching spies nearest with current signature vector distance
Levy vector.
It is for instance possible to obtain the 3 history feature vectors nearest with current signature vector, i.e. k is 3.
Method provided in this embodiment, when carrying out similarity mode, to the corresponding matching characteristic of history feature vector to
Amount is standardized, and is then weighted Euclidean distance to the matching characteristic vector after standardization and is calculated, obtains final prediction
As a result.This avoid in similarity mode, the biggish road chain of speed occupies biggish specific gravity, is conducive to improve prediction in this way
As a result accuracy.
S305: the speed in the corresponding predicted characteristics vector of the matching characteristic vector found is the road in next period of prediction
Chain m and road chain m-1 ... ... associated with road chain m, the speed of road chain m-n, specifically:
K speed of each road chain is obtained by the k predicted characteristics vector, k speed of each road chain is averaged
It is worth the speed predicted as the road chain.
It is understood that it is special to be based on k prediction in order to improve the prediction accuracy of traffic events influence, in the present embodiment
Vector is levied to predict the road chain speed in next period, i.e., not merely with a predicted characteristics vector, but utilizes multiple predictions special
Road chain speed of the result that sign vector is averaged as next period of prediction.
Based on the method that a kind of prediction traffic events that above embodiments provide influence, the embodiment of the invention also provides one
The device that kind prediction traffic events influence, is described in detail with reference to the accompanying drawing.
Installation practice one:
Referring to fig. 4, which is one schematic diagram of Installation practice that prediction traffic events provided by the invention influence.
The present embodiment provides a kind of devices that prediction traffic events influence, comprising: determining module 100, data obtaining module
200, current signature vector generation module 300, matching module 400 and the first prediction module 500；
The determining module 100, the road chain m and road chain m- associated with road chain m occurred for determining traffic events
1 ... ..., road chain m-n；The associated road chain m-1 ... ..., road chain m-n be show the way chain m occur traffic events influence model
Enclose the road chain involved；The n is preset value；
It is understood that traffic events occur on the chain m of road, other associated road chains can be involved, therefore, it is necessary to judge
The coverage of the neighbouring road road Lian Shou chain l.For example, every road chain length is 100 meters, congestion caused by traffic events is usually big
In 1000 meters or more, then n at least takes 10.The value of n depends on the coverage that traffic events need to predict, range is longer, then n
It is bigger.
The data obtaining module 200, for obtaining when traffic events occur the road chain m and associated with road chain m
Road chain m-1 ... ..., the floating car data of road chain m-n；
It should be noted that floating car data includes longitude and latitude and speed.
The current signature vector generation module 300, for according to the floating car data obtain road chain m speed and
Road chain m-1 ... ... associated with road chain m, the speed of road chain m-n, from all of above road chain speed generate current signature to
Amount；
Road chain m, and n road chain associated with road chain m have n+1 road chain altogether, are corresponding with n+1 speed；This n+1
Speed forms a vector, i.e. current signature vector.
For example, the speed v of current period road chain m can be indicated are as follows: v (tnow,m)；The speed of current period road chain m-n can
To indicate are as follows: v (tnow,m-n)；Wherein, v indicates speed, tnowIndicate that current period, m indicate road chain m.
Then current signature vector s (t when road chain m generation traffic eventsnow, m) and it indicates are as follows:
s(tnow, m) and={ v (tnow,m),v(tnow,m-1),v(tnow,m-2),......v(tnow,m-n)} (1)
Current signature vector is a vector of speed it can be seen from formula (1), that is, includes the speed on n+1 road chain
Degree.
The matching module 400 is searched and the matched matching of current signature vector similarity from historical pattern library
Feature vector；
First prediction module 500, for according in the corresponding predicted characteristics vector of matching characteristic vector found
Speed obtains the speed of road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n prediction, the historical pattern
Existing matching characteristic vector sum predicted characteristics vector in pairs is saved in library；In the matching characteristic vector sum predicted characteristics vector
The speed of road chain be to cause the speed of road chain congestion Shi Lulian to obtain according to non-traffic events in advance；Wherein matching characteristic vector
Road chain m and road chain m-1 ... ... associated with road chain m when for period t, the speed of road chain m-n, prediction characteristic vector are week
The road Qit+1Shi chain m and road chain m-1 ... ... associated with road chain m, road chain m-n speed.
It should be noted that the history feature vector in historical pattern library is to have obtained in advance, historical pattern is established
The speed of congestion road chain, the i.e. congestion of road chain are not as caused by traffic events when the data of library foundation are non-traffic events.
For example, the speed of congestion road chain establishes historical pattern library when can use the non-traffic events in one month.Because of traffic thing
Congestion caused by part influences the problem of being really congestion diffusion, caused by conventional congestion (non-traffic events) diffusion and traffic events
Congestion diffusion is substantially because caused by traffic bottlenecks.Therefore, it can use conventional congestion data and establish historical pattern library.
Wherein, existing vector in pairs is saved in history feature library to be indicated with h (t, m)={ s (t, m), s (t+1, m) }.
Wherein, s (t, m) is matching characteristic vector, and s (t+1, m) is predicted characteristics vector.
By current signature vector s (tnow, m) matched with matching characteristic vector s (t, m), if similarity mode at
Function, then the corresponding predicted characteristics vector s (t+1, m) of the matching characteristic vector is just the speed of all of above road chain in next period
The vector of composition.For example, a cycle is 5 minutes.
That is, s (t, m) is used for and s (tnow, m) and it is matched, s (t+1, m) is corresponding predicted characteristics vector, s (t+1, m)
In each road chain speed be exactly each road chain of next period predicted speed.
Prediction meanss provided in this embodiment, not judged using the unique characteristics of traffic events as parameter, and
It is that the road chain of traffic events will occur using the pairs of existing matching characteristic vector sum predicted characteristics vector having built up in advance
Speed and chain speed in road associated there composition current signature vector, with history feature vector in matching characteristic vector
The matching of similarity is carried out, the corresponding predicted characteristics vector of the matching characteristic vector matched is each of next period of prediction
The speed of a road chain.It can predict that traffic events occur on outlet chain m, and the speed of other road chains to extend influence.This
Apply for unique characteristics of this device independent of traffic events, therefore, prediction result is more accurate.
Installation practice two:
Referring to Fig. 5, which is two schematic diagram of Installation practice that prediction traffic events provided by the invention influence.
Device provided in this embodiment, further includes: unit 600 is established in historical pattern library；
Unit 600 is established in the historical pattern library, for presetting when road chain is respectively less than in the speed of period t and period t+1
When occurring on threshold speed and the road chain without traffic events, determine that the road chain is congestion road chain caused by non-traffic events, it will
Speed of the speed of congestion road chain period t as the road matching properties vector Zhong Gai chain, by the period t+1 of the congestion road chain
Speed as the speed with the matching characteristic vector existing road predicted characteristics vector Zhong Gai chain in pairs.
For example, it is necessary to meet following condition for the speed sample in historical pattern library, i.e.,And
WhereinFor pre-set velocity threshold value, then it represents that the road chain m in next period of period t and t is congestion road chain.In historical pattern library
Speed sample standard deviation is the speed of congestion road chain.
For example, pre-set velocity threshold value is 20km/h.The speed of current period is lower than 20km/h, and the speed in next period
When also below 20km/h, then illustrate that the road chain gets congestion.
It should be noted that only being introduced in Installation practice one with predicting the road chain speed in next period, Ke Yili
Solution will not only predict the road chain speed of next cycle, can generally predict subsequent multiple after traffic events generally occur
The road chain speed in period, for example, may predict to occur traffic events by taking 5 minutes a cycles as an example with the road of half an hour after
The road chain speed of 1 hour after traffic events occurs for chain speed, or prediction.
First prediction module 500 is also used to re-start highest similarity matching, predicts the above institute of subsequent cycle
There is the speed of road chain.
It is understood that in addition to predicting that next period is to establish current spy with the speed that the data of current Floating Car obtain
It levies other than vector, predicts the speed of phase week after next, and predict that the speed of the subsequent cycle of phase week after next is the previous period
The speed come has been predicted as current signature vector, has re-started highest similarity matching.
For example, the prediction meanss provided through this embodiment, can after traffic events generation, by prediction publication with
Under traffic direction information: western 400 meter of four vehicle of North 4th Ring Road institute bridge knocks into the back, and current vehicle is lined up 500 meters, it is contemplated that after 10 minutes
Vehicle queue is up to 3000 meters, and complete bridge to the estimated transit time of institute's bridge is 30 minutes, it is proposed that driver avoids this road.
Device provided in this embodiment can be predicted when traffic events occur, the speed that related road chain is got on the car, in turn
The jam situation that can predict automobile in subsequent predetermined time section, using the jam situation as Traffic information demonstration to driver, department
Machine can avoid the road chain of these congestions.Device provided in this embodiment does not contact directly with traffic events itself, therefore, fits
For all kinds of traffic events, applicability is more extensive, and prediction result is more quasi- than the method dependent on traffic events unique characteristics
Really.
Due to inherently a kind of anomalous event of traffic events, the generation of anomalous event is a small amount of, the prior art certainly
In based on traffic events unique characteristics prediction traffic events influence, need many traffic events as training sample, so
This method lack of training samples.In conventional method, a kind of mode is to use all traffic events as the training of certain stretch
Sample, but the complexity of urban road reduces the universality of this method, in turn results in prediction result inaccuracy.
Traffic events usually will cause congestion (having little significance to the traffic events predicted impact for not causing congestion), this hair
The bright middle influence by traffic events to traffic is interpreted as congestion diffusion (or low speed diffusion) problem, builds by history congestion information
Vertical historical pattern library, what which was reflected is exactly the diffusion of congestion at any time.Since history congestion is a kind of routine
Event, urban road are all getting congestion daily, therefore the sample data of this method provided by the invention is sufficient.
Installation practice three:
Referring to Fig. 6, which is three schematic diagram of Installation practice that prediction traffic events provided by the invention influence.
Device provided in this embodiment, first prediction module 500 include: computational submodule 501 and prediction submodule
502；
Matched sub-block 401, for by linear weighted function calculate in the historical pattern library with the current signature vector
K nearest matching characteristic vector of Euclidean distance；K is the integer greater than 1；It is special that the k matching characteristic vector corresponds to k prediction
Levy vector；
Highest similarity matching is matched so that Euclidean distance is nearest as an example in the present embodiment.Calculate current signature vector
With the Euclidean distance between the history feature vector in historical pattern library, obtain k with current signature vector distance going through recently
History feature vector.
It is for instance possible to obtain the 3 history feature vectors nearest with current signature vector, i.e. k is 3.
It should be noted that size of the similarity to measure inter-individual difference, including Euclidean distance, Minkowski away from
From, manhatton distance, vector space cosine similarity, Pearson correlation coefficients, Jaccard similarity factor.In engineer application,
Euclidean distance is most popular phase knowledge and magnanimity measure.
It should be noted that when carrying out similarity calculation, in order to avoid fast road chain occupies biggish calculating ratio
Weight, the low road chain of speed occupy lesser specific gravity, can be special to the matching in all history feature vectors in historical pattern library
Sign vector is standardized, and is then weighted Euclidean distance similarity calculation again.
For example, standardization each matching characteristic vector can be normalized, that is, it is transformed into 0 to 1 section.Than
Such as, it is assumed that the speed that chain m in road occurs for traffic events is 5km/h, and road chain m-5 speed is 70kn/h, a certain spy in historical pattern library
The speed for levying vector m is 10km/h, and the speed of m-5 is 50km/h.Matching of the matching compared to road chain m of road chain m-5 is in phase at this time
When calculating like degree, bigger specific gravity is occupied, such situation will influence the accuracy of final prediction result.In standardization
Afterwards, all in the range of 0-1, this phenomenon will not exist all values, so as to improve the standard of similarity calculation
Exactness, and then effectively improve the accuracy of prediction result.
Method provided in this embodiment, when carrying out similarity mode, to the corresponding matching characteristic of history feature vector to
Amount is standardized, and is then weighted Euclidean distance to the matching characteristic vector after standardization and is calculated, obtains final prediction
As a result.This avoid in similarity mode, the biggish road chain of speed occupies biggish specific gravity, is conducive to improve prediction in this way
As a result accuracy.
The prediction submodule 502, for obtaining k speed of each road chain by the k predicted characteristics vector, to every
K speed of a road chain is averaged the speed as road chain prediction.
First prediction module 500 provided in this embodiment further include: normalizer module 503；
The normalizer module 503, for being searched and the current signature vector similarity from historical pattern library
The matching characteristic vector in the historical pattern library is normalized in matched matching characteristic vector.
The computational submodule 501, for by linear weighted function calculate in the historical pattern library with the current signature
Matching characteristic vector after k nearest normalization of vector Euclidean distance；K is the integer greater than 1；The k matching characteristic to
Measure corresponding k predicted characteristics vector.
Linear weighted function calculating is carried out to the k predicted characteristics vector, the predicted characteristics vector after being calculated is prediction
Next period all of above road chain speed.
It is understood that it is special to be based on k prediction in order to improve the prediction accuracy of traffic events influence, in the present embodiment
Vector is levied to predict the road chain speed in next period, i.e., not merely with a predicted characteristics vector, but utilizes multiple predictions special
Levy road chain speed of the result after vector is weighted as next period of prediction.
Device provided in this embodiment goes out multiple history feature vectors for current signature Vectors matching, then multiple to this
The corresponding predicted characteristics vector of history feature vector is standardized, and is then weighted to the predicted characteristics vector after standardization
Calculating or final prediction result.It is matched, be can be improved by the higher multiple history feature vectors of similarity in this way
The accuracy of prediction.It is special to prediction between weighted calculation and in order to avoid the biggish road chain of speed occupies biggish weight
Sign vector is standardized, and is conducive to the accuracy for improving prediction result in this way.
The above described is only a preferred embodiment of the present invention, being not intended to limit the present invention in any form.Though
So the present invention has been disclosed as a preferred embodiment, and however, it is not intended to limit the invention.It is any to be familiar with those skilled in the art
Member, without departing from the scope of the technical proposal of the invention, all using the methods and technical content of the disclosure above to the present invention
Technical solution makes many possible changes and modifications or equivalent example modified to equivalent change.Therefore, it is all without departing from
The content of technical solution of the present invention, according to the technical essence of the invention any simple modification made to the above embodiment, equivalent
Variation and modification, all of which are still within the scope of protection of the technical scheme of the invention.
Claims (8)
1. a kind of method for predicting traffic events coverage, which comprises the following steps:
Determine road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n that traffic events occur；It is described associated
Road chain m-1 ... ..., road chain m-n is that the road chains that involve of coverage of traffic events occurs for the chain m that shows the way；The n is default
Value；
Obtain the road chain m and road chain m-1 ... ... associated with road chain m when traffic events occur, the floating of road chain m-n
Car data；
The speed and road chain m-1 ... ... associated with road chain m, road chain m-n of road chain m are obtained according to the floating car data
Speed, by all of above road chain speed generate current signature vector；
It is searched and the matched matching characteristic vector of the current signature vector similarity from historical pattern library；
According to the speed in the corresponding predicted characteristics vector of the matching characteristic vector found, road chain m and related to road chain m is obtained
The speed of the road chain m-1 ... ... of connection, road chain m-n prediction saves existing matching characteristic vector in pairs in the historical pattern library
With predicted characteristics vector；The speed of road chain in the matching characteristic vector sum predicted characteristics vector is in advance according to non-traffic thing
Part causes the speed of road chain congestion Shi Lulian to obtain；Road chain m and related to road chain m when wherein matching characteristic vector is period t
The road chain m-1 ... ... of connection, the speed of road chain m-n, road chain m and associated with road chain m when predicted characteristics vector is period t+1
Road chain m-1 ... ..., road chain m-n speed.
2. it is according to claim 1 prediction traffic events coverage method, which is characterized in that the matching characteristic to
Road chain speed in amount and predicted characteristics vector in advance according to non-traffic events when congestion road chain speed obtain, specifically:
Occur when road chain is respectively less than in pre-set velocity threshold value and the road chain in the speed of period t and period t+1 without traffic events
When, determine that the road chain is congestion road chain caused by non-traffic events, it is special using the speed of congestion road chain period t as the matching
The speed for levying the road vector Zhong Gai chain, is deposited in pairs using the speed of the period t+1 of the congestion road chain as with the matching characteristic vector
The road predicted characteristics vector Zhong Gai chain speed.
3. the method for prediction traffic events coverage according to claim 1, which is characterized in that described from historical pattern
It is searched in library with the matched matching characteristic vector of the current signature vector similarity, the matching characteristic vector found is corresponding pre-
The speed surveyed in feature vector is the road chain m and road chain m-1 ... ... associated with road chain m in next period of prediction, road chain
The speed of m-n, specifically:
It is calculated in the historical pattern library by linear weighted function and matches spy with the current signature vector Euclidean distance nearest k
Levy vector；K is the integer greater than 1；The k matching characteristic vector corresponds to k predicted characteristics vector；It is special by the k prediction
Sign vector obtains k speed of each road chain, is averaged the speed as road chain prediction to k speed of each road chain.
4. the method for prediction traffic events coverage according to claim 3, which is characterized in that from historical pattern library
Middle lookup and the matched matching characteristic vector of the current signature vector similarity, before further include:
Matching characteristic vector in the historical pattern library is normalized；
The lookup from historical pattern library and the matched matching characteristic vector of the current signature vector similarity, specifically:
K normalization nearest with the current signature vector Euclidean distance in the historical pattern library is calculated by linear weighted function
Matching characteristic vector afterwards.
5. a kind of device that prediction traffic events influence characterized by comprising determining module, data obtaining module, current spy
Levy vector generation module, matching module and the first prediction module；
The determining module, the road chain m and road chain m-1 ... ... associated with road chain m occurred for determining traffic events,
Road chain m-n；The associated road chain m-1 ... ..., road chain m-n be show the way chain m occur traffic events coverage involve
Road chain；The n is preset value；
The data obtaining module, for obtaining the road chain m and road chain m- associated with road chain m when traffic events occur
1 ... ..., the floating car data of road chain m-n；
The current signature vector generation module, for according to the floating car data obtain road chain m speed and with road chain m
Associated road chain m-1 ... ..., the speed of road chain m-n generate current signature vector by the speed of all of above road chain；
The matching module, from historical pattern library search with the matched matching characteristic of current signature vector similarity to
Amount；
First prediction module, for obtaining according to the speed in the corresponding predicted characteristics vector of matching characteristic vector found
The speed predicted to road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n saves in the historical pattern library
Existing matching characteristic vector sum predicted characteristics vector in pairs；Road chain in the matching characteristic vector sum predicted characteristics vector
Speed is to cause the speed of road chain congestion Shi Lulian to obtain according to non-traffic events in advance；Wherein matching characteristic vector is period t
Shi Lulian m and road chain m-1 ... ... associated with road chain m, the speed of road chain m-n, when predicted characteristics vector is period t+1
Road chain m and road chain m-1 ... ... associated with road chain m, road chain m-n speed.
6. the device that prediction traffic events according to claim 5 influence, which is characterized in that further include: historical pattern library
Establish unit；
Unit is established in the historical pattern library, for being respectively less than pre-set velocity threshold value in the speed of period t and period t+1 when road chain
And when occurring on the road chain without traffic events, determine that the road chain is congestion road chain caused by non-traffic events, by the congestion road
Speed of the speed of chain period t as the road matching characteristic vector Zhong Gai chain makees the speed of the period t+1 of the congestion road chain
For the speed with the matching characteristic vector existing road predicted characteristics vector Zhong Gai chain in pairs.
7. the device that prediction traffic events according to claim 5 influence, which is characterized in that the first prediction module packet
It includes: computational submodule and prediction submodule；
Computational submodule, for by linear weighted function calculate in the historical pattern library with the current signature vector Euclidean distance
K nearest matching characteristic vector；K is the integer greater than 1；The k matching characteristic vector corresponds to k predicted characteristics vector；
The prediction submodule, for obtaining k speed of each road chain by the k predicted characteristics vector, to k of each road chain
Speed is averaged the speed as road chain prediction.
8. the device that prediction traffic events according to claim 7 influence, which is characterized in that further include: normalization submodule
Block；The normalization submodule, for being searched and matched of the current signature vector similarity from historical pattern library
With feature vector, the matching characteristic vector in the historical pattern library is normalized；
The computational submodule, it is European with the current signature vector in the historical pattern library for being calculated by linear weighted function
Matching characteristic vector after k nearest normalization；K is the integer greater than 1；The k matching characteristic vector corresponds to k
Predicted characteristics vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510502663.8A CN106469503B (en) | 2015-08-14 | 2015-08-14 | A kind of method and apparatus for predicting traffic events coverage |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510502663.8A CN106469503B (en) | 2015-08-14 | 2015-08-14 | A kind of method and apparatus for predicting traffic events coverage |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106469503A CN106469503A (en) | 2017-03-01 |
| CN106469503B true CN106469503B (en) | 2019-05-24 |
Family
ID=58214656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510502663.8A Active CN106469503B (en) | 2015-08-14 | 2015-08-14 | A kind of method and apparatus for predicting traffic events coverage |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106469503B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106781511B (en) * | 2017-03-22 | 2019-07-26 | 北京工业大学 | A kind of congestion time forecasting methods based on GPS data and traffic accident type |
| CN108132056B (en) * | 2017-11-15 | 2021-04-30 | 北京通途永久科技有限公司 | Method for deducing bus route through GPS |
| CN108364463B (en) * | 2018-01-30 | 2020-07-31 | 重庆交通大学 | Traffic flow prediction method and system |
| CN108399778A (en) * | 2018-05-08 | 2018-08-14 | 深圳市易成自动驾驶技术有限公司 | Swarm intelligence congestion reminding method, system and computer readable storage medium |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7680594B2 (en) * | 2004-04-06 | 2010-03-16 | Honda Motor Co., Ltd. | Display method and system for a vehicle navigation system |
| CN102110365B (en) * | 2009-12-28 | 2013-11-06 | 日电（中国）有限公司 | Road condition prediction method and road condition prediction system based on space-time relationship |
| CN103646542B (en) * | 2013-12-24 | 2016-01-20 | 北京四通智能交通系统集成有限公司 | A kind of Forecasting Methodology of Evaluating traffic impact area and device |
| CN103839411B (en) * | 2014-03-20 | 2016-03-02 | 中国航天系统工程有限公司 | A kind of accident impact surface analysis method based on dynamic information |
| CN104464304A (en) * | 2014-12-25 | 2015-03-25 | 北京航空航天大学 | Urban road vehicle running speed forecasting method based on road network characteristics |
| CN104751630B (en) * | 2014-12-31 | 2017-01-18 | 浙江工业大学 | Road traffic state acquisition method based on Kernel-KNN matching |
-
2015
- 2015-08-14 CN CN201510502663.8A patent/CN106469503B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN106469503A (en) | 2017-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Performance evaluation of vehicle-based mobile sensor networks for traffic monitoring | |
| JP5374067B2 (en) | Traffic condition simulation apparatus and program | |
| US8755991B2 (en) | Method and structure for vehicular traffic prediction with link interactions and missing real-time data | |
| CN111712862B (en) | Method and system for generating traffic volume or traffic density data | |
| Xinghao et al. | Predicting bus real-time travel time basing on both GPS and RFID data | |
| CN106469503B (en) | A kind of method and apparatus for predicting traffic events coverage | |
| KR100820467B1 (en) | a traffic estimating system and the method considered road type | |
| JP5901838B2 (en) | How to predict future travel time on a link | |
| KR20050084501A (en) | Traffic information providing system, traffic information expression method and device | |
| JP3485774B2 (en) | Traffic flow simulation system | |
| Yang et al. | Queue estimation in a connected vehicle environment: A convex approach | |
| Aljamal et al. | Real-time estimation of vehicle counts on signalized intersection approaches using probe vehicle data | |
| Rupnik et al. | Travel time prediction on highways | |
| Waddell et al. | Utilizing low-ping frequency vehicle trajectory data to characterize delay at traffic signals | |
| CN105185107A (en) | GPS-based traffic running tendency prediction method | |
| Jones et al. | Predicting link travel times from floating car data | |
| Alrukaibi et al. | Real-time travel time estimation in partial network coverage: A case study in Kuwait City | |
| CN117496726A (en) | Traffic control coordination method and system based on intelligent traffic system | |
| Mazaré et al. | COMPUTING TRAVEL TIMES FROM FILTERED TRAFFIC STATES. | |
| CN115641721B (en) | Multi-source traffic flow index fusion and road condition calculation method based on fusion index | |
| JP2019020799A (en) | Traffic management device, traffic management system, and traffic management method | |
| Rehborn et al. | Common traffic congestion features studied in USA, UK, and Germany based on kerner's three-phase traffic theory | |
| CN105070060A (en) | Method for determining urban road traffic state based bus-mounted GPS data | |
| JP3355887B2 (en) | Travel time prediction method and travel time prediction device | |
| Tettamanti | Advanced methods for measurement and control in urban road traffic networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200507 Address after: 310052 room 508, floor 5, building 4, No. 699, Wangshang Road, Changhe street, Binjiang District, Hangzhou City, Zhejiang Province Patentee after: Alibaba (China) Co.,Ltd. Address before: 102200, No. 8, No., Changsheng Road, Changping District science and Technology Park, Beijing, China. 1-5 Patentee before: AUTONAVI SOFTWARE Co.,Ltd. |