CN107767676A - A kind of method and apparatus for contributing to Traffic signal control - Google Patents
A kind of method and apparatus for contributing to Traffic signal control Download PDFInfo
- Publication number
- CN107767676A CN107767676A CN201610710525.3A CN201610710525A CN107767676A CN 107767676 A CN107767676 A CN 107767676A CN 201610710525 A CN201610710525 A CN 201610710525A CN 107767676 A CN107767676 A CN 107767676A
- Authority
- CN
- China
- Prior art keywords
- traffic
- crossing
- conditions associated
- signal control
- grader
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
Abstract
The present invention relates to a kind of method and apparatus for contributing to Traffic signal control, the device (500) includes：One traffic acquisition module (502), the currently practical crossing traffic for obtaining a crossing are conditions associated；And, one sort module (504), for conditions associated based on the currently practical crossing traffic acquired in the traffic acquisition module (502), a kind of Traffic signal control instruction for the crossing is determined using an acquired grader, wherein, the grader is used for conditions associated for any crossing traffic for occurring, generates a kind of Traffic signal control instruction conditions associated for any crossing traffic for handling the appearance.Using this method and device, Traffic signal control can be improved to improve the patency of traffic.
Description
Technical field
The present invention relates to field of traffic, more particularly to a kind of method and apparatus for contributing to Traffic signal control.
Background technology
With quick economic development, vehicle is more and more, and traffic network is also more and more intensive.In order to ensure traffic is led to
Freely, traffic lights are set to guide vehicle to travel in order generally at the crossing of traffic network.
At present, the control generally use static traffic control strategy to traffic lights.According to static traffic control strategy,
The traffic lights duration of traffic lights generally in daily each period is all identical and changeless.
However, the traffic of traffic network and frequently changeless, but time to time change, therefore, use
Static traffic control strategy controls traffic lights not always guarantee that traffic is unobstructed, the especially working in the every workday
Period and next period are even more so.
The content of the invention
The defects of in view of prior art, embodiments of the invention provide a kind of method for contributing to Traffic signal control
And device, it can improve Traffic signal control to improve the patency of traffic.
According to a kind of method for contributing to Traffic signal control of the embodiment of the present invention, including：Obtain at least two instructions
Practice sample, each training sample is used for what is represented that a kind of crossing traffic occurred is conditions associated and occurred for handling this
A kind of conditions associated Traffic signal control instruction of crossing traffic；And based at least two training sample, train
To a grader, wherein, any crossing traffic that the grader is used to be directed to appearance is conditions associated, generates for handling
A kind of Traffic signal control that any crossing traffic of the appearance is conditions associated instructs.Here, instructed using training sample
Get and handed over for the conditions associated any crossing to generate for handling the appearance of any crossing traffic for appearance
The grader of logical conditions associated Traffic signal control instruction, so as to be directed to out using the grader for training to obtain
Existing any crossing traffic is conditions associated and generates the traffic conditions associated for any crossing traffic for handling the appearance
Signalized control instructs, it is thus possible to improves Traffic signal control to improve the patency of traffic.
Wherein, at least two training samples of the acquisition, including：Obtain at least two history training samples；And in institute
State and be based at least two training sample, after training obtains a grader, in addition to：Obtain at least two training in real time
Sample；And the grader that the described at least two real-time training samples based on acquisition obtain to training is modified.This
In, grader is corrected using real-time training sample, the classification degree of accuracy of grader can be improved.
Wherein, at least two real-time training samples of the acquisition include：A crossing based on training obtain described in point
When the Traffic signal control of class device generation instructs mismatch conditions associated with the traffic of the appearance at the crossing, described in acquisition at least
Two real-time training samples.Here there is provided a kind of condition for triggering amendment grader, when Traffic signal control instructs and should
During the conditions associated mismatch of the traffic of the appearance at crossing, it is conditions associated to illustrate that Traffic signal control instruction can not be directed to traffic
Correctly handled, now correct grader using real-time training sample, the classification degree of accuracy of grader can be improved, entered
And more accurately Traffic signal control instruction can be obtained.
Wherein, at least two training samples of the acquisition include：Obtain at least one as set by experienced traffic-police
The training sample corresponding to Traffic signal control instruction put.Here, obtain as the friendship set by experienced traffic-police
Sample corresponding to the control instructions such as messenger, the Traffic signal control instruction of the sample of acquisition more can the actual road of accurate match
Oral sex is led to conditions associated.The control instruction such as traffic signals caused by the grader obtained based on such sample also more can
It is conditions associated with actual crossing traffic.
Wherein, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing traffic amount, road
Vehicle queue length at mouthful, average speed at crossing, weather conditions at crossing, whether working day and period, wherein, it is described to be
No working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, the period refers to
The time that the crossing traffic is conditions associated when being acquired belongs to which of one day period.Here, using in information above
At least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic conditions associated.
Wherein, the grader is random forest, decision tree, SVMs, neutral net or Bayes classifier.This
In, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier, the classification of grader
The degree of accuracy is higher.
According to a kind of method for contributing to Traffic signal control of the embodiment of the present invention, including：One crossing of acquisition
Currently practical crossing traffic is conditions associated；And it is conditions associated based on acquired currently practical crossing traffic, using having obtained
A grader come determine for the crossing a kind of Traffic signal control instruction, wherein, the grader is used for pin
It is conditions associated to any crossing traffic of appearance, generate one conditions associated for any crossing traffic for handling the appearance
Kind Traffic signal control instruction.Here, it is used for using for any crossing traffic for appearance to be conditions associated to generate
Handle the appearance any crossing traffic it is conditions associated Traffic signal control instruction grader, generate for crossing
Instructed with the Traffic signal control of the conditions associated matching of currently practical crossing traffic at the crossing, so as to improve traffic letter
Signal lamp is controlled to improve the patency of traffic.
Wherein, the grader trains to obtain based at least two training samples, wherein, each training sample is used
In representing a kind of conditions associated and conditions associated for handling crossing traffic that this has occurred one kind of crossing traffic for having occurred
Traffic signal control instructs.Here, grader is to utilize the crossing traffic for being used for representing to have occurred conditions associated and for locating
The training sample for managing the conditions associated Traffic signal control instruction of crossing traffic that this has occurred trains what is obtained, so as to train
The classification degree of accuracy of obtained grader is higher.
Wherein, determine that a kind of Traffic signal control for the crossing instructs bag using acquired grader
Include：When the grader is a random forest, each decision tree obtained in the random forest is directed to the acquisition
Currently practical crossing traffic it is conditions associated and vote candidate's Traffic signal control instruction；Each described candidate is calculated to hand over
Ventilating signal lamp control instruction is voted-for number；And it is determined that it is voted-for the most candidate's Traffic signal control of number
Instruction, as the Traffic signal control instruction for the crossing.Here, determined using the grader of random forest type
The Traffic signal control instruction at crossing, can obtain the traffic lights with the conditions associated comparison match of the actual traffic at crossing
Control instruction.
According to a kind of device for contributing to Traffic signal control of the embodiment of the present invention, including：One acquisition module, use
In obtaining at least two training samples, each training sample is used to represent that a kind of crossing traffic occurred is conditions associated and uses
In handling a kind of conditions associated Traffic signal control instruction of the crossing traffic that has occurred；And a training module, use
In based at least two training sample acquired in the acquisition module, training obtains a grader, wherein, described point
Any crossing traffic that class device is used to be directed to appearance is conditions associated, generates any crossing traffic phase for handling the appearance
A kind of Traffic signal control instruction of pass situation.Here, train to obtain any for for appearance using training sample
Kind of crossing traffic is conditions associated to generate the traffic lights control conditions associated for any crossing traffic for handling the appearance
The grader of instruction is made, so as to be directed to any crossing traffic correlation-like occurred using the grader for training to obtain
Condition and generate the Traffic signal control instruction conditions associated for any crossing traffic for handling the appearance, it is thus possible to change
It is apt to Traffic signal control to improve the patency of traffic.
Wherein, the acquisition module is specifically used for obtaining at least two history training samples, and, described device is also wrapped
Include：One real-time sample acquisition module, after training to obtain the grader in the training module, obtain at least two
Real-time training sample；And a correcting module, for described at least two based on the acquisition of sample acquisition module in real time
The grader that real-time training sample trains to obtain to the training module is modified.Here, real-time training sample is utilized
To correct grader, the classification degree of accuracy of grader can be improved.
Wherein, the sample acquisition module in real time is specifically used for：In the grader that a crossing is obtained based on training
When the Traffic signal control of generation instructs mismatch conditions associated with the traffic of the appearance at the crossing, described at least two are obtained
Real-time training sample.Here there is provided a kind of condition for triggering amendment grader, when Traffic signal control instruction and the crossing
Appearance traffic conditions associated mismatch when, illustrate that Traffic signal control instruction can not be directed to the conditions associated progress of traffic
Correct processing, now corrects grader using real-time training sample, can improve the classification degree of accuracy of grader, Jin Erneng
Enough more accurately Traffic signal controls that obtains instruct.
Wherein, the acquisition module is specifically used for：Obtain at least one traffic as set by experienced traffic-police
The corresponding training sample of Signalized control instruction.Here, obtain as traffic signals set by experienced traffic-police etc.
Sample corresponding to control instruction, the Traffic signal control instruction of the sample of acquisition more can the actual crossing traffic phase of accurate match
Pass situation.The control instruction such as traffic signals also can more match actual road caused by the grader obtained based on such sample
Oral sex is led to conditions associated.
Wherein, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing traffic amount, road
Vehicle queue length at mouthful, average speed at crossing, weather conditions at crossing, whether working day and period；Wherein, it is described to be
No working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, the period refers to
The time that the crossing traffic is conditions associated when being acquired belongs to which of one day period.Here, using in information above
At least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic conditions associated.
According to a kind of device for contributing to Traffic signal control of the embodiment of the present invention, including：One traffic obtains
Modulus block, the currently practical crossing traffic for obtaining a crossing are conditions associated；And a sort module, for based on
Currently practical crossing traffic acquired in the traffic acquisition module is conditions associated, using an acquired grader come
It is determined that a kind of Traffic signal control for the crossing instructs, wherein, the grader is used for any for what is occurred
Crossing traffic is conditions associated, generates a kind of traffic lights control conditions associated for any crossing traffic for handling the appearance
System instruction.Here, using for conditions associated times to generate for handling the appearance of any crossing traffic for appearance
A kind of grader of the conditions associated Traffic signal control instruction of crossing traffic, is generated for the current with the crossing of crossing
The Traffic signal control instruction of the actual conditions associated matching of crossing traffic, so as to improve Traffic signal control to improve
The patency of traffic.
Wherein, the grader trains to obtain based at least two training samples, wherein, each training sample is used
In representing a kind of conditions associated and conditions associated for handling crossing traffic that this has occurred one kind of crossing traffic for having occurred
Traffic signal control instructs.Here, grader is to utilize the crossing traffic for being used for representing to have occurred conditions associated and for locating
The training sample for managing the conditions associated Traffic signal control instruction of crossing traffic that this has occurred trains what is obtained, so as to train
The classification degree of accuracy of obtained grader is higher.
Wherein, the sort module includes：One instruction acquisition module, for obtaining when the grader is one random
During forest, currently practical road of each decision tree acquired in the traffic acquisition module in the random forest is obtained
Candidate's Traffic signal control instruction that is conditions associated and voting is led in oral sex；One computing module, for calculating described in each
The instruction of candidate's Traffic signal control is voted-for number；And a determining module, it is most for determining to be voted-for number
Candidate's Traffic signal control instruction, as the Traffic signal control instruction for the crossing.Here, using random
The grader of Forest Types come determine the Traffic signal control at crossing instruct, the actual traffic correlation-like with crossing can be obtained
The Traffic signal control instruction of condition comparison match.
According to a kind of equipment for contributing to Traffic signal control of the embodiment of the present invention, including：Processor；And deposit
Reservoir, for storing executable instruction, wherein, the executable instruction causes the foregoing side of the computing device upon being performed
Operation included by method.
According to a kind of computer program product of the embodiment of the present invention, including：Machine readable media, being stored thereon with to hold
Row instruction, wherein, the executable instruction causes machine to perform the operation included by preceding method upon being performed.
Brief description of the drawings
Further feature, feature, advantage and the benefit of the present invention will be become more by the detailed description below in conjunction with accompanying drawing
Obviously.Wherein：
Figure 1A shows the ensemble stream of the method for contributing to Traffic signal control according to one embodiment of the present of invention
Cheng Tu；
Figure 1B shows the flow chart for being used to correct the method for grader according to one embodiment of the present of invention；
Fig. 1 C show the example of the grader before amendment；
Fig. 1 D show the example of revised grader；
Fig. 2 shows the flow chart of the method for contributing to Traffic signal control according to one embodiment of the present of invention；
Fig. 3 shows the flow chart of the method for contributing to Traffic signal control according to one embodiment of the present of invention；
Fig. 4 shows the schematic diagram of the device for contributing to Traffic signal control according to one embodiment of the present of invention；
Fig. 5 shows the schematic diagram of the device for contributing to Traffic signal control according to one embodiment of the present of invention；
And
Fig. 6 shows the schematic diagram of the equipment for contributing to Traffic signal control according to one embodiment of the present of invention.
Reference numerals list：
100：Contribute to the method for Traffic signal control； 102：Obtain sample；
104：Perform training； 106：Acquisition situation； 108：Obtain candidate instruction；
110：Calculation times； 112：Determine instruction； 114：Control signal lamp；
150：Method for correcting grader； 154：Collect sample；
158：Classified； 162：Calculate sample size；
166：The first particular leaf node is found out in trial； 170：Calculate Gini coefficients；
174：The second particular leaf node is found out in trial； 178：Calculate increment；
182：Determine Split Attribute； 186：Create child node； 20：Decision tree；
22：Leaf node； 200：Contribute to the method for Traffic signal control；
202：Obtain sample； 204：Training；
300：Contribute to the method for Traffic signal control； 302：Acquisition situation；
304：Determine control instruction； 3042：Obtain candidate instruction；
3044：Calculation times； 3046：Determine instruction；
400：Contribute to the device of Traffic signal control； 402：Acquisition module；
404：Training module； 406：Real-time sample acquisition module；
408：Correcting module； 500：Contribute to the device of Traffic signal control；
502：Traffic acquisition module； 504：Sort module；
5042：Instruction acquisition module； 5044：Computing module； 5046：Determining module；
600：Contribute to the equipment 610 of Traffic signal control：Processor；
620：Memory
Embodiment
As it was previously stated, current traffic control strategy control traffic lights do not always guarantee that traffic is unobstructed, then
How to ensure that traffic is unobstructed just becomes particularly important.It can ensure that a unobstructed feasible method of traffic is exactly according to crossing traffic
Traffic signal control conditions associated and that offer matches is instructed to control the traffic lights at crossing.
Based on above-mentioned thinking, embodiments of the invention provide a kind of scheme for contributing to Traffic signal control, and it is provided
Instructed with the conditions associated Traffic signal control to match of crossing traffic to control the traffic lights at crossing.The program utilizes
Training sample is generated for handling the appearance to train to obtain conditions associated for any crossing traffic for appearance
The grader of the conditions associated Traffic signal control instruction of any crossing traffic, then using the obtained grader of training come
Obtain and instructed with the conditions associated Traffic signal control to match of the actual crossing traffic at crossing, finally utilize acquired friendship
Ventilating signal lamp control instruction controls the traffic lights at crossing.Due to being based on training sample in the scheme of embodiments of the invention
The grader that this training obtains refers to obtain with the conditions associated Traffic signal control to match of the actual crossing traffic at crossing
Order is to control the traffic lights at crossing, therefore, the schemes of embodiments of the invention can improve Traffic signal control with
Improve the patency of traffic.
Each embodiment of the present invention is described in detail next, with reference to accompanying drawing.
Referring now to Figure 1A, it illustrates the side for contributing to Traffic signal control according to one embodiment of the present of invention
The overview flow chart of method.Method 100 shown in Figure 1A can be by any equipment with computing capability (such as, but not limited to,
Desktop computer, notebook computer, server etc.) realize.
As shown in Figure 1A, in square frame 102, at least two training samples are obtained from database.Certainly, training sample also may be used
To come tissue, such as storage etc. in the form of a file with the other forms in addition to database.Each training sample is used for
Represent a kind of conditions associated and conditions associated for handling crossing traffic that this has occurred traffic letter of crossing traffic occurred
Signal lamp control instruction.Training sample can be stored in hard disk or storage array, or can also be stored in distributed storage system
In system.
In one aspect, the Traffic signal control instruction represented by each training sample can be by experienced friendship
What logical police was set.These training samples can be related for different crossing traffics at crossing in experienced traffic-police
Situation and be provided for collecting to obtain and be stored in database during the corresponding traffic lights control instruction at crossing.It is experienced
Traffic-police is generally conditions associated to which kind of crossing traffic should to use have a deep understanding for which kind of Traffic signal control instruction,
Therefore, it, which is provided for the Traffic signal control instruction at crossing, usually can match the actual crossing traffic correlation-like at crossing
Condition, so as to improve Traffic signal control to improve the patency of traffic.
In specific implementation, the mark for the police for sending the control instructions such as traffic signals when storing training sample, can be stored
Knowledge information (such as：Police's numbering etc.).In addition, prestore the identification information of police and the corresponding relation of police's posterior infromation.
Such as：Police's numbering of one traffic-police is 020631, and the police has ten year traffics commander's experience, then can prestore
In (020631,1), the wherein corresponding relation, previous section is numbered for police, and " 1 " in aft section, which represents, 10 years and 10
Year above point duty experience.Alternatively, can also pre-set " 2 " and represent has 5 years to 9 year traffics commander's experience；By that analogy.
Alternatively, it is the police for having more than 5 years point duty experiences that can preset experienced traffic-police, certainly, can also be set
Experienced traffic-police is the police for having more than 10 years point duty experiences.
Therefore, in the step of square frame 102, when obtaining training sample, police that can first in the training sample of storage
Numbering obtain corresponding to police's posterior infromation, according still further to " experienced traffic-police " set in advance requirement from the instruction of storage
Practice in sample and filtered out training sample.
In one aspect, when Traffic signal control instruction can represent the green light of each travel direction in such as crossing
It is long.
In one aspect, crossing traffic is conditions associated can utilize vehicle queue length, crossing at crossing traffic amount, crossing
Place's average speed, weather conditions at crossing, whether working day and these attributes of period represent.Wherein, if working day is
Refer to crossing traffic is conditions associated be acquired when date whether be working day, and the period refer to it is conditions associated in crossing traffic
Time when being acquired belongs to which of one day period.In one aspect, crossing traffic amount, vehicle queue length at crossing
The various friendship general purpose transducers in the road for all directions being arranged at crossing can be for example utilized with average speed at crossing
Measurement obtains.In one aspect, weather conditions can for example obtain from internet at crossing.
In square frame 104, training is performed using acquired training sample, to obtain the random forest as grader
(random forest).It will be understood by those skilled in the art that although the grader that the present embodiment trains to obtain is random forest,
However, the invention is not limited in this.In other embodiments of the invention, grader can also be in addition to random forest
Other types of grader, such as, but not limited to, decision tree, neutral net, SVMs or Bayes classifier etc..
Those skilled in the art understand that random forest includes multiple decision trees (decision tree), wherein, it is each
Individual decision tree is a tree structure, and each internal node of the tree structure represents to survey some attribute of input
Examination, a kind of test output of each branching representation of the internal node, and, each leaf node expression pair of the tree structure
Certain classification of input.Those skilled in the art also understand that each decision tree included by random forest is based on to input
Part attribute test and will input ballot to certain classification, will input and be divided into certain classification.Using random forest
When, decision tree is the part of random forest, and certainly, decision tree also can realize the embodiment of the present invention independently as grader
The scheme of offer.
Specifically, the random forest that the present embodiment trains to obtain is used for any crossing traffic correlation-like for occurring
Condition, generate the Traffic signal control instruction conditions associated for any crossing traffic for handling the appearance.The present embodiment is instructed
Each internal node of each decision tree included by random forest got is represented to the crossing traffic as input
Some attribute in conditions associated is tested, and, each leaf node represents a certain Traffic signal control instruction.This
The conditions associated ballot of crossing traffic is arrived some traffic by each decision tree included by random forest that embodiment trains to obtain
Signalized control instructs.
Because the what state training random forest in given training sample is known to those skilled in the art,
Therefore, the description to how to train random forest is omitted herein.After training obtains as the random forest of grader, so that it may
With using random forest according to the conditions associated corresponding traffic lights control to obtain for crossing of the different crossing traffics at crossing
System instruction.
It is conditions associated in square frame 106, the currently practical crossing traffic for obtaining crossing.
In one aspect, vehicle queue length, crossing at the volume of traffic at crossing, crossing are utilized crossing traffic is conditions associated
Place's average speed, weather conditions at crossing, whether working day and period in the case of representing, can come in the following manner
The currently practical crossing traffic for obtaining crossing is conditions associated：Such as can be from the road of all directions at crossing
Various friendship general purpose transducers obtain the volume of traffic at crossings, average speed at vehicle queue length and crossing at crossing；Such as can be with
Weather conditions at crossing are obtained from internet；Such as current date and base can be obtained from the equipment of execution method 100
The value of this attribute of working day is determined whether in the acquired date；And such as can be from the equipment of execution method 100
It is upper to obtain current time and the value of this attribute of period is determined based on the acquired time.
In square frame 108, each decision tree in the random forest that training obtains is obtained for acquired currently practical
The Traffic signal control instruction that crossing traffic is conditions associated and votes.Below, the traffic signals each decision tree voted
Lamp control instruction is referred to as the instruction of candidate's Traffic signal control.
In square frame 110, calculate the instruction of each candidate's Traffic signal control is voted-for number.For example, it is assumed that candidate
N decision tree in the random forest that Traffic signal control instruction A is trained to obtain is voted, and, candidate's traffic lights
M decision tree in the random forest that control instruction B is trained to obtain is voted, then candidate's Traffic signal control instruction A's
It is n to be voted-for number, and, the candidate's Traffic signal control instruction B number that is voted-for is m.
In square frame 112, it is determined that the most candidate's Traffic signal control instruction of number is voted-for, as the friendship for crossing
Ventilating signal lamp control instruction.
In square frame 114, instructed based on the identified Traffic signal control for crossing, control the traffic signals at crossing
Lamp.
Wherein, square frame 102-104 operation is commonly referred to as the off-line training of grader, generally can be with off-line execution, to reality
When property requires not high.And square frame 106-114 operation is commonly referred to as online classification, typically held in real time for a crossing
Capable.
It can be seen from the above that the scheme of the present embodiment trains to obtain for for appearance using training sample
Any crossing traffic is conditions associated to generate the traffic signals conditions associated for any crossing traffic for handling the appearance
The grader of lamp control instruction, and, obtain the actual traffic crossing correlation-like with crossing using the obtained grader of training
The Traffic signal control instruction that condition matches instructs as the Traffic signal control for crossing, so as to the present embodiment
Scheme enables to Traffic signal control instruction for crossing and the actual traffic crossing at crossing is conditions associated matches, because
This, the scheme of the present embodiment can improve Traffic signal control to improve the patency of traffic.
In one aspect, after training obtains as the random forest of grader, can also continue to collect training sample
To correct random forest to improve the classification degree of accuracy of random forest.Here, to put it differently, being obtained for training random gloomy
The training sample of woods is referred to as history training sample, and the training sample for being used to correct random forest is referred to as training sample in real time
This.
Referring now to Figure 1B, it illustrates the stream for being used to correct the method for grader according to one embodiment of the present of invention
Cheng Tu.Method 150 shown in Figure 1B can by with computing capability any equipment (such as, but not limited to, desktop computer,
Notebook computer, server etc.) realize.
As shown in Figure 1B, in square frame 154, after training obtains as the random forest F of grader, collect for correcting
Random forest F at least two real-time training samples, wherein, each real-time training sample is used to represent a kind of road occurred
Conditions associated and conditions associated for handling crossing traffic that this has occurred Traffic signal control instruction is led in oral sex.At one
Aspect, the Traffic signal control instruction represented by each real-time training sample can be set by experienced traffic-police
's.Collected real-time training sample can also be stored in the various storage mediums mentioned by preceding step 102, and it can be with
History training sample is stored in same storage medium, for correcting grader.In one aspect, obtained regardless of training random
The instruction of Traffic signal control that forest F is provided whether with the actual crossing traffic at crossing is conditions associated matches, training
After obtaining the random forest F as grader, begin to collect real-time training sample.In another aspect, when discovery is trained
The Traffic signal control instruction and the conditions associated mismatch of actual crossing traffic at crossing that obtained random forest F is provided
When, just start to collect real-time training sample.
In square frame 158, using each decision tree included by random forest F to collected real-time training sample one by one
Classified, wherein, each real-time training sample will reach a certain in each decision tree included by random forest F
Individual leaf node.
In square frame 162, after classification is completed, each leaf node calculated in grader F all leaf nodes is arrived
The sample size reached.
In square frame 166, attempt to find out the first particular leaf node from random forest F all leaf nodes, wherein, first is special
Determine the sample size that leaf node is reached and be more than minimum samples α.The quantity of the first particular leaf node found is probably
0th, it is one or more.
In square frame 170, if finding at least one first particular leaf node, each the first particular leaf node is calculated
Gini coefficients.The Gini coefficients of each the first particular leaf node are to use the real-time training sample up to first particular leaf node
Originally calculated.For example, it is assumed that the sample for reaching the first particular leaf node Pi is N1, N2, N3, N5, N7, N9 this six samples,
So the first particular leaf node Pi Gini coefficients are calculated using sample N1, N2, N3, N5, N7, N9.Due to how to count
It is known to those skilled in the art to calculate the Gini coefficients of the node of decision tree, therefore, is omitted herein to the detailed of its
Description.
In square frame 174, attempt to find out the second particular leaf node from the first particular leaf node found, wherein, second
Specially the Gini coefficients of leaf node are more than threshold value beta.Here, the second particular leaf node is exactly the leaf node for needing to divide.Wherein, institute
The quantity of the second particular leaf node found is probably zero, one or more.
In square frame 178, if finding at least one second particular leaf node, for each the second particular leaf node
Qi, division can be used as at the second particular leaf node Qi by using the real-time training sample up to the second particular leaf node Qi to calculate
The increment of the Gini coefficients of each candidate attribute at least one candidate attribute of attribute.At least one candidate attribute is
The node of decision tree where second particular leaf node Qi other attributes outside used attribute.For example, with attribute
Collection " vehicle queue length at crossing traffic amount, crossing, average speed at crossing, weather conditions at crossing, whether working day, when
Between section " exemplified by, it is assumed that the decision tree where the second particular leaf node Qi had used attribute " weather conditions at crossing " and " had been
No working day ", then at least one candidate attribute be " crossing traffic amount ", " vehicle queue length at crossing ", " at crossing
Average speed " and " period ".Because how the increment of the Gini coefficients of computation attribute is to those skilled in the art
It is known, therefore, detailed description is omitted herein.
In square frame 182, candidate's category at least one candidate attribute with the increment of maximum Gini coefficients is determined
Property, the Split Attribute as the second particular leaf node Qi.For example, it is assumed that with other candidate attributes, " vehicle queue is grown at crossing
Degree ", " average speed at crossing " are compared with " period ", and the increment of the Gini coefficients of candidate attribute " crossing traffic amount " is maximum,
Then candidate attribute " crossing traffic amount " is confirmed as the second particular leaf node Qi Split Attribute.
In square frame 186, it is determined that after the second particular leaf node Qi Split Attribute, use up to the second particular leaf node
Q real-time training sample creates the second particular leaf node Qi child node.In this case, the second particular leaf node Qi becomes
The internal node of decision tree where into it, and the second particular leaf node Qi child node just becomes the leaf of corresponding decision tree
Node.
After Split Attribute in each the second particular leaf node Qi is determined and its child node is created, complete
Random forest F amendment.
Below, the example of the makeover process of grader is illustrated based on Fig. 1 C-1D.
Found for example, it is assumed that being calculated by square frame 154-174, as the decision tree included by the random forest F of grader
The real-time training samples number of the arrival of leaf node 22 in 20 is more than threshold value beta more than minimum samples α and Gini coefficient, that
Leaf node 22 belongs to the leaf node that needs divide.
From Fig. 1 C can be seen that property set " average speed at vehicle queue length at crossing traffic amount, crossing, crossing,
Weather conditions at crossing, whether working day, period " in, decision tree 20 used " weather conditions at crossing " and " whether
The two attributes of working day ", therefore, remaining attribute " crossing traffic amount ", " vehicle queue length at crossing ", " put down at crossing
Equal speed " and " period " are the candidate attributes that can be used as Split Attribute at leaf node 22.
Assuming that the real-time training sample for reaching leaf node 22 is N11, N12, N13, N15, wherein, real-time training sample is
Traffic signal control instruction indicated by N11, N12 is that traffic indicated by N13, N15 is believed for " short " and real-time training sample
Signal lamp control instruction is " length ", then uses real-time training sample N11, N12, N13, N15 up to leaf node 22 to calculate candidate
Attribute " crossing traffic amount ", " vehicle queue length at crossing ", " average speed at crossing " and " period " respective Gini systems
Several increments.It is assumed that the increment of the Gini coefficients of candidate attribute " crossing traffic amount " is maximum, then as shown in Fig. 2 candidate attribute
" crossing traffic amount " is confirmed as leaf node Qi Split Attribute, and, use the real-time training sample up to leaf node 22
N11, N12, N13, N15 create the two of leaf node Qi child nodes, and the two child nodes indicate respectively Traffic signal control
Instruct as " short " and " length ".
Other modifications
It will be appreciated by those skilled in the art that although square frame 102-104 operation and square frame 106-114 operation can be
Realized in same equipment, still, it is generally the case that square frame 102-104 operation and square frame 106-114 operation are respectively in difference
Equipment in realize.Such as, but not limited to, square frame 106-114 operation positioned at crossing be used for control traffic lights
Realized on computer.
Although it will be appreciated by those skilled in the art that in the above embodiments, method 100 include square frame 114 with based on
The identified Traffic signal control for crossing is instructed to control the traffic lights at crossing, however, not office of the invention
It is limited to this.In some other embodiment of the present invention, method 100 can not also include the operation of square frame 114.In such case
Under, method 100 is to determine the Traffic signal control instruction for crossing, and then traffic-police is used for according to identified
The Traffic signal control at crossing is instructed to control the traffic lights at crossing, and this is especially suitable for needing traffic-police to control manually
How the traffic lights at crossing processed but traffic-police are to control the situations lacked experience of the traffic lights at crossing.
It will be appreciated by those skilled in the art that although in the above embodiments, the grader that the training of square frame 104 obtains is
Random forest, however, the invention is not limited in this.In some other embodiment of the present invention, square frame 104 trains what is obtained
Grader can also be other types of grader, such as, but not limited to, decision tree, neutral net, SVMs or shellfish
This grader of leaf etc..Here, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier
When, the classification degree of accuracy of grader is higher.
It will be appreciated by those skilled in the art that although in the above embodiments, crossing traffic is conditions associated to utilize crossing
Locate the volume of traffic, vehicle queue length at crossing, average speed at crossing, weather conditions at crossing, whether working day and period
These attributes represent, however, the invention is not limited in this.In some other embodiment of the present invention, crossing traffic phase
Pass situation can also be vaporous merely with day at average speed at vehicle queue length at the volume of traffic at crossing, crossing, crossing, crossing
Condition, whether the part attribute in working day and period represents, or, conditions associated can utilize at crossing of crossing traffic hand over
Vehicle queue length at flux, crossing, average speed at crossing, weather conditions at crossing, whether in working day and period
At least a portion attribute and other types of suitable attribute represent, or, crossing traffic is conditions associated can utilize except
The volume of traffic at crossing, vehicle queue length at crossing, average speed at crossing, weather conditions at crossing, whether working day and when
Between other types of suitable attribute outside section represent.Here, crossing is represented using at least one in information above
Traffic is conditions associated, and it is conditions associated can relatively accurately to characterize crossing traffic.
Referring now to Fig. 2, it illustrates the side for contributing to Traffic signal control according to one embodiment of the present of invention
The flow chart of method.Method 200 shown in Fig. 2 can be realized by any equipment with computing capability.
As shown in Fig. 2 method 200 can include, in square frame 202, at least two training samples are obtained, each training sample
This is used to represent that a kind of crossing traffic occurred is conditions associated and conditions associated for handling the crossing traffic that this has occurred
A kind of Traffic signal control instruction.Square frame 202 can for example be realized using square frame 102.
Method 200 can also include, and in square frame 204, based at least two training sample, training obtains a classification
Device, wherein, any crossing traffic that the grader is used to be directed to appearance is conditions associated, generates for handling appointing for the appearance
A kind of a kind of conditions associated Traffic signal control instruction of crossing traffic.Square frame 204 for example can be using square frame 104 come real
It is existing.
Here, generated using training sample to train to obtain for any crossing traffic for appearance to be conditions associated
The grader of the conditions associated Traffic signal control instruction of any crossing traffic for handling the appearance, so as to profit
Generated with the grader for training to obtain come conditions associated for any crossing traffic occurred for handling the appearance
The conditions associated Traffic signal control instruction of any crossing traffic, it is thus possible to improve Traffic signal control and handed over improving
Logical patency.
In one aspect, square frame 202 can include, obtain at least two history training samples, and, square frame 204 it
Afterwards, method 200 can also include：At least two real-time training samples are obtained, and, described at least two based on acquisition is real-time
The grader that training sample obtains to training is modified.Here, grader is corrected using real-time training sample, can be with
Improve the classification degree of accuracy of grader.
In another aspect, at least two real-time training samples of the acquisition can include：Training is based at a crossing
When the Traffic signal control of obtained grader generation instructs mismatch conditions associated with the traffic of the appearance at the crossing,
Obtain described at least two real-time training samples.Here there is provided a kind of condition for triggering amendment grader, work as traffic lights
During the conditions associated mismatch of the traffic of the appearance at control instruction and the crossing, illustrate that Traffic signal control instructs can not be directed to
Traffic is conditions associated correctly to be handled, and now corrects grader using real-time training sample, can improve grader
The classification degree of accuracy, and then more accurately Traffic signal control instruction can be obtained.
In yet another aspect, at least two training samples of the acquisition can include：Obtain at least one by experienced
The training sample corresponding to Traffic signal control instruction set by traffic-police.Here, obtain by experienced traffic police
The samples corresponding to control instruction such as set traffic signals are examined, the Traffic signal control instruction of the sample of acquisition more can be accurate
It is conditions associated to match actual crossing traffic.The control such as traffic signals refers to caused by the grader obtained based on such sample
Order conditions associated also can more match actual crossing traffic.
In another aspect, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing
Vehicle queue length at the volume of traffic, crossing, average speed at crossing, weather conditions at crossing, whether working day and period, its
In, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, when described
Between section refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.Here, utilize
In information above at least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic correlation-like
Condition.
In still a further aspect, the grader is random forest, decision tree, SVMs, neutral net or pattra leaves
This grader.Here, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier,
The classification degree of accuracy of grader is higher.
Referring now to Fig. 3, it illustrates the side for contributing to Traffic signal control according to one embodiment of the present of invention
The flow chart of method.Method 300 shown in Fig. 3 can be realized by any equipment with computing capability.
As shown in figure 3, method 300 can include, in square frame 302, the currently practical crossing traffic phase at one crossing of acquisition
Pass situation.Square frame 302 can for example be realized using square frame 106.
Method 300 can also include, conditions associated based on acquired currently practical crossing traffic in square frame 304, utilize
An acquired grader come determine for the crossing a kind of Traffic signal control instruction, wherein, the grader
It is conditions associated for any crossing traffic for appearance, generate any crossing traffic correlation-like for handling the appearance
A kind of Traffic signal control instruction of condition.Square frame 304 can for example be realized using square frame 108-112.
Here, using for conditions associated times to generate for handling the appearance of any crossing traffic for appearance
A kind of grader of the conditions associated Traffic signal control instruction of crossing traffic, is generated for the current with the crossing of crossing
The Traffic signal control instruction of the actual conditions associated matching of crossing traffic, so as to improve Traffic signal control to improve
The patency of traffic.
In one aspect, the grader trains to obtain based at least two training samples, wherein, each training
Sample is used to represent that a kind of crossing traffic occurred is conditions associated and conditions associated for handling the crossing traffic that this has occurred
A kind of Traffic signal control instruction.Here, grader be utilize for represent the crossing traffic that has occurred it is conditions associated and
Training sample for handling the conditions associated Traffic signal control instruction of crossing traffic that this has occurred trains what is obtained, from
And train the classification degree of accuracy of obtained grader higher.
In another aspect, square frame 304 includes：In square frame 3042, when the grader is a random forest, obtain
The time that each decision tree in the random forest is conditions associated for the currently practical crossing traffic of the acquisition and votes
Traffic signal control is selected to instruct；In square frame 3044, being voted-for for each candidate's Traffic signal control instruction is calculated
Number；And in square frame 3046, it is determined that be voted-for the most candidate's Traffic signal control instruction of number, as
The Traffic signal control instruction at the crossing.Here, determine that the traffic at crossing is believed using the grader of random forest type
Signal lamp control instruction, it can obtain and be instructed with the Traffic signal control of the conditions associated comparison match of the actual traffic at crossing.
In yet another aspect, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing
Vehicle queue length at the volume of traffic, crossing, average speed at crossing, weather conditions at crossing, whether working day and period, its
In, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, when described
Between section refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.Here, utilize
In information above at least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic correlation-like
Condition.
In another aspect, the grader is random forest, decision tree, SVMs, neutral net or Bayes
Grader.Here, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier, point
The classification degree of accuracy of class device is higher.
Referring now to Fig. 4, it illustrates the dress for contributing to Traffic signal control according to one embodiment of the present of invention
The schematic diagram put.Device 400 shown in Fig. 4 can utilize software, hardware (such as integrated circuit or DSP etc.) or software and hardware combining
Mode realize.
Shown in Fig. 4, device 400 can include an acquisition module 402 and a training module 404.Acquisition module 402 is used
In obtaining at least two training samples, each training sample is used to represent that a kind of crossing traffic occurred is conditions associated and uses
In handling a kind of conditions associated Traffic signal control instruction of the crossing traffic that has occurred.Training module 404 is used to be based on
At least two training sample acquired in acquisition module 402, training obtain a grader, wherein, the grader is used
It is conditions associated in any crossing traffic for appearance, generate conditions associated for any crossing traffic for handling the appearance
A kind of Traffic signal control instruction.Acquisition module 402 can for example realize that training module 404 is for example using square frame 102
It can be realized using square frame 104.Here, train to obtain any crossing traffic for for appearance using training sample
It is conditions associated to generate point of the Traffic signal control instruction conditions associated for any crossing traffic for handling the appearance
Class device, so as to generate and use come conditions associated for any crossing traffic occurred using the grader for training to obtain
In the conditions associated Traffic signal control instruction of any crossing traffic for handling the appearance, it is thus possible to improve traffic signals
Lamp is controlled to improve the patency of traffic.
In one aspect, acquisition module 402 is specifically used for obtaining at least two history training samples, and, device 400 is also
It can include：One real-time sample obtains module 406, after obtaining the grader in the training of training module 404, obtains
At least two real-time training samples；And a correcting module 408, for obtaining the institute of the acquisition of module 406 based on real-time sample
Stating at least two real-time training samples trains the obtained grader to be modified training module 404.Here, using real-time
Training sample corrects grader, can improve the classification degree of accuracy of grader.
In another aspect, real-time sample acquisition module 406 is specifically used for：A crossing based on training obtain described in
When the Traffic signal control of grader generation instructs mismatch conditions associated with the traffic of the appearance at the crossing, described in acquisition extremely
Few two real-time training samples.Here there is provided it is a kind of trigger amendment grader condition, when Traffic signal control instruction with
During the conditions associated mismatch of the traffic of the appearance at the crossing, illustrate that Traffic signal control instruction can not be directed to traffic correlation-like
Condition is correctly handled, and now corrects grader using real-time training sample, can improve the classification degree of accuracy of grader,
And then more accurately Traffic signal control instruction can be obtained.
In yet another aspect, acquisition module 402 is specifically used for：Obtain at least one as set by experienced traffic-police
The corresponding training sample of Traffic signal control instruction.Here, obtain as the traffic set by experienced traffic-police
Sample corresponding to the control instructions such as signal, the Traffic signal control instruction of the sample of acquisition more can the actual crossing of accurate match
Traffic is conditions associated.The control instruction such as traffic signals also can more match caused by the grader obtained based on such sample
Actual crossing traffic is conditions associated.
In another aspect, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing
Vehicle queue length at the volume of traffic, crossing, average speed at crossing, weather conditions at crossing, whether working day and period；Its
In, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, when described
Between section refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.Here, utilize
In information above at least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic correlation-like
Condition.
In yet another aspect, the grader is random forest, decision tree, SVMs, neutral net or Bayes
Grader.Here, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier, point
The classification degree of accuracy of class device is higher.
Referring now to Fig. 5, it illustrates the dress for contributing to Traffic signal control according to one embodiment of the present of invention
The schematic diagram put.Device 500 shown in Fig. 5 can utilize software, hardware (such as integrated circuit or DSP etc.) or software and hardware combining
Mode realize.
Shown in Fig. 5, device 500 can include a traffic acquisition module 502 and a sort module 504.Traffic
The currently practical crossing traffic that situation acquisition module 502 is used to obtain a crossing is conditions associated.Sort module 504 is used to be based on
Currently practical crossing traffic acquired in traffic acquisition module 502 is conditions associated, using an acquired grader come
It is determined that a kind of Traffic signal control for the crossing instructs, wherein, the grader is used for any for what is occurred
Crossing traffic is conditions associated, generates a kind of traffic lights control conditions associated for any crossing traffic for handling the appearance
System instruction.Traffic acquisition module 502 can for example realize that sort module 504 can for example utilize side using square frame 106
Frame 108-112 is realized.Here, using for any crossing traffic for appearance it is conditions associated come generate be used for handle this
Occur any crossing traffic it is conditions associated Traffic signal control instruction grader, generate for crossing and the road
The Traffic signal control instruction of the conditions associated matching of currently practical crossing traffic of mouth, so as to improve traffic lights control
Make to improve the patency of traffic.
In one aspect, the grader trains to obtain based at least two training samples, wherein, each training
Sample is used to represent that a kind of crossing traffic occurred is conditions associated and conditions associated for handling the crossing traffic that this has occurred
A kind of Traffic signal control instruction.Here, grader be utilize for represent the crossing traffic that has occurred it is conditions associated and
Training sample for handling the conditions associated Traffic signal control instruction of crossing traffic that this has occurred trains what is obtained, from
And train the classification degree of accuracy of obtained grader higher.
In another aspect, sort module 504 can include 5042, computing modules 5044 of an instruction acquisition module
With a determining module 5046.Instruction obtains module 5042 and is used for when the grader is a random forest, described in acquisition
Currently practical crossing traffic correlation-like of each decision tree acquired in traffic acquisition module 502 in random forest
Condition and vote candidate's Traffic signal control instruction.Computing module 5044 is used to calculate each described candidate's traffic lights
Control instruction is voted-for number.Determining module 5046 is used to determine to be voted-for the most candidate's traffic lights control of number
System instruction, as the Traffic signal control instruction for the crossing.Here, using the grader of random forest type come really
Determine the Traffic signal control instruction at crossing, the traffic signals with the conditions associated comparison match of the actual traffic at crossing can be obtained
Lamp control instruction.
In yet another aspect, the crossing traffic it is conditions associated using in following message at least one of represent：Crossing
Vehicle queue length at the volume of traffic, crossing, average speed at crossing, weather conditions at crossing, whether working day and period, its
In, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be working day, when described
Between section refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.Here, utilize
In information above at least one of represent that crossing traffic is conditions associated, can relatively accurately characterize crossing traffic correlation-like
Condition.
In another aspect, the grader is random forest, decision tree, SVMs, neutral net or Bayes
Grader.Here, when grader is random forest, decision tree, SVMs, neutral net or Bayes classifier, point
The classification degree of accuracy of class device is higher.
Referring now to Fig. 6, it illustrates the equipment for contributing to Traffic signal control according to one embodiment of the invention
Schematic diagram.As shown in fig. 6, processor 610 and memory 620 can be included by contributing to the equipment 600 of Traffic signal control.
Memory 620 is used to store executable instruction, the executable instruction cause upon being performed the execution method 200 of processor 610 and/
Or each operation included by method 300.
Embodiments of the invention also provide a kind computer program product, including machine readable media, are stored thereon with
Executable instruction, when the executable instruction is performed so that each included by machine executed method 200 and/or method 300
Operation.
It will be appreciated by those skilled in the art that each embodiment disclosed above can be in the situation without departing from invention essence
Under make various changes and modifications.Therefore, protection scope of the present invention should be defined by the appended claims.
Claims (18)
1. a kind of method for contributing to Traffic signal control, including：
Obtain at least two training samples, each training sample be used for represent a kind of crossing traffic occurred it is conditions associated and
A kind of Traffic signal control instruction conditions associated for handling crossing traffic that this has occurred；And
Based at least two training sample, training obtains a grader, wherein, the grader is used for for appearance
Any crossing traffic is conditions associated, generates a kind of traffic letter conditions associated for any crossing traffic for handling the appearance
Signal lamp control instruction.
2. the method as described in claim 1, it is characterised in that
At least two training samples of the acquisition, including：Obtain at least two history training samples；
At least two training sample is based on described, after training obtains a grader, in addition to：
Obtain at least two real-time training samples；
The grader that the described at least two real-time training samples based on acquisition obtain to training is modified.
3. method as claimed in claim 2, it is characterised in that
At least two real-time training samples of the acquisition, including：
In the Traffic signal control instruction and the appearance at the crossing of the grader generation that a crossing is obtained based on training
Traffic conditions associated mismatch when, obtain at least two real-time training sample.
4. the method as described in any one of claims 1 to 3, wherein,
At least two training samples of the acquisition, including：Obtain at least one traffic as set by experienced traffic-police
The corresponding training sample of Signalized control instruction.
5. the method (200) as described in any one of Claims 1-4, wherein,
The crossing traffic it is conditions associated using in following message at least one of represent：
Vehicle queue length at crossing traffic amount, crossing, average speed at crossing, weather conditions at crossing, whether working day and
Period；Wherein, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be work
Day, the period refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.
6. a kind of method for contributing to Traffic signal control, including：
The currently practical crossing traffic for obtaining a crossing is conditions associated；And
It is conditions associated based on acquired currently practical crossing traffic, determine to be used for using an acquired grader described
A kind of Traffic signal control instruction at crossing,
Wherein, the grader is conditions associated for any crossing traffic for occurring, and generates for handling the appearance
A kind of conditions associated Traffic signal control instruction of any crossing traffic.
7. method as claimed in claim 6, wherein,
The grader trains to obtain based at least two training samples,
Wherein, each training sample is used to represent that a kind of crossing traffic occurred is conditions associated and for handling this has occurred
Crossing traffic it is conditions associated a kind of Traffic signal control instruction.
8. method as claimed in claim 6, wherein, determine that one kind for the crossing is handed over using acquired grader
Ventilating signal lamp control instruction includes：
When the grader is a random forest, each decision tree obtained in the random forest is directed to the acquisition
Currently practical crossing traffic it is conditions associated and vote candidate's Traffic signal control instruction；And
Calculate each candidate's Traffic signal control instruction is voted-for number；And
It is determined that the most candidate's Traffic signal control instruction of number is voted-for, as the traffic signals for the crossing
Lamp control instruction.
9. a kind of device (400) for contributing to Traffic signal control, including：
One acquisition module (402), for obtaining at least two training samples, each training sample is used to represent that one kind has been sent out
A kind of conditions associated and conditions associated for handling crossing traffic that this has occurred Traffic signal control of raw crossing traffic
Instruction；And
One training module (404), for based at least two training sample acquired in the acquisition module (402),
Training obtains a grader, wherein, any crossing traffic that the grader is used to be directed to appearance is conditions associated, and generation is used
In a kind of conditions associated Traffic signal control instruction of any crossing traffic for handling the appearance.
10. device (400) as claimed in claim 9, it is characterised in that
The acquisition module (402) is specifically used for obtaining at least two history training samples, and
Described device (400) also includes：
One real-time sample acquisition module (406), after obtaining the grader in the training module (404) training,
Obtain at least two real-time training samples；
One correcting module (408), described at least two for being obtained based on the sample acquisition module (406) in real time are real-time
The grader that training sample obtains to the training module (404) training is modified.
11. device (400) as claimed in claim 10, it is characterised in that
The sample acquisition module (406) in real time, is specifically used for：In the grader generation that a crossing is obtained based on training
Traffic signal control instruction and the appearance at the crossing traffic conditions associated mismatch when, it is real-time to obtain described at least two
Training sample.
12. device (400) as claimed in claim 9, wherein,
The acquisition module (402), is specifically used for：Obtain at least one traffic signals as set by experienced traffic-police
Training sample corresponding to lamp control instruction.
13. device (400) as claimed in claim 9, wherein,
The crossing traffic it is conditions associated using in following message at least one of represent：
Vehicle queue length at crossing traffic amount, crossing, average speed at crossing, weather conditions at crossing, whether working day and
Period；Wherein, it is described whether working day refer to the crossing traffic is conditions associated be acquired when date whether be work
Day, the period refer to the crossing traffic is conditions associated be acquired when time belong to which of one day period.
14. a kind of device (500) for contributing to Traffic signal control, including：
One traffic acquisition module (502), the currently practical crossing traffic for obtaining a crossing are conditions associated；And
One sort module (504), for being handed over based on the currently practical crossing acquired in the traffic acquisition module (502)
It is logical conditions associated, a kind of Traffic signal control instruction for the crossing is determined using an acquired grader,
Wherein, the grader is conditions associated for any crossing traffic for occurring, and generates for handling the appearance
A kind of conditions associated Traffic signal control instruction of any crossing traffic.
15. device (500) as claimed in claim 14, wherein,
The grader trains to obtain based at least two training samples, wherein, each training sample is used to represent one
A kind of conditions associated and conditions associated for handling crossing traffic that this has occurred traffic signals of the crossing traffic kind occurred
Lamp control instruction.
16. device (500) as claimed in claim 14, wherein, the sort module (504) includes：
One instruction acquisition module (5042), for when the grader is a random forest, obtaining the random forest
In each decision tree it is conditions associated for the currently practical crossing traffic acquired in the traffic acquisition module (502)
And the candidate's Traffic signal control instruction voted；And
One computing module (5044), number is voted-for for calculate each described candidate's Traffic signal control instruction；
And
One determining module (5046), for determining to be voted-for the most candidate's Traffic signal control instruction of number, make
To be instructed for the Traffic signal control at the crossing.
17. a kind of equipment (600) for contributing to Traffic signal control, including：
At least one processor (610)；And
At least one memory (620), for storing executable instruction, wherein, the executable instruction causes upon being performed
Operation included by any one at least one processor (610) perform claim requirement 1-8.
18. a kind of computer program product, including：
Machine readable media, executable instruction is stored thereon with, wherein, the executable instruction causes machine to hold upon being performed
Operation included by any one in row claim 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610710525.3A CN107767676A (en) | 2016-08-23 | 2016-08-23 | A kind of method and apparatus for contributing to Traffic signal control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610710525.3A CN107767676A (en) | 2016-08-23 | 2016-08-23 | A kind of method and apparatus for contributing to Traffic signal control |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107767676A true CN107767676A (en) | 2018-03-06 |
Family
ID=61264272
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610710525.3A Pending CN107767676A (en) | 2016-08-23 | 2016-08-23 | A kind of method and apparatus for contributing to Traffic signal control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107767676A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110322704A (en) * | 2019-06-14 | 2019-10-11 | 东南大学 | A kind of queue length real-time estimation method based on number plate identification and GPS data |
| CN111599170A (en) * | 2020-04-13 | 2020-08-28 | 浙江工业大学 | Traffic running state classification method based on time sequence traffic network diagram |
| CN114548298A (en) * | 2022-02-25 | 2022-05-27 | 阿波罗智联(北京)科技有限公司 | Model training method, traffic information processing method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104050827A (en) * | 2014-06-06 | 2014-09-17 | 北京航空航天大学 | Traffic signal lamp automatic detection and recognition method based on visual sense |
| CN105513380A (en) * | 2015-12-29 | 2016-04-20 | 中国科学院自动化研究所 | EADP controller off-line training method, EADP controller off-line training system, EADP controller online training method and EADP controller online training system |
| CN110998686A (en) * | 2017-08-08 | 2020-04-10 | 北京嘀嘀无限科技发展有限公司 | System and method for traffic light timing |
-
2016
- 2016-08-23 CN CN201610710525.3A patent/CN107767676A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104050827A (en) * | 2014-06-06 | 2014-09-17 | 北京航空航天大学 | Traffic signal lamp automatic detection and recognition method based on visual sense |
| CN105513380A (en) * | 2015-12-29 | 2016-04-20 | 中国科学院自动化研究所 | EADP controller off-line training method, EADP controller off-line training system, EADP controller online training method and EADP controller online training system |
| CN110998686A (en) * | 2017-08-08 | 2020-04-10 | 北京嘀嘀无限科技发展有限公司 | System and method for traffic light timing |
Non-Patent Citations (3)
| Title |
|---|
| 徐洪峰 等: "机动车相位固定最小绿灯时间计算方法", 《公路交通科技》 * |
| 李秀平 等: "平面交叉路口的神经网络自学习控制方案", 《信息与控制》 * |
| 胡可云 等: "《数据挖掘理论与应用》", 30 April 2008, 清华大学出版社 等 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110322704A (en) * | 2019-06-14 | 2019-10-11 | 东南大学 | A kind of queue length real-time estimation method based on number plate identification and GPS data |
| CN110322704B (en) * | 2019-06-14 | 2021-05-04 | 东南大学 | Queuing length real-time estimation method based on number plate identification and GPS data |
| CN111599170A (en) * | 2020-04-13 | 2020-08-28 | 浙江工业大学 | Traffic running state classification method based on time sequence traffic network diagram |
| CN111599170B (en) * | 2020-04-13 | 2021-12-17 | 浙江工业大学 | Traffic running state classification method based on time sequence traffic network diagram |
| CN114548298A (en) * | 2022-02-25 | 2022-05-27 | 阿波罗智联(北京)科技有限公司 | Model training method, traffic information processing method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10332007B2 (en) | Computer-implemented system and method for generating document training sets | |
| CN106610981A (en) | Verification and upgrading method and system for road information in electronic map | |
| EP2683117A1 (en) | A method and system for network transaction monitoring using transaction flow signatures | |
| CN110473594A (en) | Pathogenic microorganism genome database and its method for building up | |
| CN107767676A (en) | A kind of method and apparatus for contributing to Traffic signal control | |
| CN107391353A (en) | Complicated software system anomaly detection method based on daily record | |
| CN110119882A (en) | A kind of platform risk monitoring and control method, apparatus, equipment and storage medium | |
| CN103942312B (en) | Bus transfer lines planning method and device | |
| CN109711424B (en) | Behavior rule acquisition method, device and equipment based on decision tree | |
| Park et al. | Large imbalance data classification based on mapreduce for traffic accident prediction | |
| Aljubayrin et al. | Finding non-dominated paths in uncertain road networks | |
| CN104965784A (en) | Automatic test method and apparatus | |
| Puts et al. | Finding errors in big data | |
| CN110059952A (en) | Vehicle insurance methods of risk assessment, device, equipment and storage medium | |
| CN106056164A (en) | Classification forecasting method based on Bayesian network | |
| CN103679269A (en) | Method and device for selecting classifier sample based on active learning | |
| CN103902798A (en) | Data preprocessing method | |
| CN109471934B (en) | Financial risk clue mining method based on Internet | |
| CN117272995B (en) | Repeated work order recommendation method and device | |
| CN111325255B (en) | Specific crowd delineating method and device, electronic equipment and storage medium | |
| CN109447171A (en) | A kind of vehicle attitude classification method based on deep learning | |
| CN103605670B (en) | A kind of method and apparatus for determining the crawl frequency of network resource point | |
| CN103294828A (en) | Verification method and verification device of data mining model dimension | |
| CN114820618A (en) | Defect detection model training method, device, equipment and storage medium | |
| CN105843941A (en) | Log checking method and device |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180306 |
|
| WD01 | Invention patent application deemed withdrawn after publication |