CN110288170A - Prediction model construction method, traffic flow forecasting method, device and electronic equipment - Google Patents
Prediction model construction method, traffic flow forecasting method, device and electronic equipment Download PDFInfo
- Publication number
- CN110288170A CN110288170A CN201910690647.4A CN201910690647A CN110288170A CN 110288170 A CN110288170 A CN 110288170A CN 201910690647 A CN201910690647 A CN 201910690647A CN 110288170 A CN110288170 A CN 110288170A
- Authority
- CN
- China
- Prior art keywords
- measured
- network
- road
- data
- road net
- 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.)
- Granted
Links
- 238000010276 construction Methods 0.000 title claims abstract description 27
- 238000013277 forecasting method Methods 0.000 title abstract description 8
- 238000012549 training Methods 0.000 claims abstract description 84
- 238000012512 characterization method Methods 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 60
- 238000000034 method Methods 0.000 claims description 41
- 230000015654 memory Effects 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000007667 floating Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000005012 migration Effects 0.000 description 3
- 238000013508 migration Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The embodiment of the invention provides a kind of prediction model construction method, traffic flow forecasting method, device and electronic equipment, the prediction model construction method includes: based on road net data to be measured and source road net data training first network；Wherein, road net data to be measured characterizes the road network structure in field to be measured and the flow information in the corresponding section of road network structure, source road net data characterize source domain road network structure and the corresponding section of road network structure multiple timing nodes flow information；The second data characteristics training based on source domain is based on the second network；Wherein, the characteristic of the road network structure of the second data characteristics characterization source domain, the dimension of the second data characteristics and the first data characteristics of the characteristic for the road network structure for characterizing field to be measured are consistent；Using the output of first network as the input of the second network, the prediction model of the magnitude of traffic flow in field to be measured is obtained.Reach the prediction model obtained with characteristic is migrated, reduces the technical effect of the forecast cost of the magnitude of traffic flow.
Description
Technical field
This application involves traffic and transport fields, in particular to a kind of prediction model construction method, traffic flow forecasting
Method, apparatus and electronic equipment.
Background technique
The prediction especially short-time traffic flow forecast of traffic flow is the basis of Urban traffic control and route guidance, Accurate Prediction road
Congestion of the road traffic within short time interval (in 10-30 minutes) has very strong realistic meaning to traffic dispersion.The magnitude of traffic flow
Generating process audient's multifactor impact, including weather conditions, festivals or holidays, peak period on and off duty, city large-scale activity, city road
Road maintenance, traffic accident etc. have the characteristics that complexity and uncertainty, and carrying out Accurate Prediction to the magnitude of traffic flow is a technology
Problem.
Currently, main using the prediction model predicting traffic flow amount based on time series training, still, this mode is to friendship
The prediction of through-current capacity is strong to the dependence of road network, and prediction model, which does not have, migrates characteristic, that is, is not suitable for new structure
Road network is low to the traffic flow forecasting accuracy of the road network of new structure.For new road network, the road is needed
The a large amount of historical data re -training prediction model of road network, cost are high.
Summary of the invention
The purpose of the present invention is to provide a kind of prediction model construction method, traffic flow forecasting method, device and electronics
Equipment is intended to improve the above deficiencies in the existing technologies.
In a first aspect, the embodiment of the invention provides a kind of prediction model construction methods, which comprises
Based on road net data to be measured and source road net data training first network；Wherein, the road net data to be measured characterizes institute
The road network structure in field to be measured and the flow information in the corresponding section of the road network structure are stated, the source road net data characterizes institute
State source domain road network structure and the corresponding section of the road network structure multiple timing nodes flow information；
The second data characteristics training based on the source domain is based on the second network；Wherein, second data characteristics table
Levy the characteristic of the road network structure of the source domain, the spy of second data characteristics and the road network structure for characterizing the field to be measured
The dimension of first data characteristics of property is consistent；
Using the output of the first network as the input of the second network, the pre- of the magnitude of traffic flow in the field to be measured is obtained
Survey model.
It is optionally, described based on road net data to be measured and source road net data training first network, comprising:
The road net data to be measured is inputted into fisrt feature network, obtains the feature of the road net data to be measured；With described
Input of the feature of road net data to be measured as first-loss function is preset when the output of the first-loss function meets first
When condition, the feature based on the road net data to be measured obtains the first data characteristics；
The source road net data is inputted into second feature network, when the source road network number of second feature network output
According to feature meet second and impose a condition when, feature based on the source road net data obtains the second data characteristics；
Based on the first-loss function, first data characteristics and second data characteristics, third loss is obtained
Function；When the third loss function meets third preset condition, stop based on the road net data to be measured training described the
One character network, using the fisrt feature network as the first network.
Optionally, described to be based on the first-loss function, first data characteristics and second data characteristics, it obtains
Obtaining third loss function includes:
Obtain the Largest Mean deviation of first data characteristics and second data characteristics；
According to the Largest Mean deviation and the first-loss function, the third loss function is obtained.
Optionally, the method also includes:
The prediction model is trained based on the road net data to be measured, the prediction model after being trained.
Second aspect, the embodiment of the invention provides a kind of method for predicting, which comprises
Obtain the road net data to be measured in field to be measured；
The road net data to be measured is inputted into above-mentioned prediction model, using the output of the prediction model as described to be measured
The magnitude of traffic flow of the section in field in subsequent time period.
The third aspect, the embodiment of the invention provides a kind of prediction model construction device, described device includes:
First training module, for based on road net data to be measured and source road net data training first network；Wherein, it is described to
It surveys road net data and characterizes the road network structure in the field to be measured and the flow information in the corresponding section of the road network structure, it is described
Source road net data characterize source domain road network structure and the corresponding section of the road network structure multiple timing nodes flow
Information；
Second training module is based on the second network for the second data characteristics training based on the source domain；Wherein, institute
State the characteristic that the second data characteristics characterizes the road network structure of the source domain, second data characteristics and the characterization neck to be measured
The dimension of first data characteristics of the characteristic of the road network structure in domain is consistent；
Module is constructed, for the input using the output of the first network as the second network, obtains the field to be measured
The magnitude of traffic flow prediction model.
Optionally, first training module is also used to:
The road net data to be measured is inputted into fisrt feature network, obtains the feature of the road net data to be measured；With described
Input of the feature of road net data to be measured as first-loss function is preset when the output of the first-loss function meets first
When condition, the feature based on the road net data to be measured obtains the first data characteristics；
The source road net data is inputted into second feature network, when the source road network number of second feature network output
According to feature meet second and impose a condition when, feature based on the source road net data obtains the second data characteristics；
Based on the first-loss function, first data characteristics and second data characteristics, third loss is obtained
Function；When the third loss function meets third preset condition, stop based on the road net data to be measured training described the
One character network, using the fisrt feature network as the first network.
Fourth aspect, the embodiment of the invention provides a kind of traffic flow forecasting device, described device includes:
Module is obtained, for obtaining the road net data to be measured in field to be measured；
Prediction module, for the road net data to be measured to be inputted above-mentioned prediction model, with the defeated of the prediction model
Out as the field to be measured section subsequent time period the magnitude of traffic flow.
5th aspect, the embodiment of the invention provides a kind of computer readable storage mediums, are stored thereon with computer journey
Sequence, when which is executed by processor the step of realization any of the above-described the method.
6th aspect the embodiment of the invention provides a kind of electronic equipment, including memory, processor and is stored in storage
On device and the computer program that can run on a processor, the processor are realized described in any of the above-described when executing described program
The step of method.
Compared with the prior art, the invention has the following advantages:
The embodiment of the invention provides a kind of prediction model construction method, traffic flow forecasting method, device and electronics to set
Standby, the prediction model construction method includes: based on road net data to be measured and source road net data training first network；Wherein, to
It surveys road net data and characterizes the road network structure in field to be measured and the flow information in the corresponding section of road network structure, source road net data table
Levy source domain road network structure and the corresponding section of road network structure multiple timing nodes flow information；Based on source domain
The training of second data characteristics is based on the second network；Wherein, the characteristic of the road network structure of the second data characteristics characterization source domain, second
The dimension of data characteristics and the first data characteristics of the characteristic for the road network structure for characterizing field to be measured is consistent；With first network
Input of the output as the second network, obtain the prediction model of the magnitude of traffic flow in field to be measured.
The first network that the road net data to be measured training of source road net data and field to be measured based on source domain obtains, first
Network meets the feature of the characteristic of the road network structure of source domain and the road network structure in field to be measured, and source domain will be met by realizing
Network migration to be suitable for field to be measured first network.Second data characteristics of the source road net data based on source domain is trained
Second network arriving, for predicted flow rate, the second network meet the prediction of the magnitude of traffic flow of source domain.
Since the dimension of the second data characteristics of first data characteristics and source domain in field to be measured is consistent, Ke Yiyong
Second network predicts the magnitude of traffic flow in field to be measured based on the first data characteristics, i.e., using the output of first network as the second network
Input, obtain the prediction model of the magnitude of traffic flow in field to be measured, the second network is met to the magnitude of traffic flow in field to be measured
Prediction, without the training sample training prediction network in a large amount of field to be measured.
Therefore, it solves in the prior art that prediction model predicting traffic flow amount is strong to the dependence of road network, predicts mould
Type, which does not have, migrates characteristic, and forecasting accuracy is low, for new road network, needs a large amount of historical data weight of the road network
New training prediction model, the high technical problem of cost have reached the prediction model obtained with characteristic is migrated, have been suitable for neck to be measured
Domain only needs a small amount of road net data in field to be measured, that is, the magnitude of traffic flow in field to be measured is effectively predicted, improves the magnitude of traffic flow
Forecasting accuracy reduces the technical effect of the forecast cost of the magnitude of traffic flow.
Other feature and advantage of the embodiment of the present invention will illustrate in subsequent specification, also, partly from specification
In become apparent, or by implement understanding of the embodiment of the present invention.The objectives and other advantages of the invention can be by institute
Specifically noted structure is achieved and obtained in specification, claims and the attached drawing write.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 shows a kind of flow chart of prediction model construction method provided in an embodiment of the present invention.
Fig. 2 shows the flow charts of another prediction model construction method provided in an embodiment of the present invention.
Fig. 3 shows the flow chart of another prediction model construction method provided in an embodiment of the present invention.
Fig. 4 shows a kind of schematic diagram of the method for trained first network provided in an embodiment of the present invention.
Fig. 5 shows a kind of structural schematic diagram of prediction model provided in an embodiment of the present invention.
Fig. 6 shows a kind of method for predicting flow chart provided in an embodiment of the present invention.
Fig. 7 shows a kind of frame structure schematic diagram of prediction model construction device 200 provided in an embodiment of the present invention.
Fig. 8 shows a kind of frame structure schematic diagram of volume forecasting device 300 provided in an embodiment of the present invention.
Fig. 9 shows the frame structure schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Icon: 200- prediction model construction device；The first training module of 210-；The second training module of 220-；230- building
Module；300- volume forecasting device；310- obtains module；320- prediction module；500- bus；501- receiver；502- processing
Device；503- transmitter；504- memory；505- bus interface.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
There is spatial complex in the road network of different zones, i.e. the road network of different zones is different, current
Traffic flow forecasting method requires a large amount of historical data pair of the road network for the road network of each specific structure
Model, which is advanced, trains, and cost is very high.For new network structure, to obtain a large amount of historical data is a problem, while at
This is also higher, even if obtaining a large amount of historical data in field to be measured, training pattern it is computationally intensive, cost is also very
Height, and because computationally intensive, real-time is poor, and the effect of traffic flow forecasting is poor.
In the prior art, the reason of the effect difference of traffic flow forecasting is: the first, to the complex space of road network according to
Lai Xingqiang, model commonality is poor, only effective to specific road network structure currently based on the prediction model of time series training, no
Characteristic is migrated with model, applied to new road network feature, needs a large amount of historical data re -training, cost is very high.The
Two, with the Nonlinear Time dynamic of road conditions change, and the emergency event influence traffic of all kinds of different times, place is had concurrently
Flow, existing prediction technique are difficult to comprehensively consider all factors, cause the result of prediction not comprehensive, accuracy is low.Third, friendship
The intrinsic difficulty of through-current capacity long-term forecast.
This application provides a kind of prediction model construction method, traffic flow forecasting method, device and electronic equipment, to
It is strong to the dependence of road network to solve prediction model predicting traffic flow amount existing in the prior art, prediction model, which does not have, to be moved
It moves characteristic, forecasting accuracy is low, for new road network, needs a large amount of historical data re -training prediction of the road network
Model, the high technical problem of cost.
Embodiment
Referring to Fig. 1, Fig. 1 shows a kind of flow chart of prediction model construction method provided in an embodiment of the present invention.In advance
It surveys model and is used for predicting traffic flow amount.As an alternative embodiment, prediction model construction method includes as shown in fig. 1
S101~S103.S101~S103 is illustrated below in conjunction with Fig. 1.
S101: based on road net data to be measured and source road net data training first network.
Wherein, road net data to be measured characterizes the road network structure in field to be measured and the flow letter in the corresponding section of road network structure
Breath, source road net data characterizes the road network structure of source domain and the corresponding section of road network structure is believed in the flow of multiple timing nodes
Breath.
S102: the second data characteristics training based on source domain is based on the second network.
Wherein, the characteristic of the road network structure of the second data characteristics characterization source domain, the second data characteristics and characterization neck to be measured
The dimension of first data characteristics of the characteristic of the road network structure in domain is consistent.
S103: using the output of first network as the input of the second network, the prediction of the magnitude of traffic flow in field to be measured is obtained
Model.
By using above scheme, the road net data to be measured of source road net data and field to be measured based on source domain is trained
The first network arrived, first network meet the feature of the characteristic of the road network structure of source domain and the road network structure in field to be measured, real
The network migration of source domain will be met to the first network for being suitable for field to be measured by having showed.First network is for extracting field to be measured
The first data characteristics, realize the migration of model extraction feature.Second data of the source road net data based on source domain are special
Second network that sign training obtains, for predicted flow rate, the second network meet the prediction of the magnitude of traffic flow of source domain.Due to
The dimension of second data characteristics of first data characteristics and source domain in field to be measured is consistent, and can be based on the second network
First data characteristics predicts the magnitude of traffic flow in field to be measured, i.e., is obtained using the output of first network as the input of the second network
The prediction model of the magnitude of traffic flow in field to be measured, the second network meet the prediction of the magnitude of traffic flow to field to be measured, without
Want the training sample training prediction network in a large amount of field to be measured.Therefore, prediction model prediction traffic in the prior art is solved
Flow is strong to the dependence of road network, and prediction model, which does not have, migrates characteristic, and forecasting accuracy is low, for new road network
Network, needs a large amount of historical data re -training prediction model of the road network, and the high technical problem of cost has reached acquisition
Prediction model, which has, migrates characteristic, is suitable for field to be measured, only needing a small amount of road net data in field to be measured, that is, be effectively predicted to
The magnitude of traffic flow in survey field improves the forecasting accuracy of the magnitude of traffic flow, reduces the forecast cost and cost of the magnitude of traffic flow.
Wherein, if only the road net data training dependent on field to be measured obtains the spy of the data for extracting field to be measured
The first network of sign, then there are following deficiencies for first network: the first, training data is few, and training sample is insufficient, what training obtained
The accuracy that first network extracts the data characteristics in field to be measured is low.Second, a large amount of training sample is obtained, it is at high cost, together
When training sample it is more, computationally intensive, cost is high.
And the field (becoming source domain herein) of the traffic flow forecasting of degree of precision has been obtained for one, have
The training sample data (history road net data, referred to as source road net data) in enough fields, in order to which the field for reducing new is (to be measured
Field) traffic flow forecasting cost, may rely on the prediction of the magnitude of traffic flow in source domain road net data training field to be measured
Model.But due to having differences between different road networks, road network feature is different, for having the road network structure with source domain
The field to be measured of different new road network structures depends only on the prediction model of source domain road net data training to field to be measured
The magnitude of traffic flow forecasting accuracy it is low, i.e., the prediction model do not have migrates characteristic.
To solve the above-mentioned problems, based on source road net data and a small amount of road net data to be measured (road network number in field to be measured
According to) train first network, then the first network obtained combines the characteristic of the road network structure of source domain and the characteristic in field to be measured,
It realizes to have from the network for being suitable for source domain and migrates characteristic and be suitable for field to be measured.
As an alternative embodiment, being obtained based on road net data to be measured and source road net data training first network
Mode with the first network for migrating characteristic specifically can specifically combine Fig. 2 refering to S101-1~S101~4 in Fig. 2
To being described below for S101-1~S101~4:
S101-1: road net data to be measured is inputted into fisrt feature network, obtains the feature of road net data to be measured；With road to be measured
Input of the feature of network data as first-loss function, when the output of first-loss function meets the first preset condition, base
The first data characteristics is obtained in the feature of road net data to be measured.
S101-2: source road net data is inputted into second feature network, when the source road net data that second feature network exports
When feature meets the second setting condition, the feature based on source road net data obtains the second data characteristics.
S101-3: being based on first-loss function, the first data characteristics and the second data characteristics, obtains third loss function.
S101-4: when third loss function meets third preset condition, stop being based on road net data fisrt feature to be measured
Network, using fisrt feature network as first network.
Wherein, do not limit S101-1 and S101-2 executes sequence, can first carry out S101-1 and execute S101-2, can also
To be performed simultaneously S101-1 and S101-2.Road net data to be measured is inputted into fisrt feature network, source road net data is defeated simultaneously
Enter second feature network, i.e., trains fisrt feature network and second feature network simultaneously.
Wherein, S101-2 specifically may is that source road net data inputting second feature network, obtain source road net data
Feature；Using the feature of source road net data as the input of the second loss function, when the output of the second loss function meets second in advance
If when condition, to obtain the second data characteristics based on the feature of source road net data.
Wherein, first preset condition can be, the output of first-loss function convergence or first-loss function
Less than the first given threshold.Second preset condition can be, the convergence of the second loss function or the second loss function
Output is less than the second given threshold.The third preset condition can be, and the convergence of third loss function or third lose letter
Several output is less than third given threshold.Wherein, the first given threshold, the second given threshold and third given threshold can phases
Deng can also be different.The fisrt feature network and second feature network can be the identical network of structure.
In order to enable first network, which has, migrates characteristic, the loss function to training first network is needed to be adjusted, such as
Described in S101-3, S101-3 specific embodiment may is that the maximum for obtaining the first data characteristics and the second data characteristics is equal
It is worth deviation；According to Largest Mean deviation and first-loss function, third loss function is obtained.Specific calculation is as follows
Formula (1) and formula (2) are described:
Wherein, FsiIndicate i-th of second data characteristicses of the road network structure of source domain, FtjIndicate the road network in field to be measured
J-th of first data characteristicses of structure, M indicate the quantity of the second data characteristics of the road network structure of source domain, and N indicates neck to be measured
The quantity of first data characteristics of the road network structure in domain, LossMMDIndicate Largest Mean deviation.
Loss=LossMMD+μ*Lossorg (2)
Wherein, LossorgIndicate that first-loss function, μ indicate that impact factor, Loss indicate third loss function.
The feature based on road net data to be measured obtains the first data characteristics, specifically, to the spy of road net data to be measured
It levies (road network structure) and carries out registration process, using the feature of the road net data to be measured after registration process as the first data characteristics.Base
The second data characteristics is obtained in the feature of source road net data, specifically, the feature to source road net data carries out registration process, with right
The feature for source road net data that treated together is as the second data characteristics.The first data characteristics and the second data after registration process
The dimension of feature is consistent.
By using above scheme, the fisrt feature of the second data characteristics, field to be measured based on source domain and the first damage
It loses function and obtains third loss function, based on third loss function and road net data to be measured training fisrt feature network, after training
Fisrt feature network receive source domain source road net data influence, while be suitable for the feature of road net data to be measured is mentioned
It takes, has and migrate characteristic, and the accuracy of feature extraction is high.The road net data to be measured in a large amount of field to be measured is not needed, just yet
The fisrt feature network suitable for field to be measured can be obtained, cost is reduced, reduces costs.
In order to improve the validity and accuracy of first network, after obtaining first network, the prediction model building
Method further include: (fine-tuning) is finely adjusted to prediction model based on road net data to be measured.It is finely adjusted (fine-
Tuning the accuracy of the magnitude of traffic flow of the prediction model output after) is high.Specifically, the mode of fine tuning is using fine-
The mode of tuning, is trained prediction model based on road net data to be measured, the prediction model after being trained, specifically:
Using road net data to be measured as training data input prediction model, the parameter of prediction model is modified, prediction model is trained,
When the output of prediction model meets four preset conditions, deconditioning prediction model, the prediction model after being trained.By
The road net data to be measured training prediction model in field to be measured improves prediction model so that prediction model is suitable for field to be measured
Predict the precision of the magnitude of traffic flow in field to be measured.
In embodiments of the present invention, when fisrt feature network is identical with second feature network, fisrt feature network and
Two character networks may include figure convolutional neural networks (Graph Convolutional Network, GCN) and connect entirely suitable
Answer network (FC-ADAPT).Using scheme convolutional neural networks output as adapt to network input, obtain fisrt feature network with
Second feature network.Wherein, figure convolutional neural networks are used to extract the feature and extraction source road net data of road net data to be measured
Feature adapts to network and is used to be aligned the feature of road net data to be measured and the feature of source road net data, to be measured after alignment
The feature of road net data is identical with the dimension of the feature of source road net data.
As an alternative embodiment, the second network includes long Memory Neural Networks (Long Short-term in short-term
Memory Networks, LSTM) and fully-connected network (FC).Using the output of long Memory Neural Networks in short-term as fully connected network
The input of network, to obtain the second network.Wherein, long Memory Neural Networks in short-term are used for based on the data characteristics for adapting to network output
The magnitude of traffic flow in (the first data characteristics or the second data characteristics) corresponding region of prediction data feature, fully-connected network is for defeated
The magnitude of traffic flow out.
In order to clearly demonstrate the prediction network of the application, as one embodiment, fisrt feature network include GCN-T and
FC-ADAPT-T, second feature network include GCN-S and FC-ADAPT-S.Second network includes LSTM and FC.That is prediction model
Including GCN-T, FC-ADAPT-T, LSTM and FC after training, i.e., using the output of the GCN-T after training as the FC- after training
The input of ADAPT-T, using the output of the FC-ADAPT-T after training as the input of the LSTM after training, with the LSTM after training
Input of the output as FC.
Based on above-mentioned embodiment, as one embodiment, Fig. 3, Fig. 4 and Fig. 5 are please referred to, Fig. 3 shows one
S301~S305 that kind prediction model construction method includes.
S301: road net data to be measured is inputted into GCN-T, GCN-T exports the feature of road net data to be measured, first-loss function
For input be road net data to be measured feature the first preset condition of output when, by FC-ADAPT-T to the road network number to be measured
According to feature carry out registration process, obtain the first data characteristics.
S302: source road net data is inputted into GCN-S, the feature of GCN-S output source road net data, the second loss function is directed to
When input is that the output of the feature of source road net data meets the second preset condition, by FC-ADAPT-S to the source road net data
Feature carries out registration process, obtains the second data characteristics.
Source road net data feature includes flow information of the corresponding section of road network structure in multiple timing nodes of source domain,
As an alternative embodiment, the second data characteristics includes multiple sections segmentum intercalaris when multiple in the road network structure of source domain
The velocity information of point, each section correspond to multiple velocity informations, i.e. each section multiple speed for corresponding to multiple timing nodes
Information.First data characteristics can be multiple sections in road network structure to be measured and believe in the speed of one or more timing node
Breath, i.e., daily section can be with only one velocity information (timing nodes).Second data characteristics and the first data characteristics
Dimension be consistent, can be the road network structure alignment of the first data characteristics and the second data characteristics.
S303: inputting LSTM for the second data characteristics, when the prediction result of LSTM output meets four preset conditions, stops
LSTM is only trained, output is connected to the prediction result that LSTM is exported by FC entirely.
Wherein, the magnitude of traffic flow of the source domain of the prediction result characterization prediction of LSTM output.
S304, using the output of the GCN-T after training as the input of the FC-ADAPT-T after training, with the FC- after training
Input of the output of ADAPT-T as the LSTM after training, using the output of the LSTM after training as the input of FC.
That is, predicting neck to be measured by being based on the first data characteristics based on the LSTM after the training after the training of the second data characteristics
The magnitude of traffic flow in domain passes through the result of the magnitude of traffic flow of FC output prediction.
S305: being trained the prediction model including GCN-T, GCN-S, FC-ADAPT-T, LSTM and FC after training,
Prediction model after being trained.
Wherein, the sequence that executes of S301, S302 and S303 are not limited, and S302 may be performed simultaneously S301 before S303
And S302.Road net data to be measured is inputted into GCN-T simultaneously and source road net data is inputted into GCN-S.
Wherein, in Fig. 4 and Fig. 5, vsi, the source road net data at i=0,2 ... ..., n+1 i-th of moment of expression, n is total
The quantity of timing node.The specific can be that in the road network structure of source domain i-th of moment speed, vtj, j=0,2 ... ...,
N+1 indicate j-th of moment road net data to be measured, the specific can be that in the road network structure in field to be measured j-th of moment speed
Degree.Ft is the first data characteristics, and Fs is the second data characteristics.
By using above scheme, road net data to be measured is inputted into GCN-T, GCN-T exports the feature of road net data to be measured,
When the first-loss function that feature based on road net data to be measured obtains meets the first preset condition, by FC-ADAPT-T to this
The feature of road net data to be measured is aligned, and the first data characteristics is obtained.Source road net data is inputted into GCN-S, GCN-S output source
The feature of road net data when the second loss function that the feature based on source road net data obtains meets the second preset condition, is passed through
FC-ADAPT-S is aligned the feature of the source road net data, obtains the second data characteristics.Second data characteristics is inputted
LSTM, when the prediction result of LSTM output meets four preset conditions, deconditioning LSTM is exported LSTM by FC pre-
It surveys result and connects output entirely.By being predicted based on the LSTM after the training after the training of the second data characteristics based on the first data characteristics
The magnitude of traffic flow in field to be measured passes through the result of the magnitude of traffic flow of FC output prediction.It is carried out to including the prediction model after training
Fine tuning, the prediction model after being finely tuned.The first, the prediction model after the fine tuning can be with each road in Accurate Prediction field to be measured
The magnitude of traffic flow of section.The second, a large amount of training data for not needing field to be measured, reduces the cost of training pattern, reduces
Cost improves the real-time of predicting traffic flow amount.Third is predicted based on source road net data and road net data to be measured training
Model, prediction model can traffic data distribution under adaptive all kinds of influence factors so that magnitude of traffic flow precision of its prediction is high.
A kind of prediction model for predicting traffic flow amount proposed based on the above embodiment, in embodiments of the present invention,
Propose a kind of method for predicting based on the prediction model.Please refer to the S401 and S402 for including in Fig. 6.
S401: the road net data to be measured in field to be measured is obtained.
S402: road net data input prediction model to be measured is existed using the output of prediction model as the section in field to be measured
The magnitude of traffic flow of subsequent time period.
Preferably, road net data to be measured is inputted into the prediction model after fine tuning, to obtain more accurately field to be measured
Section subsequent time period the magnitude of traffic flow.
Wherein, the short length of the future time can be in the time range for the 10-30 minute that this moment starts.
For S401, the road net data to be measured in field to be measured can pass through the automobile of the highway bayonet in acquisition field to be measured
RFID data, taxi real time GPS information, the GPS information of bus are to obtain the real time position of these vehicles (when including
Between, location information), and then the average speed in vehicle place section is calculated.By the average speed in each section and in advance
The road network structure in the region to be measured of building is combined, and obtains road net data to be measured.
It, can be using GPS speed calculating method, Floating Car (out for the mode of the average speed in section where obtaining vehicle
Hire a car, bus) radio frequency identification (Radio Frequency Identification, RFID) data fusion speed calculating method
Deng fluctuating big feature for Floating Car hourage, section trip of the inter-zone trip time to Floating Car provided using RFID
The row time is fitted and smoothly (herein using exponential smoothing), then is divided on each section, solves floating car data
Unusual fluctuations obtain more stable velocity information.
The magnitude of traffic flow includes the vehicular movement speed in section.The method for predicting further include: according to road
The vehicular movement speed of section and the congestion record of history obtain traffic information and are gone through according to traffic information history of forming road conditions table
History road conditions table includes the vehicular movement speed in section and the congestion record of history.
In embodiments of the present invention, the acquisition modes of the source road net data of the source domain in prediction model construction method can be with
It is to implement to obtain, is also possible to obtain from tranining database.Source road net data includes the road network structure of source domain, and each
Velocity information of the section in multiple timing nodes.The acquisition modes of velocity information can be with the average speed in the road net data of source
Acquisition modes are identical.
A kind of prediction model construction method is provided for above-described embodiment, the embodiment of the present application also correspondence provides one kind and is used for
The executing subject of above-mentioned step is executed, which can be the prediction model construction device 200 in Fig. 7.Please refer to figure
7, which includes:
First training module 210, for based on road net data to be measured and source road net data training first network；Wherein, institute
It states road net data to be measured and characterizes the road network structure in the field to be measured and the flow information in the corresponding section of the road network structure,
The road network structure of the source road net data characterization source domain and the corresponding section of the road network structure are in multiple timing nodes
Flow information.
Second training module 220 is based on the second network for the second data characteristics training based on the source domain；It is described
Second data characteristics characterizes the characteristic of the road network structure of the source domain, second data characteristics and the characterization field to be measured
The dimension of the first data characteristics of characteristic of road network structure be consistent.
Module 230 is constructed, for the input using the output of the first network as the second network, obtains the neck to be measured
The prediction model of the magnitude of traffic flow in domain.
Optionally, first training module 210 is also used to: the road net data to be measured is inputted into fisrt feature network,
Obtain the feature of the road net data to be measured；Using the feature of the road net data to be measured as the input of first-loss function, when
When the output of the first-loss function meets the first preset condition, the feature based on the road net data to be measured obtains the first number
According to feature；The source road net data is inputted into second feature network, when the source road network number of second feature network output
According to feature meet second and impose a condition when, feature based on the source road net data obtains the second data characteristics；Based on described
First-loss function, first data characteristics and second data characteristics obtain third loss function；When the third is damaged
When mistake function meets third preset condition, stop based on the road net data training to be measured fisrt feature network, with described
Fisrt feature network is as the first network.
Optionally, first training module 210 is specifically also used to: obtaining first data characteristics and second number
According to the Largest Mean deviation of feature；According to the Largest Mean deviation and the first-loss function, the third is obtained
Loss function.
Optionally, the prediction model construction device 200 further include:
Module is finely tuned, for being trained based on the road net data to be measured to the prediction model, after being trained
Prediction model.
A kind of method for predicting is provided for above-described embodiment, the also corresponding one kind that provides of the embodiment of the present application is for executing
The executing subject of above-mentioned step, the executing subject can be the volume forecasting device 300 in Fig. 8.Referring to FIG. 8, the flow
Prediction meanss include:
Module 310 is obtained, for obtaining the road net data to be measured in field to be measured.
Prediction module 320, for the road net data to be measured to be inputted prediction model described in any of the above embodiments, with described
Prediction model output as the field to be measured section subsequent time period the magnitude of traffic flow.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
The embodiment of the invention also provides a kind of electronic equipment, as shown in figure 9, include memory 504, processor 502 and
It is stored in the computer program that can be run on memory 504 and on processor 502, the processor 502 executes described program
The step of Shi Shixian method for predicting described previously and either or prediction model construction method method.
Wherein, in Fig. 9, bus architecture (is represented) with bus 500, and bus 500 may include any number of interconnection
Bus and bridge, bus 500 will include the one or more processors represented by processor 502 and what memory 504 represented deposits
The various circuits of reservoir link together.Bus 500 can also will peripheral equipment, voltage-stablizer and management circuit etc. it
Various other circuits of class link together, and these are all it is known in the art, therefore, no longer carry out further to it herein
Description.Bus interface 505 provides interface between bus 500 and receiver 501 and transmitter 503.Receiver 501 and transmitter
503 can be the same element, i.e. transceiver, provide the unit for communicating over a transmission medium with various other devices.Place
It manages device 502 and is responsible for management bus 500 and common processing, and memory 504 can be used for storage processor 502 and execute behaviour
Used data when making.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer program, the journey
Realized when sequence is executed by processor method for predicting described previously and or either prediction model construction method method step
Suddenly.
Algorithm and display are not inherently related to any particular computer, virtual system, or other device provided herein.
Various general-purpose systems can also be used together with teachings based herein.As described above, it constructs required by this kind of system
Structure be obvious.In addition, the present invention is also not directed to any particular programming language.It should be understood that can use various
Programming language realizes summary of the invention described herein, and the description done above to language-specific is to disclose this hair
Bright preferred forms.
In the instructions provided here, numerous specific details are set forth.It is to be appreciated, however, that implementation of the invention
Example can be practiced without these specific details.In some instances, well known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this specification.
Similarly, it should be understood that in order to simplify the disclosure and help to understand one or more of the various inventive aspects,
Above in the description of exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes
In example, figure or descriptions thereof.However, the disclosed method should not be interpreted as reflecting the following intention: i.e. required to protect
Shield the present invention claims features more more than feature expressly recited in each claim.More precisely, as following
Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
Thus the claims for following specific embodiment are expressly incorporated in the specific embodiment, wherein each claim itself
All as a separate embodiment of the present invention.
Those skilled in the art will understand that can be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more devices different from this embodiment.It can be the module or list in embodiment
Member or component are combined into a module or unit or component, and furthermore they can be divided into multiple submodule or subelement or
Sub-component.Other than such feature and/or at least some of process or unit exclude each other, it can use any
Combination is to all features disclosed in this specification (including adjoint claim, abstract and attached drawing) and so disclosed
All process or units of what method or apparatus are combined.Unless expressly stated otherwise, this specification is (including adjoint power
Benefit require, abstract and attached drawing) disclosed in each feature can carry out generation with an alternative feature that provides the same, equivalent, or similar purpose
It replaces.
In addition, it will be appreciated by those of skill in the art that although some embodiments in this include institute in other embodiments
Including certain features rather than other feature, but the combination of the feature of different embodiment means in the scope of the present invention
Within and form different embodiments.For example, in the following claims, embodiment claimed it is any it
One can in any combination mode come using.
Various component embodiments of the invention can be implemented in hardware, or to run on one or more processors
Software module realize, or be implemented in a combination thereof.It will be understood by those of skill in the art that can be used in practice
Microprocessor or digital signal processor (DSP) realize some or all portions in device according to an embodiment of the present invention
The some or all functions of part.The present invention is also implemented as a part or complete for executing method as described herein
The device or device program (for example, computer program and computer program product) in portion.It is such to realize program of the invention
It can store on a computer-readable medium, or may be in the form of one or more signals.Such signal can be with
It downloads from internet website, is perhaps provided on the carrier signal or is provided in any other form.
Claims (10)
1. a kind of prediction model construction method, which is characterized in that the described method includes:
Based on road net data to be measured and source road net data training first network；Wherein, the road net data to be measured characterizes neck to be measured
The road network structure in domain and the flow information in the corresponding section of the road network structure, the road of the source road net data characterization source domain
The flow information of web frame and the corresponding section of the road network structure in multiple timing nodes；
The second data characteristics training based on the source domain is based on the second network；Wherein, second data characteristics characterizes institute
State the characteristic of the road network structure of source domain, second data characteristics and the characteristic for the road network structure for characterizing the field to be measured
The dimension of first data characteristics is consistent；
Using the output of the first network as the input of the second network, the prediction mould of the magnitude of traffic flow in the field to be measured is obtained
Type.
2. the method according to claim 1, wherein described based on road net data to be measured and the training of source road net data
First network, comprising:
The road net data to be measured is inputted into fisrt feature network, obtains the feature of the road net data to be measured；With described to be measured
Input of the feature of road net data as first-loss function, when the output of the first-loss function meets the first preset condition
When, the feature based on the road net data to be measured obtains the first data characteristics；
The source road net data is inputted into second feature network, when the source road net data that the second feature network exports
When feature meets the second setting condition, the feature based on the source road net data obtains the second data characteristics；
Based on the first-loss function, first data characteristics and second data characteristics, third loss function is obtained；
When the third loss function meets third preset condition, stop based on the road net data training to be measured fisrt feature
Network, using the fisrt feature network as the first network.
3. according to the method described in claim 2, it is characterized in that, described be based on the first-loss function, first number
According to feature and second data characteristics, obtaining third loss function includes:
Obtain the Largest Mean deviation of first data characteristics and second data characteristics；
According to the Largest Mean deviation and the first-loss function, the third loss function is obtained.
4. method according to claim 1-3, which is characterized in that the method also includes:
The prediction model is trained based on the road net data to be measured, the prediction model after being trained.
5. a kind of method for predicting, which is characterized in that the described method includes:
Obtain the road net data to be measured in field to be measured；
The road net data to be measured is inputted into the described in any item prediction models of claim 1-4, with the defeated of the prediction model
Out as the field to be measured section subsequent time period the magnitude of traffic flow.
6. a kind of prediction model construction device, which is characterized in that described device includes:
First training module, for based on road net data to be measured and source road net data training first network；Wherein, the road to be measured
Network data characterizes the road network structure in field to be measured and the flow information in the corresponding section of the road network structure, the source road network number
According to characterization source domain road network structure and the corresponding section of the road network structure multiple timing nodes flow information；
Second training module is based on the second network for the second data characteristics training based on the source domain；Wherein, described
Two data characteristicses characterize the characteristic of the road network structure of the source domain, second data characteristics and the characterization field to be measured
The dimension of first data characteristics of the characteristic of road network structure is consistent；
Module is constructed, for the input using the output of the first network as the second network, obtains the friendship in the field to be measured
The prediction model of through-current capacity.
7. device according to claim 6, which is characterized in that first training module is also used to:
The road net data to be measured is inputted into fisrt feature network, obtains the feature of the road net data to be measured；With described to be measured
Input of the feature of road net data as first-loss function, when the output of the first-loss function meets the first preset condition
When, the feature based on the road net data to be measured obtains the first data characteristics；
The source road net data is inputted into second feature network, when the source road net data that the second feature network exports
When feature meets the second setting condition, the feature based on the source road net data obtains the second data characteristics；
Based on the first-loss function, first data characteristics and second data characteristics, third loss function is obtained；
When the third loss function meets third preset condition, stop based on the road net data training to be measured fisrt feature
Network, using the fisrt feature network as the first network.
8. a kind of traffic flow forecasting device, which is characterized in that described device includes:
Module is obtained, for obtaining the road net data to be measured in field to be measured；
Prediction module, for the road net data to be measured to be inputted the described in any item prediction models of claim 1-4, with described
Prediction model output as the field to be measured section subsequent time period the magnitude of traffic flow.
9. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor
The step of any one of claim 1-5 the method is realized when row.
10. a kind of electronic equipment, which is characterized in that on a memory and can be in processor including memory, processor and storage
The computer program of upper operation, the processor realize the step of any one of claim 1-5 the method when executing described program
Suddenly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910690647.4A CN110288170B (en) | 2019-07-29 | 2019-07-29 | Prediction model construction method, traffic flow prediction device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910690647.4A CN110288170B (en) | 2019-07-29 | 2019-07-29 | Prediction model construction method, traffic flow prediction device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110288170A true CN110288170A (en) | 2019-09-27 |
| CN110288170B CN110288170B (en) | 2020-04-14 |
Family
ID=68024147
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910690647.4A Active CN110288170B (en) | 2019-07-29 | 2019-07-29 | Prediction model construction method, traffic flow prediction device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110288170B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110956808A (en) * | 2019-12-23 | 2020-04-03 | 北京交通大学 | Heavy truck traffic flow prediction method based on non-full-sample positioning data |
| CN111145289A (en) * | 2019-12-30 | 2020-05-12 | 北京爱康宜诚医疗器材有限公司 | Extraction method and device of pelvis three-dimensional data |
| CN113256985A (en) * | 2021-06-28 | 2021-08-13 | 智道网联科技(北京)有限公司 | Traffic congestion prediction method and device and electronic equipment |
| CN113496305A (en) * | 2020-04-03 | 2021-10-12 | 浙江大学 | Data processing method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967532A (en) * | 2017-10-30 | 2018-04-27 | 厦门大学 | The Forecast of Urban Traffic Flow Forecasting Methodology of integration region vigor |
| CN109740785A (en) * | 2018-10-22 | 2019-05-10 | 北京师范大学 | The method of node state prediction based on figure convolutional neural networks |
| CN109887282A (en) * | 2019-03-05 | 2019-06-14 | 中南大学 | A kind of road network traffic flow prediction technique based on level timing diagram convolutional network |
-
2019
- 2019-07-29 CN CN201910690647.4A patent/CN110288170B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967532A (en) * | 2017-10-30 | 2018-04-27 | 厦门大学 | The Forecast of Urban Traffic Flow Forecasting Methodology of integration region vigor |
| CN109740785A (en) * | 2018-10-22 | 2019-05-10 | 北京师范大学 | The method of node state prediction based on figure convolutional neural networks |
| CN109887282A (en) * | 2019-03-05 | 2019-06-14 | 中南大学 | A kind of road network traffic flow prediction technique based on level timing diagram convolutional network |
Non-Patent Citations (3)
| Title |
|---|
| RUI FU 等: "Using LSTM and GRU Neural Network Methods for Traffic Flow Prediction", 《31ST YOUTH ACADEMIC ANNUAL CONFERENCE OF CHINESE ASSOCIATION OF AUTOMATION》 * |
| 乔松林 等: "基于深度学习的短时交通流量预测", 《青岛大学学报（自然科学版）》 * |
| 邓烜堃 等: "基于深度学习的交通流量预测研究", 《计算机工程与应用》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110956808A (en) * | 2019-12-23 | 2020-04-03 | 北京交通大学 | Heavy truck traffic flow prediction method based on non-full-sample positioning data |
| CN111145289A (en) * | 2019-12-30 | 2020-05-12 | 北京爱康宜诚医疗器材有限公司 | Extraction method and device of pelvis three-dimensional data |
| CN113496305A (en) * | 2020-04-03 | 2021-10-12 | 浙江大学 | Data processing method and device |
| CN113256985A (en) * | 2021-06-28 | 2021-08-13 | 智道网联科技(北京)有限公司 | Traffic congestion prediction method and device and electronic equipment |
| CN113256985B (en) * | 2021-06-28 | 2021-09-17 | 智道网联科技(北京)有限公司 | Traffic congestion prediction method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110288170B (en) | 2020-04-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110288170A (en) | Prediction model construction method, traffic flow forecasting method, device and electronic equipment | |
| Fan et al. | Digital twin empowered mobile edge computing for intelligent vehicular lane-changing | |
| Eberlein et al. | The holding problem with real–time information available | |
| Yang et al. | Short-term traffic prediction for edge computing-enhanced autonomous and connected cars | |
| Wang et al. | A distributed dynamic route guidance approach based on short-term forecasts in cooperative infrastructure-vehicle systems | |
| CN103440422A (en) | Bus behind-schedule recovering method based on arrival time predication with time window | |
| EP2610782A1 (en) | Method for planning the route of a vehicle | |
| CN103177561A (en) | Method and system for generating bus real-time traffic status | |
| Roopa et al. | Dynamic management of traffic signals through social IoT | |
| CN101807348B (en) | Dynamic network navigation system and method | |
| Liu et al. | A distributed and privacy-aware speed advisory system for optimizing conventional and electric vehicle networks | |
| CN117546218A (en) | Distributed multitasking machine learning for traffic prediction | |
| Lin et al. | New transit signal priority scheme for intersections with nearby bus rapid transit median stations | |
| Araldo et al. | Implementation & policy applications of AMOD in multi-modal activity-driven agent-based urban simulator simmobility | |
| CN104680821A (en) | Intelligent traffic system and navigation method thereof | |
| Fujimoto et al. | Dynamic data driven application simulation of surface transportation systems | |
| CN103839424B (en) | A kind of taxi empty navigational system and method based on recruiting car expectation value | |
| Zhao et al. | Dynamic service entity placement for latency sensitive applications in transportation systems | |
| CN103236180B (en) | A kind of Vehicular information interactive interface method | |
| Stancel et al. | Fleet Management System for Truck Platoons-Generating an Optimum Route in Terms of Fuel Consumption | |
| Si et al. | Data-based sorting algorithm for variable message sign location: Case study of Beijing | |
| Marchetta et al. | S 2-move: Smart and social move | |
| Feng et al. | Optimising departure intervals for multiple bus lines with a multi‐objective model | |
| CN110942622A (en) | Parking lot planning method based on real-time operation big data of parking lot | |
| Reyes et al. | On‐line passenger estimation in a metro system using particle filter |
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 |