CN109785629A - A kind of short-term traffic flow forecast method - Google Patents
A kind of short-term traffic flow forecast method Download PDFInfo
- Publication number
- CN109785629A CN109785629A CN201910153156.6A CN201910153156A CN109785629A CN 109785629 A CN109785629 A CN 109785629A CN 201910153156 A CN201910153156 A CN 201910153156A CN 109785629 A CN109785629 A CN 109785629A
- Authority
- CN
- China
- Prior art keywords
- data
- network
- seq2seq
- model
- prediction model
- 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
Landscapes
- Traffic Control Systems (AREA)
Abstract
The embodiment of the invention provides a kind of short-term traffic flow forecast method, method includes: the historical data acquired according to path sensor in road network and road network topology structured data, and building is used for the data set of prediction model；The data set is normalized, and checks the missing values in the data set in measurement data；Using fully-connected network and seq2seq network struction neural network as short-term traffic flow forecast model, and prediction model is trained using the data set after normalized, chooses the smallest model of error neural network prediction model the most final；Using the final neural network prediction model, the real-time traffic data of multiple path sensors acquisition are handled, predicted flow rate data are obtained.The invention discloses a kind of short-term traffic flow forecast methods for having merged fully-connected network and seq2seq network in deep learning, improve precision of prediction, provide a kind of new thinking for Forecast of Urban Traffic Flow forecasting problem.
Description
Technical field
The present invention relates to technical field of transportation more particularly to a kind of short-term traffic flow forecast methods.
Background technique
In recent years, as society develops rapidly, urbanization process continues to promote, and population is resided in city and car ownership flies
Speed increases.This brings great challenge to urban transportation infrastructure, and out-of-date traffic programme is difficult the friendship for adapting to be skyrocketed through
Logical demand, causes urban transport problems to become increasingly complex.That is, in order to improve traffic circulation efficiency, only by increase
Road quantity and infrastructure be it is far from being enough, will also by improve traffic control system operational efficiency, to give full play to
The function of existing road.
Therefore intelligent transportation system is come into being, and it is desirable to efficiently solve above-mentioned problem by intelligent transportation system.
System for traffic guiding and traffic control system are the important components in intelligent transportation system, and reasonable traffic state judging
And accurately traffic flow forecasting can be improved the accuracy of traffic guidance and the validity of traffic control, in intelligent transportation system
In have critical role, be the key that complete induction and control and guarantee.
Defect in the prior art are as follows: 1) existing method mostly only focus on single detector collected traffic information, and
It has ignored the topological structure of road network, has ignored between adjacent detector and correlation in this way, and model can not be verified true
Physical circuit online operational efficiency and accuracy；2) existing method does not mostly account for the variation tendency of flow, without benefit
With the correlation of surrounding time point in time series data, so that the case where model is easily trapped into over-fitting, model generalization ability
It is bad；3) existing method is not high to the utilization rate of initial data, needs a large amount of data to reach perfect precision, improves pair
The requirement of equipment.
Summary of the invention
The embodiment provides a kind of short-term traffic flow forecast methods, to overcome the deficiencies of existing technologies.
To achieve the goals above, this invention takes following technical solutions.
A kind of short-term traffic flow forecast method, which comprises
The historical data and road network topology structured data acquired according to path sensor in road network, building are used for prediction model
Data set；
The data set is normalized, and checks the missing values in the data set in measurement data；
Using fully-connected network and seq2seq network struction neural network as short-term traffic flow forecast model, and benefit
Prediction model is trained with the data set after normalized, chooses the smallest model of error neural network the most final
Prediction model；
Using the final neural network prediction model, the real-time traffic data of multiple path sensors acquisition are carried out
Processing, obtains predicted flow rate data.
Preferably, the history data set of the path sensor acquisition includes time data and traffic data, historical data
Collect XglobalWhether middle time data include: sampling instant time t and are festivals or holidays tholidays；Traffic data includes: current detection
Section detector number εc, current detection section number of track-lines εl, through street number k, current detection section where current detector
Magnitude of traffic flow V in unit periodcur(t) (veh/h), average speed v in current detection section unit periodcur(t) (km/h), when
The cart quantity c passed through in preceding detection section unit periodcur(t) (veh), averagely occupy in current detection section unit period
Rate ocur(t) (%)；
The road network topology structured data includes: magnitude of traffic flow V in the connected detection section unit period in upstreampre(t)
(veh), upstream, which is connected, detects average speed v in section unit periodpre(t) (km/h), upstream, which are connected, detects section unit period
The cart quantity c inside passed throughpre(t) (veh), upstream, which are connected, detects average occupancy o in section unit periodpre(t) (%).
Preferably, the method further include:
The sampling time interval of the history data set is extended for 15min, changes in flow rate sums up, average speed
Degree and occupation rate take mean value, and the data obtained is divided by through street number k, as trend data collectionWherein i
∈k；
Convert trend data collection to the format for meeting seq2seq network inputs.
Preferably, the sampling time interval of the history data set is 5min.
Preferably, described to convert trend data collection to the format for meeting seq2seq network inputs, i.e., every 5 samplings when
It carves acquisition data and merges into an input sample, the label of each sample is the true magnitude of traffic flow of rear 5 sampling instants.
Preferably, the data set is normalized, and checks the missing values in the measurement data, comprising:
Each field x that data are concentratediIt is normalized, has respectivelyWherein xi′It is each
Result after field normalization；
It checks the missing values and exceptional value in each field, periphery section sensor number is used to the missing values of measurement data
According to average value be filled, for be in measurement data 0 exceptional value using the front and back occurred extremely before and after same detector
Moment measurement mean value is replaced.
Preferably, described pre- as Short-Term Traffic Flow using fully-connected network and seq2seq network struction neural network
Model is surveyed, and prediction model is trained using the data set after normalized, chooses the smallest model of error as most
Whole neural network prediction model, comprising:
Prediction model is constructed, the prediction model includes the full articulamentum that 4 layers of activation primitive are ReLU, k 5 layers
The full articulamentum of seq2seq network and 1 layer line；
Use the trend data collection for meeting seq2seq network inputs formatK seq2seq net is respectively trained
Network；
The objective function of seq2seq network is mean square error function MSE, the calculation of MSE error are as follows:
Wherein, y is real traffic,For predicted flow rate, n is total number of samples；
The encoder section for the seq2seq model that caching training is completed；
By the history data set XglobalInput full articulamentum, while by corresponding trend data collectionIn
Data input trained seq2seq model, and the output of full articulamentum and the input of seq2seq models encoder part are closed
And；
By the full articulamentum of Data In-Line after merging, and using MSE as the objective function of entire prediction model, training
The parameter of full articulamentum and linear full linking layer, training are completed to obtain final neural network prediction model.
Preferably, using the final neural network prediction model, to the real-time traffic of multiple path sensors acquisition
Data are handled, and predicted flow rate data are obtained, comprising:
For the data on flows acquired in real time, building meets the data of seq2seq structure input；
Initial data and building meet seq2seq structure input the data on flows acquired in real time input simultaneously described in short-term
Magnitude of traffic flow neural network prediction model obtains final prediction result.
As can be seen from the technical scheme provided by the above-mentioned embodiment of the present invention, the embodiment of the present invention is directed to traffic flow quantitative change
Change has the characteristics that non-linear and periodic, proposes and a kind of has merged (sampling in short-term for fully-connected network and seq2seq network
Be spaced in half an hour or less) magnitude of traffic flow neural network prediction model, merge deep learning in fully-connected network to global feature
Extractability guaranteeing the case where information content is not lost with seq2seq network to the learning ability of time series variation trend
Under, so that the more periodic features of model learning, improve precision of prediction, provided for Forecast of Urban Traffic Flow forecasting problem a kind of new
Thinking.
The additional aspect of the present invention and advantage will be set forth in part in the description, these will become from the following description
Obviously, or practice through the invention is recognized.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, required use in being described below to embodiment
Attached drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill of field, without creative efforts, it can also be obtained according to these attached drawings others
Attached drawing.
Fig. 1 is a kind of flow chart of short-term traffic flow forecast method provided in an embodiment of the present invention；
Fig. 2 is the short-term traffic flow forecast of fusion fully-connected network and seq2seq2 network provided in an embodiment of the present invention
Flow chart；
Fig. 3 is seq2seq schematic network structure provided in an embodiment of the present invention；
Fig. 4 is short-term traffic flow forecast schematic network structure provided in an embodiment of the present invention.
Specific embodiment
Embodiments of the present invention are described below in detail, the example of the embodiment is shown in the accompanying drawings, wherein from beginning
Same or similar element or element with the same or similar functions are indicated to same or similar label eventually.Below by ginseng
The embodiment for examining attached drawing description is exemplary, and for explaining only the invention, and is not construed as limiting the claims.
Those skilled in the art of the present technique are appreciated that unless expressly stated, singular " one " used herein, " one
It is a ", " described " and "the" may also comprise plural form.It is to be further understood that being arranged used in specification of the invention
Diction " comprising " refer to that there are the feature, integer, step, operation, element and/or component, but it is not excluded that in the presence of or addition
Other one or more features, integer, step, operation, element, component and/or their group.It should be understood that when we claim member
Part is " connected " or when " coupled " to another element, it can be directly connected or coupled to other elements, or there may also be
Intermediary element.In addition, " connection " used herein or " coupling " may include being wirelessly connected or coupling.Wording used herein
"and/or" includes one or more associated any cells for listing item and all combinations.
Those skilled in the art of the present technique are appreciated that unless otherwise defined, all terms used herein (including technology art
Language and scientific term) there is meaning identical with the general understanding of those of ordinary skill in fields of the present invention.Should also
Understand, those terms such as defined in the general dictionary, which should be understood that, to be had and the meaning in the context of the prior art
The consistent meaning of justice, and unless defined as here, it will not be explained in an idealized or overly formal meaning.
In order to facilitate understanding of embodiments of the present invention, it is done by taking several specific embodiments as an example below in conjunction with attached drawing further
Explanation, and each embodiment does not constitute the restriction to the embodiment of the present invention.
The embodiment of the invention provides a kind of process flow of short-term traffic flow forecast method as shown in Figure 1, including such as
Lower processing step:
The historical data and road network topology structured data acquired according to path sensor in road network, building are used for prediction model
Data set；
Data set is normalized, and checks the missing values in data set in measurement data；
Using fully-connected network and seq2seq network struction neural network as short-term traffic flow forecast model, and benefit
Prediction model is trained with the data set after normalized, chooses the smallest model of error neural network the most final
Prediction model；
Using final neural network prediction model, at the real-time traffic data of multiple path sensors acquisition
Reason, obtains predicted flow rate data.
As shown in Fig. 2, the Short-Term Traffic Flow of fusion fully-connected network and seq2seq network provided in an embodiment of the present invention
The algorithm flow chart of prediction technique, forecasting system include the predicted portions of the training part and real time traffic data of model.?
In training part, trend data collection is constructed by historical traffic data first and preliminary treatment is carried out to data, is adopted by increasing
The variation tendency of the mode retention time sequence of sample time, and data are simply divided by through street number, it mentions
The accuracy of high single model；Secondly, seq2seq net corresponding with through street number is respectively trained in usage trend data set
Network, and retain the encoder section of network；Next, historical traffic data collection is aligned with trend data collection, i.e., each history
Traffic data sample corresponds to trend data and concentrates last timing node, the two data sets are separately input to fully-connected network
In trained seq2seq network, it is input in the full articulamentum that activation primitive is linear function after obtained result combination
The parameter of all fully-connected networks of training, reaches satisfied precision.In prediction module, using above-mentioned training pattern to reality
When traffic flow data handled, and predict the magnitude of traffic flow of future time point.
Specifically includes the following steps:
(1) historical data and road network topology structure feature acquired according to path sensor in road network is constructed for predicting
The data set of model；
Path sensor acquisition history data set includes time data and traffic data, wherein history data set XglobalIn
Whether time data include: sampling instant time t and are festivals or holidays tholidays；Traffic data includes: the detection of current detection section
Device number εc；The number of track-lines ε of current detection sectionl；Through street number k where current detector；Current detection section unit period
Interior magnitude of traffic flow Vcur(t)(veh/h)；Average speed v in current detection section unit periodcur(t)(km/h)；Current detection is disconnected
The cart quantity c passed through in the unit period of facecur(t) average occupancy o in (veh) and current detection section unit periodcur(t)
(%)；Wherein the sampling time interval of history data set is 5min；
Road network topology structured data includes: magnitude of traffic flow V in the connected detection section unit period in upstreampre(t)(veh)；On
Trip, which is connected, detects average speed v in section unit periodpre(t)(km/h)；Pass through in the connected detection section unit period in upstream
Cart quantity cpre(t)(veh)；Upstream, which is connected, detects average occupancy o in section unit periodpre(t) (%)；
The above topology information can concentrate retrieval to obtain in historical data.
In order to obtain the Long-term change trend information of the magnitude of traffic flow, the sampling time interval of history data set is extended for 15min,
Changes in flow rate sums up, and average speed and occupation rate take mean value；And divide data by through street number k, make
For trend data collectionWherein i ∈ k.
Convert trend data collection to the format for meeting seq2seq network inputs, i.e., every 5 sampling instants obtain data and close
And be an input sample, the label of each sample is the true magnitude of traffic flow of rear 5 sampling instants.
Trend data collection format is as shown in table 1:
1 trend data collection format of table
Label corresponding with trend data collection is as shown in table 2:
2 trend data collection label of table
In table 1, table 2, T1 to T9 sample time order is incremented by, and the sampling interval is 15min.
(2) data set is normalized, and checks the missing values in data；
Each field x that data are concentratedi(such as flow, average speed) is normalized respectively, hasWherein xi′For the result after the normalization of each field；
It checks the missing values and exceptional value in each field, periphery section sensor number is used to the missing values of measurement data
According to average value be filled, for be in measurement data 0 exceptional value using the front and back occurred extremely before and after same detector
Moment measurement mean value is replaced.
(3) using fully-connected network and seq2seq network struction neural network as short-term traffic flow forecast model, and
Prediction model is trained using the data obtained in step (2), chooses the smallest model of error prediction mould the most final
Type, specifically includes the following steps:
Prediction model is constructed, which includes the full articulamentum that 4 layers of activation primitive are ReLU (line rectification function), k a 5
The seq2seq network of layer and the full articulamentum of 1 layer line；
Using being converted into the trend data collection for meeting the format of seq2seq network inputsI ∈ k is respectively trained k
Seq2seq network；
The objective function of seq2seq network is mean square error function MSE:
Wherein, y is real traffic,For predicted flow rate, n is total number of samples；
The encoder section for the seq2seq model that caching training is completed；
Fig. 3 is complete seq2seq schematic network structure, uses input LSTM cell number for 5, after the completion of trained
Save encoder section parameter.
Fig. 4 is short-term traffic flow forecast network structure schematic diagram, as shown in figure 4,
By history data set XglobalInput full articulamentum, while by corresponding trend data collectionIn data
Seq2seq model is inputted, and full articulamentum is exported and is merged with the input of seq2seq models encoder part；
By the full articulamentum of Data In-Line after merging, and using MSE as the objective function of entire prediction model, training
The parameter of full articulamentum and linear full linking layer.
(4) using the neural network model trained, at the real-time traffic data of multiple path sensors acquisition
Reason, obtains predicted flow rate data, i.e., for the data acquired in real time, building meets the data of seq2seq structure input, will be original
Data and prediction model obtained by building the data obtained simultaneously input step (2), obtain final prediction result.
Those skilled in the art will be understood that above-mentioned lifted sampling time interval and input LSTM cell number is only
The technical solution of the embodiment of the present invention is better described, rather than to the restriction that the embodiment of the present invention is made.It is any real according to user
Border situation adjustment time interval and input LSTM cell number, are all contained in the range of the embodiment of the present invention.
In conclusion the embodiment of the present invention passes through first in conjunction with path sensor acquisition Information Center road network topology structure building
Short-term traffic flow forecast model history data set；Secondly trend data collection is constructed according to history data set, wherein trend data
Collection meets the input format of seq2seq network；Following usage trend data set is trained seq2seq network and retains net
Historical traffic data collection is finally aligned by the encoder section of network with trend data collection, what training was connected with seq2seq encoder
Fully-connected network.The short of fully-connected network and seq2seq network in deep learning has been merged the embodiment of the invention discloses a kind of
When traffic flow forecasting method, merge in deep learning that fully-connected network is to the extractability of global feature, with seq2seq network
To the learning ability of time series variation trend, in the case where guarantor's global characteristics are not lost, so that multiple more targeted
The more periodic features of seq2seq e-learning improve precision of prediction, provide for Forecast of Urban Traffic Flow forecasting problem a kind of new
Thinking.
Those of ordinary skill in the art will appreciate that: attached drawing is the schematic diagram of one embodiment, module in attached drawing or
Process is not necessarily implemented necessary to the present invention.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device or
For system embodiment, since it is substantially similar to the method embodiment, so describing fairly simple, related place is referring to method
The part of embodiment illustrates.Apparatus and system embodiment described above is only schematical, wherein the conduct
The unit of separate part description may or may not be physically separated, component shown as a unit can be or
Person may not be physical unit, it can and it is in one place, or may be distributed over multiple network units.It can root
According to actual need that some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Ordinary skill
Personnel can understand and implement without creative efforts.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with scope of protection of the claims
Subject to.
Claims (8)
1. a kind of short-term traffic flow forecast method, which is characterized in that the described method includes:
The historical data and road network topology structured data acquired according to path sensor in road network, building are used for the number of prediction model
According to collection；
The data set is normalized, and checks the missing values in the data set in measurement data；
Using fully-connected network and seq2seq network struction neural network as short-term traffic flow forecast model, and utilizes and return
One changes that treated that data set is trained prediction model, chooses the smallest model of error neural network prediction the most final
Model；
Using the final neural network prediction model, at the real-time traffic data of multiple path sensors acquisition
Reason, obtains predicted flow rate data.
2. prediction technique according to claim 1, which is characterized in that the history data set packet of the path sensor acquisition
Include time data and traffic data, history data set XglobalWhether middle time data include: sampling instant time t and are that section is false
Day tholidays；Traffic data includes: current detection section detector number εc, current detection section number of track-lines εl, current inspection
Magnitude of traffic flow V in survey device place through street number k, current detection section unit periodcur(t) (veh/h), current detection section
Average speed v in unit periodcur(t) (km/h), the cart quantity c passed through in current detection section unit periodcur(t)
(veh), average occupancy o in current detection section unit periodcur(t) (%)；
The road network topology structured data includes: magnitude of traffic flow V in the connected detection section unit period in upstreampre(t) (veh), on
Trip, which is connected, detects average speed v in section unit periodpre(t) (km/h), upstream are connected in detection section unit period and pass through
Cart quantity cpre(t) (veh), upstream, which are connected, detects average occupancy o in section unit periodpre(t) (%).
3. prediction technique according to claim 2, which is characterized in that the method further include:
The sampling time interval of the history data set is extended for 15min, changes in flow rate sums up, average speed and
Occupation rate takes mean value, and the data obtained is divided by through street number k, as trend data collectionWherein i ∈ k；
Convert trend data collection to the format for meeting seq2seq network inputs.
4. prediction technique according to claim 2, which is characterized in that the sampling time interval of the history data set is
5min。
5. prediction technique according to claim 3, which is characterized in that described convert trend data collection to meets
The format of seq2seq network inputs, i.e., every 5 sampling instants obtain data and merge into an input sample, the mark of each sample
Label are the true magnitude of traffic flow of rear 5 sampling instants.
6. prediction technique according to claim 1, which is characterized in that be normalized, and examine to the data set
Look into the missing values in the measurement data, comprising:
Each field x that data are concentratediIt is normalized, has respectivelyWherein xi' it is each field
Result after normalization；
It checks the missing values and exceptional value in each field, periphery section sensing data is used to the missing values of measurement data
Average value is filled, and the exceptional value in measurement data being 0 uses the front and back moment occurred extremely before and after same detector
Measurement mean value is replaced.
7. prediction technique according to claim 1, which is characterized in that described to utilize fully-connected network and seq2seq network
Neural network is constructed as short-term traffic flow forecast model, and prediction model is carried out using the data set after normalized
Training chooses the smallest model of error as final neural network prediction model, comprising:
Prediction model is constructed, the prediction model includes the full articulamentum that 4 layers of activation primitive are ReLU, k 5 layers of seq2seq
The full articulamentum of network and 1 layer line；
Use the trend data collection for meeting seq2seq network inputs formatK seq2seq network is respectively trained；
The objective function of seq2seq network is mean square error function MSE, the calculation of MSE error are as follows:
Wherein, y is real traffic,For predicted flow rate, n is total number of samples；
The encoder section for the seq2seq model that caching training is completed；
By the history data set XglobalInput full articulamentum, while by corresponding trend data collectionIn data
Trained seq2seq model is inputted, and full articulamentum is exported and is merged with the input of seq2seq models encoder part；
By the full articulamentum of Data In-Line after merging, and using MSE as the objective function of entire prediction model, training connects entirely
The parameter of layer and linear full linking layer is connect, training is completed to obtain final neural network prediction model.
8. prediction technique according to claim 1, which is characterized in that the final neural network prediction model is utilized,
The real-time traffic data of multiple path sensors acquisition are handled, predicted flow rate data are obtained, comprising:
For the data on flows acquired in real time, building meets the data of seq2seq structure input；
The data on flows acquired in real time that initial data and building meet the input of seq2seq structure inputs the traffic in short-term simultaneously
Flow neural network prediction model obtains final prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910153156.6A CN109785629A (en) | 2019-02-28 | 2019-02-28 | A kind of short-term traffic flow forecast method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910153156.6A CN109785629A (en) | 2019-02-28 | 2019-02-28 | A kind of short-term traffic flow forecast method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109785629A true CN109785629A (en) | 2019-05-21 |
Family
ID=66486044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910153156.6A Pending CN109785629A (en) | 2019-02-28 | 2019-02-28 | A kind of short-term traffic flow forecast method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109785629A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276949A (en) * | 2019-06-20 | 2019-09-24 | 苏州大学 | Forecasting Approach for Short-term Traffic Flow based on machine learning |
| CN110288205A (en) * | 2019-06-04 | 2019-09-27 | 北京世纪高通科技有限公司 | Traffic impact analysis method and device |
| CN110428613A (en) * | 2019-07-09 | 2019-11-08 | 中山大学 | A kind of intelligent transportation trend prediction method of machine learning |
| CN110474815A (en) * | 2019-09-23 | 2019-11-19 | 北京达佳互联信息技术有限公司 | Bandwidth prediction method, apparatus, electronic equipment and storage medium |
| CN110517485A (en) * | 2019-08-09 | 2019-11-29 | 大连理工大学 | A kind of Short-time Traffic Flow Forecasting Methods based on Time segments division |
| CN110717627A (en) * | 2019-09-29 | 2020-01-21 | 浙江大学 | Full traffic prediction method based on dual graph framework |
| CN111292534A (en) * | 2020-02-13 | 2020-06-16 | 北京工业大学 | Traffic state estimation method based on clustering and deep sequence learning |
| CN111815046A (en) * | 2020-07-06 | 2020-10-23 | 北京交通大学 | Traffic flow prediction method based on deep learning |
| CN112182961A (en) * | 2020-09-23 | 2021-01-05 | 中国南方电网有限责任公司超高压输电公司 | Large-scale fading modeling prediction method for wireless network channel of converter station |
| CN112434260A (en) * | 2020-10-21 | 2021-03-02 | 北京千方科技股份有限公司 | Road traffic state detection method and device, storage medium and terminal |
| CN112489426A (en) * | 2020-11-26 | 2021-03-12 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN112884190A (en) * | 2019-11-29 | 2021-06-01 | 杭州海康威视数字技术股份有限公司 | Flow prediction method and device |
| CN114373299A (en) * | 2021-12-27 | 2022-04-19 | 宁波大学 | Channel traffic flow prediction method based on AIS data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251625A (en) * | 2016-08-18 | 2016-12-21 | 上海交通大学 | Three-dimensional urban road network global state Forecasting Methodology under big data environment |
| CN107103758A (en) * | 2017-06-08 | 2017-08-29 | 厦门大学 | A kind of city area-traffic method for predicting based on deep learning |
| CN108986470A (en) * | 2018-08-20 | 2018-12-11 | 华南理工大学 | The Travel Time Estimation Method of particle swarm algorithm optimization LSTM neural network |
| CN109243172A (en) * | 2018-07-25 | 2019-01-18 | 华南理工大学 | Traffic flow forecasting method based on genetic algorithm optimization LSTM neural network |
-
2019
- 2019-02-28 CN CN201910153156.6A patent/CN109785629A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251625A (en) * | 2016-08-18 | 2016-12-21 | 上海交通大学 | Three-dimensional urban road network global state Forecasting Methodology under big data environment |
| CN107103758A (en) * | 2017-06-08 | 2017-08-29 | 厦门大学 | A kind of city area-traffic method for predicting based on deep learning |
| CN109243172A (en) * | 2018-07-25 | 2019-01-18 | 华南理工大学 | Traffic flow forecasting method based on genetic algorithm optimization LSTM neural network |
| CN108986470A (en) * | 2018-08-20 | 2018-12-11 | 华南理工大学 | The Travel Time Estimation Method of particle swarm algorithm optimization LSTM neural network |
Non-Patent Citations (2)
| Title |
|---|
| TIAN Y，ET AL: "Predicting Short-Term Traffic Flow by Long Short-Term Memory Recurrent Neural Network", 《IEEE INTERNATIONAL CONFERENCE ON SMART CITY/SOCIALCOM/SUSTAINCOM TOGETHER WITH DATACOM 2015 AND SC2 2015》 * |
| 张文刚: "基于深度学习的交通预测技术及其在通信中的应用研究", 《中国优秀硕士学位论文全文数据库（工程科技Ⅱ辑）》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110288205A (en) * | 2019-06-04 | 2019-09-27 | 北京世纪高通科技有限公司 | Traffic impact analysis method and device |
| CN110288205B (en) * | 2019-06-04 | 2021-07-27 | 北京世纪高通科技有限公司 | Traffic influence evaluation method and device |
| CN110276949A (en) * | 2019-06-20 | 2019-09-24 | 苏州大学 | Forecasting Approach for Short-term Traffic Flow based on machine learning |
| CN110428613A (en) * | 2019-07-09 | 2019-11-08 | 中山大学 | A kind of intelligent transportation trend prediction method of machine learning |
| CN110517485B (en) * | 2019-08-09 | 2021-05-07 | 大连理工大学 | Short-term traffic flow prediction method based on time interval division |
| CN110517485A (en) * | 2019-08-09 | 2019-11-29 | 大连理工大学 | A kind of Short-time Traffic Flow Forecasting Methods based on Time segments division |
| CN110474815A (en) * | 2019-09-23 | 2019-11-19 | 北京达佳互联信息技术有限公司 | Bandwidth prediction method, apparatus, electronic equipment and storage medium |
| US11374825B2 (en) | 2019-09-23 | 2022-06-28 | Beijing Daijia Internet Information Technology Co., Ltd. | Method and apparatus for predicting bandwidth |
| CN110474815B (en) * | 2019-09-23 | 2021-08-13 | 北京达佳互联信息技术有限公司 | Bandwidth prediction method and device, electronic equipment and storage medium |
| CN110717627A (en) * | 2019-09-29 | 2020-01-21 | 浙江大学 | Full traffic prediction method based on dual graph framework |
| CN110717627B (en) * | 2019-09-29 | 2022-01-25 | 浙江大学 | Full traffic prediction method based on dual graph framework |
| CN112884190A (en) * | 2019-11-29 | 2021-06-01 | 杭州海康威视数字技术股份有限公司 | Flow prediction method and device |
| CN112884190B (en) * | 2019-11-29 | 2023-11-03 | 杭州海康威视数字技术股份有限公司 | Flow prediction method and device |
| CN111292534A (en) * | 2020-02-13 | 2020-06-16 | 北京工业大学 | Traffic state estimation method based on clustering and deep sequence learning |
| CN111815046B (en) * | 2020-07-06 | 2024-03-22 | 北京交通大学 | Traffic flow prediction method based on deep learning |
| CN111815046A (en) * | 2020-07-06 | 2020-10-23 | 北京交通大学 | Traffic flow prediction method based on deep learning |
| CN112182961B (en) * | 2020-09-23 | 2023-01-31 | 中国南方电网有限责任公司超高压输电公司 | Converter station wireless network channel large-scale fading modeling prediction method |
| CN112182961A (en) * | 2020-09-23 | 2021-01-05 | 中国南方电网有限责任公司超高压输电公司 | Large-scale fading modeling prediction method for wireless network channel of converter station |
| CN112434260A (en) * | 2020-10-21 | 2021-03-02 | 北京千方科技股份有限公司 | Road traffic state detection method and device, storage medium and terminal |
| CN112489426A (en) * | 2020-11-26 | 2021-03-12 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN112489426B (en) * | 2020-11-26 | 2021-11-09 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN114373299A (en) * | 2021-12-27 | 2022-04-19 | 宁波大学 | Channel traffic flow prediction method based on AIS data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109785629A (en) | A kind of short-term traffic flow forecast method | |
| Yin et al. | Deep learning on traffic prediction: Methods, analysis, and future directions | |
| Liu et al. | Short‐term traffic speed forecasting based on attention convolutional neural network for arterials | |
| CN106874432B (en) | A kind of public transport passenger trip space-time trajectory extracting method | |
| CN109635928A (en) | A kind of voltage sag reason recognition methods based on deep learning Model Fusion | |
| CN106650725A (en) | Full convolutional neural network-based candidate text box generation and text detection method | |
| CN106650913A (en) | Deep convolution neural network-based traffic flow density estimation method | |
| CN109255505A (en) | A kind of short-term load forecasting method of multi-model fused neural network | |
| Zhene et al. | Deep convolutional mesh RNN for urban traffic passenger flows prediction | |
| CN109118020A (en) | A kind of subway station energy consumption short term prediction method and its forecasting system | |
| CN110188936A (en) | Short-time Traffic Flow Forecasting Methods based on multifactor spatial choice deep learning algorithm | |
| CN112668375B (en) | Tourist distribution analysis system and method in scenic spot | |
| Gu et al. | [Retracted] Application of Fuzzy Decision Tree Algorithm Based on Mobile Computing in Sports Fitness Member Management | |
| CN107145991A (en) | A kind of time-varying random network dynamic route searching method of consideration section correlation | |
| Deng et al. | The pulse of urban transport: Exploring the co-evolving pattern for spatio-temporal forecasting | |
| CN114492978A (en) | Time-space sequence prediction method and device based on multi-layer attention mechanism | |
| CN115391553A (en) | Method for automatically searching time sequence knowledge graph complement model | |
| CN113064989B (en) | Method for extracting perception features of public transport sentiments | |
| CN111985727B (en) | Method and system for predicting weather based on loop parting model | |
| CN115565376B (en) | Vehicle journey time prediction method and system integrating graph2vec and double-layer LSTM | |
| CN111678531A (en) | Subway path planning method based on LightGBM | |
| CN116824851A (en) | Path-based urban expressway corridor traffic jam tracing method | |
| Chen et al. | Multi-modal neural network for traffic event detection | |
| CN103886169A (en) | Link prediction algorithm based on AdaBoost | |
| Miao et al. | A queue hybrid neural network with weather weighted factor for traffic flow prediction |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190521 |
|
| RJ01 | Rejection of invention patent application after publication |