CN111754019A - Road section feature representation learning algorithm based on space-time diagram information maximization model - Google Patents
Road section feature representation learning algorithm based on space-time diagram information maximization model Download PDFInfo
- Publication number
- CN111754019A CN111754019A CN202010382570.7A CN202010382570A CN111754019A CN 111754019 A CN111754019 A CN 111754019A CN 202010382570 A CN202010382570 A CN 202010382570A CN 111754019 A CN111754019 A CN 111754019A
- Authority
- CN
- China
- Prior art keywords
- representation
- road section
- road
- time
- learning algorithm
- 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
- 238000010586 diagram Methods 0.000 title claims abstract description 12
- 230000003993 interaction Effects 0.000 claims abstract description 12
- 238000011144 upstream manufacturing Methods 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 claims description 29
- 238000012549 training Methods 0.000 claims description 18
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 12
- 239000013598 vector Substances 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000003068 static effect Effects 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010304 firing Methods 0.000 claims description 3
- 230000006698 induction Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000006641 stabilisation Effects 0.000 claims description 3
- 238000011105 stabilization Methods 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims 1
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 230000007246 mechanism Effects 0.000 abstract description 3
- 239000000284 extract Substances 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 3
- PFFIDZXUXFLSSR-UHFFFAOYSA-N 1-methyl-N-[2-(4-methylpentan-2-yl)-3-thienyl]-3-(trifluoromethyl)pyrazole-4-carboxamide Chemical compound S1C=CC(NC(=O)C=2C(=NN(C)C=2)C(F)(F)F)=C1C(C)CC(C)C PFFIDZXUXFLSSR-UHFFFAOYSA-N 0.000 description 2
- 238000002679 ablation Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q50/14—Travel agencies
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Marketing (AREA)
- Databases & Information Systems (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Primary Health Care (AREA)
- Remote Sensing (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a road section feature representation learning algorithm based on a space-time diagram information maximization model, which considers the timeliness of road section states, deeply excavates the time information of road sections, adopts a maximization mutual information mechanism, extracts the mutual influence and interaction relation among the road section information, the time information and the traffic information and utilizes the relation in the learning algorithm based on a neural network. The obtained road section represents the real-time global traffic condition, and learns the upstream and downstream real-time dependency relationship among the road sections, so that the precision of travel time prediction is greatly improved.
Description
Technical Field
The invention relates to the relevant fields of graph neural networks and the like, in particular to a road section feature representation learning algorithm based on a space-time graph information maximization model.
Background
With the rapid increase of the number of motor vehicles, the urban traffic jam condition is increasingly severe, and a series of problems of low travel efficiency, resource waste and the like are caused. The travel time prediction plays a crucial role in traffic management, path planning, car sharing, vehicle dispatching and other applications. Almost all travel service applications today have this functionality, such as google maps, hundredth maps, drip, etc. Under the support of accurate travel time estimation, a user can reasonably plan an individual path, and time waste on a congested road section is avoided. Simultaneously, the city can also rationally carry out route guidance, effectively slows down the jam problem. Therefore, many researchers are working on timely and effective travel time estimation. However, due to the complexity of travel time estimation, providing accurate estimates remains a challenging task. One method is used for most travel time prediction tasks, that is, link-based travel time prediction. The method can greatly relieve the risk of data sparsity brought by path-based travel time prediction, and has received much attention. In the past, methods for predicting travel time based on road sections have many problems, and one of the great problems is that learning about road section feature representation is not accurate enough, so that the prediction precision is not satisfactory. With the rise of the neural network, the Embedding is applied to the road section feature representation and plays a great role in extracting static information of road sections such as road types and topological relations among road sections, and a non-naive network representation learning method exists. However, the road network is a complex network, and the method still has several problems in the road network, namely 1) it can cause the adjacent road sections to be indistinguishable. The aim of the proposed method is to make the node representation and the adjacent node representation more similar, but in the actual road network, only one of the two adjacent road sections is congested, so that they should be more distinguishable; 2) it does not take into account the interplay of road segments and traffic conditions. Global traffic conditions can affect each road section in a road network, and the states of some key road sections also determine the traffic conditions in turn; 3) it does not take into account the time-varying nature of the road segment itself. The status of the road segments may not be consistent over different time periods, for example, the road segments are congested in the morning and clear in the afternoon.
Disclosure of Invention
The invention provides a road section feature representation learning algorithm based on a space-time diagram information maximization model, which can realize real-time feature representation of a learned road section.
In order to achieve the technical effects, the technical scheme of the invention is as follows:
a road section feature representation learning algorithm based on a space-time diagram information maximization model comprises the following steps:
s1: extracting road section attributes from a road network, generating road section initial vectors, and constructing a time adjacency matrix based on tracks in historical data;
s2: adopting CNN and max-firing operation to the traffic condition, and extracting corresponding traffic condition/flow representation;
s3: inputting the data of S1, S2 and S3 into an encoder for training to obtain a real-time road section representation;
s4: and taking the obtained road section representation as a target, and obtaining the dynamic representation of the road section through the full connection layer.
Further, the specific process of step S1 is:
s11: performing data preprocessing, and acquiring static attributes of each road section through a road network, wherein the three attributes of the road section type, the number of lanes and whether the road section is a one-way road are used;
s12: generating corresponding one-hot vectors for the three attributes, splicing, and obtaining a road section initial vector R ═ R through a full connection layer1,r2,…,rN}；
S13: dividing the road track of the historical data according to time periods, and obtaining a time adjacency matrix A according to the tracks of different time periods(t)That is, if some road segments are driven multiple times and have upstream and downstream relations obtained from the historical track in a certain time period, the corresponding road segments have the adjacent relations, and the adjacent relations are not simply determined from the topological relations.
Further, the specific process of step S2 is:
s21: carrying out grid division on the corresponding city, and calculating the congestion condition and traffic flow of the corresponding grid;
s22: inputting the grid data into the CNN to obtain the representation of the traffic state and the traffic flow;
learning from the grid data a representation that can demonstrate real-time traffic conditions based on the CNN; using the same method to obtain a representation of the mesh inflow and outflow; the specific calculation formula is as follows:
s(t)＝CNN(S(t))。
further, the specific process of step S3 is:
s31: obtaining an adjacency representation h of a link from an initial vector of the link and a temporal adjacency matrix using a graph convolutional neural network(t)；
S32: by negative sampling, repeating the step of S31 results in a broken road segment adjacency representation
S33: generalizing the adjacency representation of the links using a readout function to obtain a global representation g of the graph(t)；
S34: splicing the global representation, the traffic state, the inflow and the outflow of the graph to obtain real-time high-order summary of the graph
S35: and carrying out model training by using gradient descent maximization according to the following objective function through the obtained adjacency representation, negative sampling adjacency representation and high-order graph induction, wherein the adjacency representation of the road section obtained after the training is stable is the final road section representation, and the function formula is as follows:
further, in practice, the interaction information of the segment representation and the global representation is maximized based on the Jensen-Shannon divergence, i.e., the J-S divergence, between the positive and negative examples, and the obtained adjacent representation tends to retain the interaction information of the global representation, and find and retain the similarity at the local level, such as a long-distance segment with similar structural features.
Further, the specific process of step S4 is:
s41: in consideration of the fact that the road section representation has a time cycle rule, mapping the time of the road section static representation and the time of the one-hot code to a low-dimensional representation through a full connection layer to obtain a dynamic road section representation;
s42: acquiring a road segment representation by using an encoder after training stabilization;
s43: minimizing the difference between the L loss and the L loss, optimizing the parameters of the full link layer, and obtaining dynamic representation of the road section based on the full link layer after the training is stable;
s44: in fact, according to the periodicity of the road section state and considering the data sparsity problem, the dynamic representation of the road section is compressed, and the main formula is as follows:
H(t)＝(R+A(t))
compared with the prior art, the technical scheme of the invention has the beneficial effects that:
the invention considers the timeliness of the road section state, deeply excavates the time information of the road section, and adopts a maximum mutual information mechanism to extract and utilize the mutual influence and interaction among the road section information, the time information and the traffic information in the learning algorithm based on the neural network. The obtained road section represents the real-time global traffic condition, and learns the upstream and downstream real-time dependency relationship among the road sections, so that the precision of travel time prediction is greatly improved.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a schematic flow chart of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
As shown in fig. 1, a road segment feature representation learning algorithm based on a space-time diagram information maximization model includes the following steps:
s1: extracting road section attributes from a road network, generating road section initial vectors, and constructing a time adjacency matrix based on tracks in historical data;
s2: adopting CNN and max-firing operation to the traffic condition, and extracting corresponding traffic condition/flow representation;
s3: inputting the data of S1, S2 and S3 into an encoder for training to obtain a real-time road section representation;
s4: and taking the obtained road section representation as a target, and obtaining the dynamic representation of the road section through the full connection layer.
Further, the specific process of step S1 is:
s11: performing data preprocessing, and acquiring static attributes of each road section through a road network, wherein the three attributes of the road section type, the number of lanes and whether the road section is a one-way road are used;
s12: generating corresponding one-hot vectors for the three attributes, splicing, and obtaining a road section initial vector R ═ R through a full connection layer1,r2,…,rN}；
S13: dividing the road track of the historical data according to time periods, and obtaining a time adjacency matrix A according to the tracks of different time periods(t)That is, if some road segments are driven multiple times and have upstream and downstream relations obtained from the historical track in a certain time period, the corresponding road segments have the adjacent relations, and the adjacent relations are not simply determined from the topological relations.
The specific process of step S2 is:
s21: carrying out grid division on the corresponding city, and calculating the congestion condition and traffic flow of the corresponding grid;
s22: inputting the grid data into the CNN to obtain the representation of the traffic state and the traffic flow;
learning from the grid data a representation that can demonstrate real-time traffic conditions based on the CNN; using the same method to obtain a representation of the mesh inflow and outflow; the specific calculation formula is as follows:
s(t)＝CNN(S(t))。
the specific process of step S3 is:
s31: obtaining an adjacency representation h of a link from an initial vector of the link and a temporal adjacency matrix using a graph convolutional neural network(t)；
S32: by negative sampling, repeating the step of S31 results in a broken road segment adjacency representation
S33: generalizing the adjacency representation of the links using a readout function to obtain a global representation g of the graph(t)；
S34: splicing the global representation, the traffic state, the inflow and the outflow of the graph to obtain real-time high-order summary of the graph
S35: and carrying out model training by using gradient descent maximization according to the following objective function through the obtained adjacency representation, negative sampling adjacency representation and high-order graph induction, wherein the adjacency representation of the road section obtained after the training is stable is the final road section representation, and the function formula is as follows:
further, in practice, the interaction information of the segment representation and the global representation is maximized based on the Jensen-Shannon divergence, i.e., the J-S divergence, between the positive and negative examples, and the obtained adjacent representation tends to retain the interaction information of the global representation, and find and retain the similarity at the local level, such as a long-distance segment with similar structural features.
The specific process of step S4 is:
s41: in consideration of the fact that the road section representation has a time cycle rule, mapping the time of the road section static representation and the time of the one-hot code to a low-dimensional representation through a full connection layer to obtain a dynamic road section representation;
s42: acquiring a road segment representation by using an encoder after training stabilization;
s43: minimizing the difference between the L loss and the L loss, optimizing the parameters of the full link layer, and obtaining dynamic representation of the road section based on the full link layer after the training is stable;
s44: in fact, according to the periodicity of the road section state and considering the data sparsity problem, the dynamic representation of the road section is compressed, and the main formula is as follows:
H(t)＝(R+A(t))
as shown in fig. 2, the core object of the present invention is the effect of the road segment representation on traffic prediction. Then we first study the impact of the road segment representation on travel time predictions and determine the dataset, we use the urban trip dataset provided by the dribble-out "gaia" data development plan-the ensemble and the west ampere dataset, published in https:// gaia. Table 1 shows the number of road segments and the number of tracks for the three sets of data sets.
Criteria for travel time prediction are then determined, where the predicted effect of the model is represented by the commonly used RMSE and MAE in the field. Namely, when we do difference evaluation to the predicted time and the real time of the path.
According to the judgment standard, three groups of data sets are divided into a training set, a verification set and a test set, wherein the training set is a track of the first 17 days corresponding to the dripping data set, the data of the last 10 days are used as the test set, and the rest data are used as the verification set; the Harbin data set is the track of the first 3 days, the data of the last 1 day is used as a test set, and the rest data is used as a verification set.
TABLE 1 data set dimension information and interaction information
Before the patent, common road section representation learning methods are all based on a nonprofile network representation learning algorithm, and although the algorithm learns the static attributes of road sections of a road network, the algorithm does not consider the time factors of the road sections and the traffic factors of the road network, and the influence on the precision of travel time prediction is obvious. All of our proposed algorithms use a convolutional neural network based network representation learning algorithm and maximize mutual information to consider the interaction between road segments, road network global and traffic conditions.
For comparison with previous methods, we also calculated the RMSE and MAE performance of these methods on three data sets, and the way in which the training, validation and test sets were partitioned remains the same as our method.
In addition, to test the effect of each part of the model on the model, we performed ablation experiments, yielding the following three model variants, respectively: 1) ST-DGI/S: removing static information of the road section, namely regarding the road section as a node without attribute, and carrying out random initialization on the road section representation; 2) ST-DGI/T: disregarding time factors for the road segment; 3) ST-DGI/G: traffic factors of a road network are ignored, and only generalized representation of a road network graph is used.
TABLE 2 Performance of multiple models on two sets of data
The result shows that the method has obvious improvement compared with the prior method, which is to a great extent because the method extracts the interaction information of the local road section and the global road network to the maximum extent by maximizing a mutual information mechanism from the space-time attribute of the road section, thereby learning the correlation between the road section and the traffic condition and the dynamic dependency relationship between the road sections. We can also confirm the importance of the three factors to the model effect through ablation experiments. Based on accurate dynamic link representation, we can improve the accuracy of travel time prediction.
The same or similar reference numerals correspond to the same or similar parts;
the positional relationships depicted in the drawings are for illustrative purposes only and are not to be construed as limiting the present patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (6)
1. A road section feature representation learning algorithm based on a space-time diagram information maximization model is characterized by comprising the following steps:
s1: extracting road section attributes from a road network, generating road section initial vectors, and constructing a time adjacency matrix based on tracks in historical data;
s2: adopting CNN and max-firing operation to the traffic condition, and extracting corresponding traffic condition/flow representation;
s3: inputting the data of S1, S2 and S3 into an encoder for training to obtain a real-time road section representation;
s4: and taking the obtained road section representation as a target, and obtaining the dynamic representation of the road section through the full connection layer.
2. The road segment feature representation learning algorithm based on the space-time diagram information maximization model as claimed in claim 1, wherein the specific process of step S1 is:
s11: performing data preprocessing, and acquiring static attributes of each road section through a road network, wherein the three attributes of the road section type, the number of lanes and whether the road section is a one-way road are used;
s12: generating corresponding one-hot vectors for the three attributes, splicing, and obtaining a road section initial vector R ═ R through a full connection layer1,r2,…,rN}；
S13: dividing the road track of the historical data according to time periods, and obtaining a time adjacency matrix A according to the tracks of different time periods(t)That is, if some road segments are driven multiple times and have upstream and downstream relations obtained from the historical track in a certain time period, the corresponding road segments have the adjacent relations, and the adjacent relations are not simply determined from the topological relations.
3. The road segment feature representation learning algorithm based on the space-time diagram information maximization model as claimed in claim 2, wherein the specific process of step S2 is:
s21: carrying out grid division on the corresponding city, and calculating the congestion condition and traffic flow of the corresponding grid;
s22: inputting the grid data into the CNN to obtain the representation of the traffic state and the traffic flow;
learning from the grid data a representation that can demonstrate real-time traffic conditions based on the CNN; using the same method to obtain a representation of the mesh inflow and outflow; the specific calculation formula is as follows:
S(t)＝CNN(S(t))。
4. the road segment feature representation learning algorithm based on the space-time diagram information maximization model as claimed in claim 3, wherein the specific process of step S3 is:
s31: obtaining an adjacency representation h of a link from an initial vector of the link and a temporal adjacency matrix using a graph convolutional neural network(t)；
S32: by negative sampling, repeating the step of S31 results in a broken road segment adjacency representation
S33: generalizing the adjacency representation of the links using a readout function to obtain a global representation g of the graph(t)；
S34: splicing the global representation, the traffic state, the inflow and the outflow of the graph to obtain real-time high-order summary of the graph
S35: and carrying out model training by using gradient descent maximization according to the following objective function through the obtained adjacency representation, negative sampling adjacency representation and high-order graph induction, wherein the adjacency representation of the road section obtained after the training is stable is the final road section representation, and the function formula is as follows:
5. the road segment feature representation learning algorithm based on the space-time diagram information maximization model as claimed in claim 4, wherein the interaction information of the road segment representation and the global representation is maximized based on Jensen-Shannon divergence, i.e. J-S divergence, between the positive sample and the negative sample, so that the obtained adjacent representation tends to retain the interaction information of the global representation more, and local level similarity such as a long-distance road segment with similar structural features is found and retained.
6. The road segment feature representation learning algorithm based on the space-time diagram information maximization model as claimed in claim 5, wherein the specific process of step S4 is:
s41: in consideration of the fact that the road section representation has a time cycle rule, mapping the time of the road section static representation and the time of the one-hot code to a low-dimensional representation through a full connection layer to obtain a dynamic road section representation;
s42: acquiring a road segment representation by using an encoder after training stabilization;
s43: minimizing the difference between the L loss and the L loss, optimizing the parameters of the full link layer, and obtaining dynamic representation of the road section based on the full link layer after the training is stable;
s44: in fact, according to the periodicity of the road section state and considering the data sparsity problem, the dynamic representation of the road section is compressed, and the main formula is as follows:
H(t)＝(R+A(t))
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010382570.7A CN111754019B (en) | 2020-05-08 | 2020-05-08 | Road section feature representation learning algorithm based on space-time diagram information maximization model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010382570.7A CN111754019B (en) | 2020-05-08 | 2020-05-08 | Road section feature representation learning algorithm based on space-time diagram information maximization model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111754019A true CN111754019A (en) | 2020-10-09 |
| CN111754019B CN111754019B (en) | 2023-11-07 |
Family
ID=72673259
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010382570.7A Active CN111754019B (en) | 2020-05-08 | 2020-05-08 | Road section feature representation learning algorithm based on space-time diagram information maximization model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111754019B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256980A (en) * | 2021-05-28 | 2021-08-13 | 佳都科技集团股份有限公司 | Road network state determination method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688850A (en) * | 2017-08-08 | 2018-02-13 | 北京深鉴科技有限公司 | A kind of deep neural network compression method |
| CN109830102A (en) * | 2019-02-14 | 2019-05-31 | 重庆邮电大学 | A kind of short-term traffic flow forecast method towards complicated urban traffic network |
| US10349208B1 (en) * | 2018-08-17 | 2019-07-09 | xAd, Inc. | Systems and methods for real-time prediction of mobile device locations |
| CN110491129A (en) * | 2019-09-24 | 2019-11-22 | 重庆城市管理职业学院 | The traffic flow forecasting method of divergent convolution Recognition with Recurrent Neural Network based on space-time diagram |
| CN110929962A (en) * | 2019-12-13 | 2020-03-27 | 中国科学院深圳先进技术研究院 | Traffic flow prediction method and device based on deep learning |
-
2020
- 2020-05-08 CN CN202010382570.7A patent/CN111754019B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688850A (en) * | 2017-08-08 | 2018-02-13 | 北京深鉴科技有限公司 | A kind of deep neural network compression method |
| US10349208B1 (en) * | 2018-08-17 | 2019-07-09 | xAd, Inc. | Systems and methods for real-time prediction of mobile device locations |
| CN109830102A (en) * | 2019-02-14 | 2019-05-31 | 重庆邮电大学 | A kind of short-term traffic flow forecast method towards complicated urban traffic network |
| CN110491129A (en) * | 2019-09-24 | 2019-11-22 | 重庆城市管理职业学院 | The traffic flow forecasting method of divergent convolution Recognition with Recurrent Neural Network based on space-time diagram |
| CN110929962A (en) * | 2019-12-13 | 2020-03-27 | 中国科学院深圳先进技术研究院 | Traffic flow prediction method and device based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 罗文慧 等: "基于CNN-SVR混合深度学习模型的短时交通流预测", 《交通运输系统工程与信息》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256980A (en) * | 2021-05-28 | 2021-08-13 | 佳都科技集团股份有限公司 | Road network state determination method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111754019B (en) | 2023-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Contextualized spatial–temporal network for taxi origin-destination demand prediction | |
| Hu et al. | Stochastic origin-destination matrix forecasting using dual-stage graph convolutional, recurrent neural networks | |
| CN112489426B (en) | Urban traffic flow space-time prediction scheme based on graph convolution neural network | |
| Wang et al. | A big data approach for smart transportation management on bus network | |
| US20240054321A1 (en) | Traffic prediction | |
| CN113762595B (en) | Traffic time prediction model training method, traffic time prediction method and equipment | |
| Fang et al. | Ssml: Self-supervised meta-learner for en route travel time estimation at baidu maps | |
| CN113704956A (en) | Urban road online microscopic simulation method and system based on digital twin technology | |
| CN115440032A (en) | Long-term and short-term public traffic flow prediction method | |
| CN112071062A (en) | Driving time estimation method based on graph convolution network and graph attention network | |
| CN115545758B (en) | Method and system for self-adaptive incremental site selection of urban service facilities | |
| Yuan et al. | Route travel time estimation on a road network revisited: Heterogeneity, proximity, periodicity and dynamicity | |
| Zhang et al. | Gof-tte: Generative online federated learning framework for travel time estimation | |
| Maleki et al. | Developing a supervised learning-based simulation method as a decision support tool for rebalancing problems in bike-sharing systems | |
| Yokoyama et al. | A framework for large-scale train trip record analysis and its application to passengers’ flow prediction after train accidents | |
| CN111754019A (en) | Road section feature representation learning algorithm based on space-time diagram information maximization model | |
| Li et al. | Ridesplitting demand prediction via spatiotemporal multi-graph convolutional network | |
| Hu et al. | SOUP: A fleet management system for passenger demand prediction and competitive taxi supply | |
| Song et al. | Statistical traffic generation methods for urban traffic simulation | |
| Sun et al. | Alleviating data sparsity problems in estimated time of arrival via auxiliary metric learning | |
| CN113688274B (en) | Riding route generation method, device, equipment and storage medium | |
| CN112686457B (en) | Route arrival time estimation method and device, electronic equipment and storage medium | |
| JP7543944B2 (en) | Machine learning program, machine learning method, and estimation device | |
| Xu et al. | Metro train operation plan analysis based on station travel time reliability | |
| CN113409574A (en) | Bayesian network-based traffic health diagnosis method |
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 |