CN112382081A - Traffic flow prediction method based on multiple tasks - Google Patents
Traffic flow prediction method based on multiple tasks Download PDFInfo
- Publication number
- CN112382081A CN112382081A CN202011058017.4A CN202011058017A CN112382081A CN 112382081 A CN112382081 A CN 112382081A CN 202011058017 A CN202011058017 A CN 202011058017A CN 112382081 A CN112382081 A CN 112382081A
- Authority
- CN
- China
- Prior art keywords
- data
- traffic flow
- time
- highway section
- graph
- 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
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000012544 monitoring process Methods 0.000 claims abstract description 50
- 238000004140 cleaning Methods 0.000 claims abstract description 7
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 238000000547 structure data Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000002864 sequence alignment Methods 0.000 claims description 3
- 238000007418 data mining Methods 0.000 abstract description 2
- 238000007781 pre-processing Methods 0.000 abstract 1
- 238000007726 management method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- 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"
-
- G06Q50/40—
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Abstract
The invention discloses a traffic flow prediction method based on multiple tasks, and relates to the field of data mining. The invention comprises the following steps: s1, setting a plurality of highway section monitoring points, and acquiring historical traffic flow data of each highway section monitoring point; s2, aligning the acquired historical traffic flow data time sequence, and performing data cleaning and data preprocessing; s3, constructing a spatio-temporal dependency relationship image used for expressing the monitoring points of each highway section, and obtaining graph data; and S4, inputting the obtained graph data into the multitask model according to time sequence, and predicting the traffic flow and the travel time of a future traffic network. The method has the characteristics of improving the traffic flow prediction precision capability and improving the data utilization efficiency.
Description
Technical Field
The invention relates to the field of data mining, in particular to a traffic flow prediction method based on multiple tasks.
Background
Traffic flow and travel time prediction are important for a traffic management platform, and the traffic management platform can be helped to better complete the spatial scheduling of vehicles. At present, when people predict traffic flow and travel time, the two problems are generally regarded as one task respectively, the adopted method is to learn one task at a time and then combine the learning tasks, but the method ignores the high correlation between the traffic flow and the travel time, further influences the result precision of predicting the traffic flow and the travel time, and simultaneously reduces the efficiency of traffic management.
Therefore, it is necessary to provide a traffic flow prediction method based on multiple tasks.
Disclosure of Invention
The invention provides a multitask-based traffic flow prediction method capable of improving traffic flow prediction accuracy and improving data utilization efficiency, and aims to solve the problems that in the prior art, the conventional traffic demand prediction method neglects high correlation between traffic flow and travel time, so that the result accuracy of traffic flow and travel time prediction is influenced, and the traffic management efficiency is reduced.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides a traffic flow prediction method based on multiple tasks, which comprises the following steps:
s1, setting a plurality of highway section monitoring points, and acquiring historical traffic flow data of each highway section monitoring point; s2, aligning the acquired historical traffic flow data time sequence, cleaning the data, accumulating and combining the data according to set interval time to obtain the traffic flow data of each set interval time, and dividing the traffic flow data into a training data set, a verification data set and a test data set; s3, constructing a spatio-temporal dependency relationship image used for expressing the monitoring points of each highway section, and obtaining graph data; and S4, inputting the obtained graph data into a multitask model according to a time sequence, wherein the multitask model comprises a space-time graph convolution network module and a multitask learning module, and the trained multitask model predicts the traffic flow and the travel time of a future traffic network.
Optionally, a plurality of highway section monitoring points are set, and historical traffic flow data of each highway section monitoring point is obtained, for example: the data such as lane occupancy rate, vehicle number of each lane, vehicle traffic information of each lane and the like comprises the following steps:
determining the specific number of monitoring points of the highway section, extracting required data from the monitored traffic data according to a predicted target, and obtaining time sequence data of the monitoring point of the ith highway sectioni∈[1,N]The historical traffic flow data is used for reflecting the traffic condition of the highway section in the historical period; whereinj∈[1,M]Representing the data in time sequence for the jth period.
Optionally, step S2 further includes the following steps:
s21, cleaning missing values, abnormal values and error records in the history traffic flow data after time sequence alignment; and S22, merging the washed historical traffic flow data according to the set interval time, and simultaneously carrying out normalization processing.
Optionally, step S22 further includes the following steps:
s221, accumulating and combining the historical traffic flow data collected and cleaned by the monitoring points of each highway section according to set interval time to obtain the traffic flow data of each set interval time;
s222, selecting the maximum traffic flow value in the traffic flow data of each set interval time, and taking the ratio of the traffic flow data of each set interval time to the maximum traffic flow value as a traffic flow data set after normalization processing.
Optionally, step S3 further includes the following steps:
s31, regarding each highway section monitoring point as a node, and constructing an image according to the road network distance between the highway section monitoring points; s32, taking the traffic data of each monitoring point obtained in the step S222 as graph node vertex set data, taking the adjacent matrix as an edge set in the graph and storing the edge set as graph structure data, wherein the graph structure data is taken as high-speed traffic data; the adjacency matrix of the road network is calculated by the distance of monitoring stations in the traffic network, and the weighted adjacency matrix W is calculated by the following formula:
wherein wijIs the weight of the edge, by dijObtaining, i.e. the distance, σ, between the ith highway section monitoring point and the jth highway section monitoring point2And oa is a threshold value to control the distribution and sparsity of the matrix W.
Optionally, the space-time graph convolutional network module includes two space-time convolutional blocks and a fully-connected output layer connected to the end; each time-space convolution block comprises two time gate convolution layers, and a space map convolution layer is arranged between the two time gate convolution layers.
Optionally, the multi-task learning module includes a shared hidden layer, and is used for predicting tasks for different traffic flow demands. The multi-task learning module can simultaneously and respectively reserve the output layer mode of each task, and multi-task learning is achieved.
Therefore, the invention has the following beneficial effects:
(1) the method adopts a graph structure on the time-space modeling, adopts an adjacent matrix W to depict the graph node connection, and can construct the relationship between non-adjacent monitoring points; rather than merely establishing a connection between adjacent monitoring points as in the issued patent. The invention introduces the multi-task learning idea, adds the multi-task learning layer, and provides a more accurate prediction result by using the shared weight in the network, and simultaneously improves the utilization efficiency of data;
(2) the invention uses the space-time graph convolution network for traffic flow and travel time prediction, and improves the precision of the prediction result by learning the space-time correlation; on a traffic space model, a model of space characteristics is extracted by a space map convolution layer in the patent; in traffic time sequence, the relation between two adjacent time points is considered compared with the existing RNN; the time gate convolution adopts a convolution operation mode to extract the relation between two or more time points.
(3) The prediction model can realize the prediction of the traffic flow and the travel time of the traffic network, and has wide application prospect in the fields of travel planning, traffic management and the like.
Drawings
Fig. 1 is a flowchart of a traffic flow prediction method based on multiple tasks according to the present invention.
FIG. 2 is a block diagram of a multitasking model of the present invention.
Detailed Description
The invention is further described with reference to the following detailed description and accompanying drawings:
as shown in fig. 1, the present invention provides a traffic flow prediction method based on multitask, including the following steps:
s1, setting a plurality of highway section monitoring points, and acquiring historical traffic flow data of each highway section monitoring point. Taking the high-speed traffic flow data as an example, specifically, according to the step S1, selecting a specific number of monitoring points of the highway section, and according to the prediction target, extracting the required data from the high-speed traffic data acquired from the monitoring points of each highway section as historical traffic flow data, where the historical traffic flow data is past at a certain interval time on the monitoring points or the highway section of the particular highway section, such as: lane occupancy, number of vehicles in each lane, traffic information of vehicles in each lane, and the like. Determining the specific number N of monitoring points of the highway section, extracting required data from the monitored traffic data according to a predicted target, and obtaining time sequence data of the monitoring point of the ith highway sectioni∈[1,N]As historical traffic flow data; whereinj∈[1,M]Representing the data in time sequence for the jth period.
And S2, aligning the acquired historical traffic flow data time sequence, cleaning the data, accumulating and combining the data according to set interval time respectively to obtain the traffic flow data of each set interval time, and dividing the traffic flow data into a training data set, a verification data set and a test data set. Specifically, step S2 further includes the following steps:
s21, cleaning missing values, abnormal values and error records in the history traffic flow data after time sequence alignment; after data washing, a linear interpolation method is used to fill in missing values.
And S22, merging the washed historical traffic flow data according to the set interval time, and simultaneously carrying out normalization processing. The set interval time can be made according to actual prediction requirements.
Step S22 further includes the following steps:
s221, accumulating and combining the historical traffic flow data collected and cleaned by the monitoring points of each highway section according to set interval time to obtain the traffic flow data of each set interval time;
s222, selecting the maximum traffic flow value in the traffic flow data of each set interval time, and taking the ratio of the traffic flow data of each set interval time to the maximum traffic flow value as a traffic flow data set after normalization processing.
The normalized traffic flow data set is divided into a training data set, a validation data set, and a test data set. The data set gives consideration to timing information and spatial information at the same time, specifically, 60% of all traffic flow data is divided into a training data set, 30% is used as a verification data set, and the remaining 10% is used as a test data set.
S3, constructing a spatio-temporal dependency relationship image used for expressing the monitoring points of each highway section, and obtaining graph data; specifically, step S3 further includes the following steps:
s31, regarding each monitoring point of the highway section as a node, and constructing a graph according to the distance of the road network among the monitoring points of the highway section; specifically, if the interval of the historical traffic flow data is set to 5 minutes and the specific number of the selected highway section monitoring points is 288, 288 data points are located at each vertex in the 288 monitoring station road network every day. The loop system can be numerically formed into a directed graph by checking the correlation and the direction and OD (origin-destination) point of each way.
And S32, taking the traffic data of each monitoring point obtained in the step S222 as graph node vertex set data, taking the adjacent matrix as an edge set in the graph and storing the edge set as graph structure data, wherein the graph structure data is high-speed traffic data.
The adjacency matrix of the road network is calculated by the distance of the monitoring stations in the traffic network. The weighted adjacency matrix W is calculated by the following equation:
wherein wijIs the weight of the edge, by dijThe distance between the ith highway section monitoring point and the jth highway section monitoring point is obtained. Sigma2And oa is a threshold value that controls the distribution and sparsity of the matrix W, in which formula σ is set2Is 10, oa is 0.5. The W matrix is used for representing the spatial relation degree of each node.
And S4, inputting the obtained graph data into a multitask model according to a time sequence, wherein the multitask model comprises a space-time graph convolution network module and a multitask learning module, and the trained multitask model predicts the traffic flow and the travel time of a future traffic network.
In the step, the multitask model comprises a space-time graph convolution network module and a multitask learning module. The space-time graph convolution network module comprises two space-time convolution blocks and a fully-connected output layer connected at the tail end; each time-space convolution block comprises two time gate convolution layers, and a space map convolution layer is arranged between the two time gate convolution layers.
Modeling the space characteristics of graph structure input data through a space graph convolution operation, capturing time dimension characteristics through standard two-dimensional convolution, activating a function through a linear correction unit, updating the information of graph nodes by the information of adjacent time slices of the graph nodes after one layer of time dimension convolution, and enabling the information of the graph nodes and the information of the adjacent time slices of the graph nodes to contain the information of the adjacent graph nodes at the same time after the graph convolution operation.
Thus, after a layer of spatio-temporal convolution operations, both the time and space dimensional features of the data and the spatio-temporal correlations are captured. And using multilayer space-time convolution to extract farther information on space-time dimension, and enabling the space-time convolution result to be consistent with the predicted target dimension through full-connection operation, wherein the full-connection module also uses a linear correction unit as an activation function.
In the space-time graph convolution network module, a residual error connection and a bottleeck strategy are used in each space-time volume block. Input vt-M+1,...vtAnd the space-time dependency relationship is obtained by the space-time convolution block processing. All features are integrated by an output layer to generate the final prediction
The multi-task learning module in the multi-task model comprises a shared hidden layer for further extracting the space-time dependency relationship prediction obtained by the space-time graph convolution network moduleTo characterize the respective relevant characteristics and the interaction among different tasks among different traffic tasks, namely a traffic flow task and a travel time prediction task. And the output layers of all tasks are respectively reserved in the multi-task learning module and used for receiving the information extracted from the shared hidden layer and respectively giving predictions of different traffic problems so as to realize multi-task learning. The multitask model selects the mean square error as a loss function for the gradient descent algorithm training model.
This is achieved by sharing the hidden layer between all tasks while preserving the output layer for several specific tasks. The method uses the convolution network of the space-time diagram for predicting the traffic flow and the travel time, improves the accuracy of the prediction result by learning the space-time correlation through multi-task learning, and has the effectiveness that
(1) Many people's related tasks are learned together, with related but also unrelated parts. When a task (Main task) is learned, the part irrelevant to the task is equivalent to noise in the learning process, and therefore, the generalization (generation) effect of learning can be improved by introducing the noise.
(2) In single-task learning, the back-propagation of the gradient tends to fall into local minima. Local minimum values of different tasks in multi-task learning are in different positions, and the shared hidden layer can be helped to escape from the local minimum values through interaction.
(3) The added task can change the dynamic characteristics of weight value updating, and the network is probably more suitable for multi-task learning. For example, the learning rate of the shared hidden layer is improved by multi-task parallel learning, and the learning effect is improved by a larger learning rate.
(4) Multiple tasks are shared and expressed in a shallow layer, the network capacity is possibly weakened, network overfitting is reduced, and the generalization effect is improved.
The prediction of future traffic conditions is a multi-tasking problem. In the structure of the invention, these tasks are trained together and parameters are adjusted by back propagation, so that each task will get a more accurate result.
The invention improves the utilization efficiency of data by introducing the multi-task learning idea, adding a multi-task learning layer and utilizing the sharing weight in the network; the invention also uses the space-time graph convolution network for traffic flow and travel time prediction, and improves the precision of the prediction result by learning the space-time correlation. 1. The learning layer may select a fully connected neural network for obtaining features in the spatio-temporal dependency that reflect the problem being solved.
2. Learning one task at a time may not take into account the interaction between tasks, learning multiple tasks simultaneously, and the auxiliary task may help improve the learning performance of the primary task. The prediction of the traffic flow and travel time of the future traffic network is complementary.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (7)
1. A traffic flow prediction method based on multiple tasks is characterized by comprising the following steps:
s1, setting a plurality of highway section monitoring points, and acquiring historical traffic flow data of each highway section monitoring point;
s2, aligning the acquired historical traffic flow data time sequence, cleaning the data, accumulating and combining the data according to set interval time to obtain the traffic flow data of each set interval time, and dividing the traffic flow data into a training data set, a verification data set and a test data set;
s3, constructing a spatio-temporal dependency relationship image used for expressing the monitoring points of each highway section, and obtaining graph data;
and S4, inputting the obtained graph data into a multitask model according to a time sequence, wherein the multitask model comprises a space-time graph convolution network module and a multitask learning module, and the trained multitask model predicts the traffic flow and the travel time of a future traffic network.
2. The multitask-based traffic flow predicting method according to claim 1, wherein the step S1 sets a plurality of highway section monitoring points, and acquires historical traffic flow data of each highway section monitoring point, and includes the steps of:
determining the specific number N of monitoring points of the highway section, extracting required data from the monitored traffic data according to a predicted target, and obtaining time sequence data of the monitoring point of the ith highway sectioni∈[1,N]As historical traffic flow data; whereinj∈[1,M]Representing the data in time sequence for the jth period.
3. The multitask based traffic flow prediction method according to claim 1, wherein the step S2 further includes the steps of:
s21, cleaning missing values, abnormal values and error records in the history traffic flow data after time sequence alignment;
and S22, merging the washed historical traffic flow data according to the set interval time, and simultaneously carrying out normalization processing.
4. The multitask based traffic flow prediction method according to claim 3, wherein the step S22 further includes the steps of:
s221, accumulating and combining the historical traffic flow data collected and cleaned by the monitoring points of each highway section according to set interval time to obtain the traffic flow data of each set interval time;
s222, selecting the maximum traffic flow value in the traffic flow data of each set interval time, and taking the ratio of the traffic flow data of each set interval time to the maximum traffic flow value as a traffic flow data set after normalization processing.
5. The multitask based traffic flow prediction method according to claim 4, wherein the step S3 further includes the steps of:
s31, regarding each monitoring point of the highway section as a node, and constructing a graph according to the distance of the road network among the monitoring points of the highway section;
s32, taking the traffic data of each monitoring point obtained in the step S222 as graph node vertex set data, taking the adjacent matrix as an edge set in the graph and storing the edge set as graph structure data, wherein the graph structure data is taken as high-speed traffic data; the adjacency matrix of the road network is calculated by the distance of monitoring stations in the traffic network, and the weighted adjacency matrix W is calculated by the following formula:
6. The multitask based traffic flow prediction method according to claim 1, wherein the space-time graph convolution network module comprises two space-time convolution blocks and a fully connected output layer connected at the end; each time-space convolution block comprises two time gate convolution layers, and a space map convolution layer is arranged between the two time gate convolution layers.
7. The multitask based traffic flow prediction method according to claim 1, wherein the multitask learning module comprises a shared hidden layer for predicting tasks for different traffic flow demands.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011058017.4A CN112382081A (en) | 2020-09-30 | 2020-09-30 | Traffic flow prediction method based on multiple tasks |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011058017.4A CN112382081A (en) | 2020-09-30 | 2020-09-30 | Traffic flow prediction method based on multiple tasks |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112382081A true CN112382081A (en) | 2021-02-19 |
Family
ID=74580891
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011058017.4A Pending CN112382081A (en) | 2020-09-30 | 2020-09-30 | Traffic flow prediction method based on multiple tasks |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112382081A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905856A (en) * | 2021-05-08 | 2021-06-04 | 浙江高速信息工程技术有限公司 | Method for constructing high-speed traffic data set with space-time dependence |
CN113470352A (en) * | 2021-06-17 | 2021-10-01 | 之江实验室 | Traffic big data analysis and prediction system and method based on multitask learning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN110674987A (en) * | 2019-09-23 | 2020-01-10 | 北京顺智信科技有限公司 | Traffic flow prediction system and method and model training method |
CN111081010A (en) * | 2018-10-18 | 2020-04-28 | 阿里巴巴集团控股有限公司 | Method and device for predicting traffic flow or travel time |
CN111091712A (en) * | 2019-12-25 | 2020-05-01 | 浙江大学 | Traffic flow prediction method based on cyclic attention dual graph convolution network |
CN111653088A (en) * | 2020-04-21 | 2020-09-11 | 长安大学 | Vehicle driving quantity prediction model construction method, prediction method and system |
-
2020
- 2020-09-30 CN CN202011058017.4A patent/CN112382081A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111081010A (en) * | 2018-10-18 | 2020-04-28 | 阿里巴巴集团控股有限公司 | Method and device for predicting traffic flow or travel time |
CN110674987A (en) * | 2019-09-23 | 2020-01-10 | 北京顺智信科技有限公司 | Traffic flow prediction system and method and model training method |
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 |
CN111091712A (en) * | 2019-12-25 | 2020-05-01 | 浙江大学 | Traffic flow prediction method based on cyclic attention dual graph convolution network |
CN111653088A (en) * | 2020-04-21 | 2020-09-11 | 长安大学 | Vehicle driving quantity prediction model construction method, prediction method and system |
Non-Patent Citations (1)
Title |
---|
ZHANG KUNPENG ET AL: "A Multitask Learning Model for Traffic Flow and Speed Forecasting", 《IEEE ACCESS》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905856A (en) * | 2021-05-08 | 2021-06-04 | 浙江高速信息工程技术有限公司 | Method for constructing high-speed traffic data set with space-time dependence |
CN113470352A (en) * | 2021-06-17 | 2021-10-01 | 之江实验室 | Traffic big data analysis and prediction system and method based on multitask learning |
CN113470352B (en) * | 2021-06-17 | 2022-10-21 | 之江实验室 | Traffic big data analysis and prediction system and method based on multitask learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111009129B (en) | Urban road traffic flow prediction method and device based on space-time deep learning model | |
CN112330215B (en) | Urban vehicle demand prediction method, equipment and storage medium | |
CN111210633A (en) | Short-term traffic flow prediction method based on deep learning | |
CN112766597B (en) | Bus passenger flow prediction method and system | |
CN114299723B (en) | Traffic flow prediction method | |
CN113313947A (en) | Road condition evaluation method of short-term traffic prediction graph convolution network | |
CN113326974B (en) | Multi-source traffic flow prediction method based on multi-task hypergraph convolutional network | |
US20240054321A1 (en) | Traffic prediction | |
CN111639791A (en) | Traffic flow prediction method, system, storage medium and terminal | |
CN114287022A (en) | Multi-step traffic prediction | |
CN113762595B (en) | Traffic time prediction model training method, traffic time prediction method and equipment | |
CN113362637B (en) | Regional multi-field-point vacant parking space prediction method and system | |
CN112150207A (en) | Online taxi appointment order demand prediction method based on space-time context attention network | |
CN112382081A (en) | Traffic flow prediction method based on multiple tasks | |
Kuang et al. | Intelligent traffic signal control based on reinforcement learning with state reduction for smart cities | |
CN116307152A (en) | Traffic prediction method for space-time interactive dynamic graph attention network | |
CN112613630B (en) | Short-term traffic demand prediction method integrating multi-scale space-time statistical information | |
CN114066503A (en) | Method for forecasting taxi demand based on virtual service area construction | |
CN116307293B (en) | Urban space-time data prediction method based on hybrid perception and causal depolarization | |
Alecsandru et al. | Hybrid model-based and memory-based traffic prediction system | |
CN115762147B (en) | Traffic flow prediction method based on self-adaptive graph meaning neural network | |
CN117116045A (en) | Traffic flow prediction method and device based on space-time sequence deep learning | |
CN114169659A (en) | Criminal spatiotemporal risk prediction and decision scheduling method and device | |
Xu et al. | An LSTM approach for predicting the short-time passenger flow of urban bus | |
CN116957166B (en) | Tunnel traffic condition prediction method and system based on Hongmon system |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210219 |