CN113887704A - Traffic information prediction method, device, equipment and storage medium - Google Patents

Traffic information prediction method, device, equipment and storage medium Download PDF

Info

Publication number
CN113887704A
CN113887704A CN202111147932.5A CN202111147932A CN113887704A CN 113887704 A CN113887704 A CN 113887704A CN 202111147932 A CN202111147932 A CN 202111147932A CN 113887704 A CN113887704 A CN 113887704A
Authority
CN
China
Prior art keywords
traffic
target
information prediction
traffic information
source
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
Application number
CN202111147932.5A
Other languages
Chinese (zh)
Inventor
余剑峤
黄芸洁
宋晓壮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southwest University of Science and Technology
Southern University of Science and Technology
Original Assignee
Southwest University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southwest University of Science and Technology filed Critical Southwest University of Science and Technology
Priority to CN202111147932.5A priority Critical patent/CN113887704A/en
Publication of CN113887704A publication Critical patent/CN113887704A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Pure & Applied Mathematics (AREA)
  • Biophysics (AREA)
  • Mathematical Optimization (AREA)
  • Economics (AREA)
  • Mathematical Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Chemical & Material Sciences (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Analytical Chemistry (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention provides a traffic information prediction method, a device, equipment and a storage medium, which relate to the field of data processing, wherein the method comprises the following steps: the method comprises the steps of obtaining a target urban traffic road map, dividing the target urban traffic road map into a plurality of target sub-maps, forming a plurality of target sub-maps and corresponding target traffic state information into target input data, and inputting the target input data into a pre-trained traffic information prediction model to obtain a traffic information prediction value, wherein the traffic information prediction model is obtained by training a source urban traffic road map in advance, and the number of nodes of the target urban traffic road map is less than that of nodes of the source urban traffic road network. The traffic information prediction model is trained by using the source urban traffic network data with sufficient data volume, and then applied to the target city with less data volume, so that the traffic information prediction precision and prediction efficiency of the target city can be remarkably improved.

Description

Traffic information prediction method, device, equipment and storage medium
Technical Field
The embodiment of the application relates to the technical field of data processing, in particular to a traffic information prediction method, a traffic information prediction device, traffic information prediction equipment and a storage medium.
Background
The rapid development of urbanization enables people to live more modernized, however, the rapid development of urbanization also brings more problems such as traffic jam, environmental pollution and overuse of land to modern cities. The diffusion of big data and the rapid development of computing power provide possibilities for solving these problems using data science and computing techniques, and urban computing aims to perform smart city system construction using massive data generated in cities.
For example, traffic speed prediction is a challenge in the construction of smart city systems. The related art includes a classical statistical method and a deep learning model, and the traditional statistical method utilizes, for example, autoregressive integrated moving average and historical average to predict traffic speed, but the methods are limited to non-stationary sequences and have very limited capability of processing highly complex non-linear time sequence data. Deep learning models, such as data-driven models, utilize big data to make traffic speed predictions, but this approach faces the problem of scarcity of city data. This lack of data is mainly due to the cost of building an urban-wide traffic sensor complex and the considerable time required to collect the data.
Disclosure of Invention
The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.
The embodiment of the application provides a traffic information prediction method, a device, equipment and a storage medium, which can train a traffic information prediction model by using city data with sufficient data volume and transfer the traffic information prediction model to a city with rare data volume for traffic information prediction, thereby effectively improving the traffic information prediction precision and the prediction efficiency.
In a first aspect, an embodiment of the present application provides a traffic information prediction method, including:
acquiring a target city traffic road map, and dividing the target city traffic road map into a plurality of target subgraphs;
forming target input data by the target subgraphs and the corresponding target traffic state information;
and inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value, wherein the traffic information prediction model is obtained by training a source urban traffic road graph in advance, and the number of nodes of the target urban traffic road graph is different from that of the nodes of the source urban traffic road network.
In an optional implementation manner, the obtaining a target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs includes:
reducing the size of the target city traffic road map according to different proportions to obtain a plurality of thumbnails in different proportions;
preliminarily dividing the thumbnail with the minimum proportion according to the dividing number to obtain a first target subgraph set;
and dividing the target city traffic road map into corresponding target subgraphs according to the node mapping relation in the first target subgraph set.
In an optional implementation manner, before the obtaining a target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs, the method further includes:
and when the node number of the target city traffic road graph cannot be divided by the division number, completing the data of the target city traffic road graph by using element zero.
In an alternative implementation, the traffic information prediction model includes a spatial correlation sub-model and a temporal correlation sub-model;
the step of inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value comprises the following steps:
acquiring spatial information by using the spatial correlation submodel according to the target input data;
and acquiring time information by using the time correlation sub-model according to the spatial information so as to obtain the traffic information predicted value according to the spatial information and the time information.
In an alternative implementation, the spatial correlation sub-model is formed by a graph neural network model, and the temporal correlation sub-model is formed by a gated cyclic unit model.
In an optional implementation manner, the training process of the traffic information prediction model includes:
constructing a source city training data set, the source city training data set comprising: the source input data comprise source subgraphs corresponding to the source city traffic road map and corresponding source traffic state information;
inputting the source city training data set into the traffic information prediction model to obtain source prediction traffic state information;
and optimizing a loss function according to the error value between the source predicted traffic state information and the label, and training to obtain the traffic information prediction model.
In an optional implementation manner, the method further includes adjusting a second loss function by using a second error value between the traffic information prediction value and the real target traffic state information in the prediction time period, and optimizing parameters of the traffic information prediction model.
In a second aspect, an embodiment of the present application provides a traffic information prediction apparatus, including:
the target city traffic road map acquisition module is used for acquiring a target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs;
the target input data composition module is used for composing a plurality of target subgraphs and corresponding target traffic state information into target input data;
and the traffic information prediction module is used for inputting the target input data into a traffic information prediction model which is trained in advance to obtain a traffic information prediction value, the traffic information prediction model is obtained by training a source urban traffic road graph in advance, and the number of nodes of the target urban traffic road graph is different from that of the source urban traffic road network.
In a third aspect, a computer device includes a processor and a memory;
the memory is used for storing programs;
the processor is configured to execute the traffic information prediction method according to any one of the first aspect according to the program.
In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions for performing the traffic information prediction method according to any one of the first aspect.
Compared with the related art, the traffic information prediction method provided by the first aspect of the embodiment of the application obtains the traffic information prediction value by obtaining the target city traffic road map, dividing the target city traffic road map into a plurality of target sub-maps, then forming target input data by the plurality of target sub-maps and corresponding target traffic state information, and finally inputting the target input data into a traffic information prediction model trained in advance, wherein the traffic information prediction model is obtained by training the source city traffic road map in advance, and the number of nodes of the target city traffic road map is less than that of nodes of the source city traffic road network. The traffic information prediction model is trained by using the source urban traffic network data with sufficient data volume, and then applied to the target city with less data volume, so that the traffic information prediction precision and prediction efficiency of the target city can be remarkably improved.
It is to be understood that the advantageous effects of the second aspect to the fourth aspect compared to the related art are the same as the advantageous effects of the first aspect compared to the related art, and reference may be made to the related description of the first aspect, which is not repeated herein.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the embodiments or the related technical descriptions will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive labor.
FIG. 1 is a schematic diagram of an exemplary system architecture provided by one embodiment of the present application;
FIG. 2 is a flow chart of a traffic information prediction method provided by an embodiment of the present application;
FIG. 3 is a flow chart of a traffic information prediction method according to an embodiment of the present application;
fig. 4 is a schematic diagram of sub-graph partitioning of a traffic information prediction method according to an embodiment of the present application;
FIG. 5 is a schematic illustration of road traffic provided by an embodiment of the present application;
FIG. 6 is a flow chart of a traffic information prediction method according to an embodiment of the present application;
FIG. 7 is a flow chart of a traffic information prediction method according to an embodiment of the present application;
FIG. 8 is a comparison of performance of a traffic information prediction method according to an embodiment of the present application;
FIG. 9 is a comparison of further performance of a traffic information prediction method according to an embodiment of the present application;
fig. 10 is a block diagram showing a configuration of a traffic information prediction apparatus according to an embodiment of the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the embodiments of the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the embodiments of the present application with unnecessary detail.
It should be noted that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different from that in the flowcharts. The terms first, second and the like in the description and in the claims, and the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
It should also be appreciated that reference throughout the specification to "one embodiment" or "some embodiments," or the like, means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," or the like, in various places throughout this specification are not necessarily all referring to the same embodiment, but rather "one or more but not all embodiments" unless specifically stated otherwise. The terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless expressly specified otherwise.
The rapid development of urbanization brings problems such as traffic jam to modern cities, and for example, traffic speed prediction is a challenge in the construction of a smart city system. The related art includes a classical statistical method and a deep learning model, and the traditional statistical method utilizes, for example, autoregressive integrated moving average and historical average to predict traffic speed, but the methods are limited to non-stationary sequences and have very limited capability of processing highly complex non-linear time sequence data.
As the range of traffic data expands and diversity increases, data-driven traffic prediction methods have proven to be superior to traditional simulation-based methods. Such as basic time series models like auto-regressive moving-complex averages, kalman filters and their variants, spatio-temporal regularization regression models, support vector regression, etc., which work well in classification and regression tasks using large data, but these methods require artificial feature processing. With the increase of data volume, some deep learning models requiring a large amount of data are gradually started to be used for traffic prediction tasks, for example, a data-driven model can utilize large data to predict traffic speed, and a deep learning method can effectively extract high-dimensional features from the large amount of data. Data-driven deep learning traffic prediction models, such as deep belief networks and stacked autoencoders, show superior performance in traffic flow prediction. In recent years, a recurrent neural network has also been widely used in the time series prediction task due to its performance in time-dependent modeling. Although deep learning has been developed due to the proliferation of urban data, these methods face the problem of scarcity of urban data, mainly due to the cost of building an urban-wide traffic sensor integrated system and the considerable time required to collect data. However, due to the lack of data collection methods, many small cities have a limited amount of traffic data available for building data-driven models. With the acceleration of the urbanization process, the demand of traffic construction of medium and small cities is urgent.
Therefore, the embodiment of the application provides a traffic information prediction method, and compared with the related technology, a traffic information prediction value is obtained by obtaining a target city traffic road map, dividing the target city traffic road map into a plurality of target subgraphs, then forming target input data by the plurality of target subgraphs and corresponding target traffic state information, and finally inputting the target input data into a traffic information prediction model trained in advance, wherein the traffic information prediction model is obtained by training a source city traffic road map in advance, and the number of nodes of the target city traffic road map is less than that of nodes of the source city traffic road network. The traffic information prediction model is trained by using the source urban traffic network data with sufficient data volume, and then applied to the target city with less data volume, so that the traffic information prediction precision and prediction efficiency of the target city can be remarkably improved.
It is understood that the traffic information prediction method provided in the embodiments of the present application may be implemented by various electronic devices with computing processing capability, for example, various user terminals such as a laptop computer, a tablet computer, a desktop computer, a set-top box, a mobile device (e.g., a mobile phone, a portable music player, a personal digital assistant, a dedicated messaging device, a portable game device), and the like, and may also be implemented by a server.
It should be noted that the server may be an independent physical server, may also be a server cluster or a distributed system formed by a plurality of physical servers, and may also be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a network service, cloud communication, a middleware service, a domain name service, a security service, and a big data and artificial intelligence platform, which is not limited herein.
In order to facilitate understanding of the technical solutions provided in the embodiments of the present application, an application scenario in which the traffic information prediction method provided in the embodiments of the present application is applied to a server is described below as an example.
Referring to fig. 1, fig. 1 is a schematic view of an application scenario of a traffic information prediction method provided in an embodiment of the present application.
As shown in fig. 1, system architecture 100 may include a database 101, a network 102, and a server 103. Network 102 is the medium used to provide communication links between database 101 and server 103. Network 102 may include various connection types, such as wired communication links, wireless communication links, and so forth.
In an embodiment of the present invention, the server 103 obtains the target city traffic road map from the database 101, divides the target city traffic road map into a plurality of target sub-maps, then composes the plurality of target sub-maps and corresponding target traffic state information into target input data, and obtains a traffic information prediction value by inputting the target input data into a traffic information prediction model trained in advance (a prediction process of traffic information prediction will be described in detail below). The traffic information prediction model is trained by using the source urban traffic network data with sufficient data volume, and then applied to the target city with less data volume, so that the traffic information prediction precision and prediction efficiency of the target city can be remarkably improved.
It should be noted that the traffic information prediction method provided by the embodiment of the present invention is generally executed by the server 103, and accordingly, the traffic information prediction apparatus is generally disposed in the server 103. However, in other embodiments of the present invention, the terminal device may also have a similar function as the server, so as to execute the traffic information prediction scheme provided by the embodiment of the present invention.
The system architecture and the application scenario described in the embodiment of the present application are for more clearly illustrating the technical solution of the embodiment of the present application, and do not form a limitation on the technical solution provided in the embodiment of the present application, and it is known by those skilled in the art that the technical solution provided in the embodiment of the present application is also applicable to similar technical problems with the evolution of the system architecture and the appearance of new application scenarios. Those skilled in the art will appreciate that the system architecture shown in FIG. 1 is not intended to be limiting of embodiments of the present application and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
Based on the system architecture, various embodiments of the traffic information prediction method of the embodiment of the application are provided.
As shown in fig. 2, fig. 2 is a flowchart of a traffic information prediction method according to an embodiment of the present application, including but not limited to step S110 and step S130.
And step S110, acquiring the target city traffic road map, and dividing the target city traffic road map into a plurality of target subgraphs.
In one embodiment, the traffic network may be represented as a topology similar to a graph structure, nodes of the graph structure may represent observation points set in the traffic network, and sensors or cameras may be set at different positions of the observation points to record traffic conditions of each observation point in a past period of time. The edges of the graph structure are directed edges and may represent roads.
In one embodiment, T represents a target city with a small amount of data (e.g., fewer observation points and less available information), and N is providedTThe target urban traffic network of the target urban road network system of each node is represented as GT。
GT＝(VT，ET，AT)
Wherein, VTN in network system for representing target urban roadTAnd (4) collecting node information.
ETRepresents the connection NTA set of directed edges for a node.
ATAn adjacency matrix representing connectivity between nodes in the target urban road network system is represented as:
Figure BDA0003286110040000051
wherein i and j represent node numbers, if any, NiPoint to NjAnd is absent of NjPoint to NiThen, then
Figure BDA0003286110040000052
In one embodiment, the target urban traffic road network G is obtainedTAnd then carrying out sub-graph division on the data. Referring to fig. 3, step S110 includes, but is not limited to, the following steps:
and step S111, reducing the size of the target city traffic road map according to different proportions to obtain a plurality of thumbnails with different proportions.
In one embodiment, referring to fig. 4, a diagram of sub-graph division is shown, and a diagram of a target urban traffic road is shown as G0In different proportions, G0Zoom to G1、G2、G3、G4Four thumbnails at different scales.
And step S112, preliminarily dividing the thumbnail with the minimum proportion according to the dividing number to obtain a first target subgraph.
In one embodiment, the number of divisions is k, and the smallest-scale thumbnail in FIG. 4 is G4Into the middle of G4Performing primary division to obtain a first target subgraph set G4’。
And S113, dividing the target city traffic road map into corresponding target subgraphs according to the node mapping relation in the first target subgraph set.
In one embodiment, sub-graph G is based on a first target sub-graph4' the node mapping relation of the reaction in (1) maps the target city traffic road map G0Is likewise divided into G0', in which the number k is dividedNamely the number of clusters, the target subgraph comprises k or k types of traffic structures, for example (a fork, a circular intersection and the like in a road), the basic structures generally exist in each city, and each traffic graph comprises various different traffic structures, so that the model can be migrated between cities with different data volumes.
In one embodiment, the sub-graph partitioning of step S110 is performed using a METIS graph partitioning method. METIS is a powerful graph cut software package developed by Karypis Lab. Precisely, METIS is a software package that is cut into a serial graph. The algorithm design of the METIS is mainly based on a multi-level recursive bisection method, a multi-level K-way bisection method and a multi-constraint division mechanism, and when a user uses the METIS software package, the user can select a corresponding division mode according to needs.
The main characteristics of METIS are as follows: first, METIS has high quality segmentation results, purportedly 10% -50% more accurate than normal spectral clustering; secondly, the METIS has very high execution efficiency, is 1-2 orders of magnitude faster than a common division calculation method, and can be divided into 256 classes within a few seconds; finally, METIS has very low injection elements, thereby reducing storage load and computational load.
In one embodiment, the target urban traffic road network GTPartitioning into similarly sized MTAnd (4) each target subgraph. When the target city traffic road map GTCannot be divided into numbers (i.e., M in this example)T) When dividing completely, using element zero to make target city traffic road map GTAnd (5) performing data completion.
Referring to fig. 5, a schematic view of road traffic according to which the road traffic is represented in topological form in an embodiment of the present application is shown.
In fig. 5 there are two traffic roads, one bidirectional road and one unidirectional road. The segments of each road are labeled from 1 to 5, which are defined as nodes of the target city traffic road map.
For node 1 (segment 1), no other node (segment) is directly connected to its node.
For node 2 (segment 2), there are two nodes (i.e., 3 and 4) connected to this node (node 2).
Thus, there are edges (3,2), (4,2), respectively, which are denoted as A, according to the adjacency matrix of the connectivity between the individual nodes in the target urban road network system32＝A42＝A15＝1。
The matrix form is represented as:
Figure BDA0003286110040000061
in one embodiment, since the use of METIS requires the input of a symmetric adjacency matrix, when performing sub-graph division, the adjacency matrix is subjected to a symmetry process, which is expressed as:
Figure BDA0003286110040000071
in one embodiment, the adjacency matrix is represented as:
Figure BDA0003286110040000072
it is converted into a sparse matrix form at the METIS input, represented as: where Input [ x ] indicates the number of all nodes connected to the xth node, and (0, 4) is 1 for the above-mentioned adjacency matrix, (0, 4) indicates that there is a connection between the 0 th node and the 4 th node, and therefore the content of the corresponding sparse matrix Input [0] is [4 ].
Assume that the content output through METIS is membership ═ 0,1, 1, 0, 1], where membership [ x ] indicates which class the first node belongs to, so that sub-graph partitioning is performed according to its class.
And step S120, forming target input data by the plurality of target subgraphs and the corresponding target traffic state information.
In one embodiment, the target traffic state information with time step t is represented as
Figure BDA0003286110040000073
Where N denotes a node and F denotes the number of traffic information of interest, which in this embodiment includes a traffic speed, i.e., takes F1, which is a traffic information prediction, i.e., a traffic speed prediction. It is understood that the traffic information is only illustrated here, and is not meant to be limited to traffic speed, and other traffic flow information may also be predicted by applying the method of the present embodiment.
In one embodiment, the target traffic status information includes target historical traffic status information for the first H time steps, represented as:
Figure BDA0003286110040000074
the target input information is represented as: [ G ]T，XT]。
And step S130, inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value.
In one embodiment, the output is the predicted traffic state information of the next Q time steps, i.e. the predicted value of the traffic information of the target city, which is expressed as:
Figure BDA0003286110040000075
in one embodiment, the traffic information prediction model comprises a spatial correlation sub-model and a temporal correlation sub-model, wherein the spatial correlation sub-model is used for acquiring spatial information, and inputting the spatial information into the temporal correlation sub-model to acquire temporal information, so as to obtain a traffic information prediction value according to the spatial information and the temporal information.
In one embodiment, the spatial correlation sub-model is constructed from a graph neural network model and the temporal correlation sub-model is constructed from a gated cyclic unit model.
A Graph neural Network (GCN) of a spatial correlation sub-model, also called a Graph convolution neural Network, is a feature extractor, and only the object to be processed is Graph data. Features (e.g., the spatial information described above) can be extracted from the graph data so that the features can be used to perform node classification, graph classification, edge prediction, or the like on the graph data.
The graph neural network GCN is an effective method for establishing spatial correlation of traffic networks. The shortcoming that the convolutional neural network CNN only processes Euclidean space data is overcome. Since the graph structure better reflects the topology structure of the traffic network, the graph neural network GCN can capture the spatial information features of finer granularity in the traffic network.
The GCN can adopt spectrogram convolution, which is based on spectrogram theory and uses spectral clustering to construct a filter in a Fourier domain to research the properties of the map through eigenvalues and eigenvectors of a Laplace matrix of the map. The laplacian matrix is symmetric, with the matrix having non-zero elements only at the vertices themselves and their 1-hop neighbors, which makes the convolution process simpler.
The specific model input and output process is as follows.
Assuming that a batch of graph data is provided, wherein N nodes are provided, each node has its own spatial information characteristic, the characteristics of the nodes are configured to form an N × D dimensional characteristic matrix X, and then the relationship between the nodes forms an N × N dimensional matrix a, which is also called an adjacency matrix, with X and a as the input of the model.
First, the laplace matrix of the graph is calculated using the adjacency matrix, which is expressed as:
Figure BDA0003286110040000081
wherein the content of the first and second substances,
Figure BDA0003286110040000082
representing an adjacency matrix with self-connection, INThe unit matrix is represented by a matrix of units,
Figure BDA0003286110040000083
representing a diagonal matrix with self-connected adjacency matrices.
Since the graph neural network GCN is also a neural network layer, there is H for each convolutional layer(l+1)＝σ(LH(l)θ(l)) Wherein L represents the number of layers, L represents the number of layers, H(l)Denotes the output of the l-th layer, θ(l)Represents the weight parameter of the l < th > layer, and sigma (·) represents sigmoid function.
The output is f (X, A), expressed as:
f(X,A)＝σ(LReLU(LXW0)W1)
wherein X represents a feature matrix, W0Representing a weight matrix mapping of the input to the hidden unit, W1Represents the weight parameter of the next layer, and ReLU (·) represents the corrected linear unit.
For the graphs with many nodes, the graph neural network GCN has higher requirements on machine performance, and the complexity of the solving process is improved, so that in the embodiment, the graph neural network GCN is enabled to pay more attention to information extraction on the subgraph by dividing the subgraph, the equipment consumption is reduced, and the accuracy is improved.
When the traffic information is traffic speed, the traffic speed prediction is essentially a time-series prediction task. It is therefore desirable to build a time correlation sub-model that is applicable to time series data.
In one embodiment, the temporal correlation sub-model is constructed from a gated loop element model.
A Gated Round Unit (GRU) is used to capture time information. The gated cyclic unit GRU is a variant of the recurrent neural network RNN. The method can reduce the problems of gradient explosion and gradient disappearance existing in an RNN network, and compared with another RNN variant (namely Long-Short Term Memory, LSTM), the method has the advantages that the model parameters of the gated cyclic unit GRU are fewer, and the structure is simpler.
Therefore, in this embodiment, the graph neural network GCN is used to form a spatial correlation sub-model, the gated cyclic unit GRU is used to form a temporal correlation sub-model, and the spatial correlation sub-model and the temporal correlation sub-model together form a traffic information prediction model.
In an embodiment, the traffic information prediction model is obtained by training a source urban traffic road map in advance, the number of nodes of the target urban traffic road map is different from the number of nodes of the source urban traffic road network, for example, the number of nodes of the source urban traffic road network can be far greater than the number of nodes of the target urban traffic road network, so that the traffic information prediction model is trained by using source urban traffic road network data with sufficient data volume, and then the traffic information prediction model is applied to a target city with small data volume, so that the traffic information prediction precision and the prediction efficiency of the target city can be remarkably improved.
Referring to fig. 6, a schematic diagram of a training process of the traffic information prediction model in this embodiment is shown, where the training process includes, but is not limited to, the following steps:
step S610, constructing a source city training data set, wherein the source city training data set comprises: source input data and corresponding tags.
In one embodiment, the source input data includes a source subgraph corresponding to the source city traffic road graph and corresponding source traffic state information, and the label is the actual source traffic state information of the prediction time period corresponding to the source subgraph.
For example, a source city with a large amount of data is represented by S, and N is providedSEach node, the source urban traffic network of the source urban road network system is represented as GS。
GS＝(VS，ES，AS)
Wherein, VSN in representation source urban road network systemSAnd (4) collecting node information.
ESRepresents the connection NSA set of directed edges for a node.
ASAn adjacency matrix representing connectivity between individual nodes in a source urban road network system is represented as:
Figure BDA0003286110040000091
wherein i and j represent node numbers, if any, NiPoint to NjAnd is absent of NjPoint to NiThen, then
Figure BDA0003286110040000092
Meanwhile, the source traffic state information having a time step of t is represented as
Figure BDA0003286110040000093
In this embodiment, the traffic information includes a traffic speed, that is, F is 1, it is understood that the traffic information is only illustrated here and is not meant to be limited to only the traffic speed, and other traffic flow information may also be predicted by applying the method of this embodiment.
Wherein, the subgraph division mode is the same as the division mode of the target urban traffic road graph, and the source urban traffic road network G is obtained by adopting the METIS methodSThe source subgraph of (1).
In one embodiment, the source input data further includes source historical traffic status information for the first H time steps, represented as:
Figure BDA0003286110040000094
that is, the source input data of the training samples in the source city training dataset is represented as: [ G ]S，XS]。
The label corresponding to the source input data is the real source traffic state information of the prediction time period corresponding to the source subgraph.
Step S611, inputting the source city training data set into the traffic information prediction model to obtain the source prediction traffic state information.
In one embodiment, a large number of source city training data sets (including the first H time steps) are input into the traffic information prediction model, and the output is predicted traffic status information of the next Q time steps, i.e., source predicted traffic status information, which is expressed as:
Figure BDA0003286110040000095
step S612, a traffic information prediction model is obtained through training according to the error value optimization loss function between the source prediction traffic state information and the label.
In the training process, the performance of the traffic information prediction model is judged according to the error value between the obtained predicted value and the true value (label), a loss function is defined, the loss value of the loss function is calculated according to the error value, the parameters of the traffic information prediction model are adjusted until the loss value meets the preset convergence condition, and then the training of the traffic information prediction model is completed.
Referring to fig. 7, a traffic information prediction flow chart of an embodiment of the present application is shown.
The first is the training phase.
Using source city traffic road map GSAnd (3) training a traffic information prediction model (represented by ST-block in the figure).
A source city training data set (containing a plurality of training samples) is constructed and input.
Source city traffic road map GSWherein G isS＝(VS，ES，AS) Then utilizing METIS method to make source urban traffic road network GSPartitioning into similarly sized MSSource subgraphs (each source subgraph representing a road type).
The source subgraph is represented as:
Figure BDA0003286110040000101
the label of the training sample in the source city training data set is the source traffic state information corresponding to the source subgraph, the source traffic state information corresponding to each source subgraph comprises the source historical traffic state information of the previous H time steps, and the label is expressed as:
Figure BDA0003286110040000102
the set of source traffic status information is represented as:
Figure BDA0003286110040000103
thus, Source City trainingThe source input data in the training samples in the training dataset are represented as: [ G ]Si，XSi]The label is represented as
Figure BDA0003286110040000104
Respectively inputting the training samples into a traffic information prediction model ST-block, training to obtain corresponding source prediction traffic state information, and expressing as follows:
Figure BDA0003286110040000105
according to the obtained predicted value
Figure BDA0003286110040000106
Sum true value (Label)
Figure BDA0003286110040000107
And determining the performance of the traffic information prediction model, defining a loss function, calculating the loss value of the loss function according to the error value, and adjusting the parameters of the traffic information prediction model until the loss value meets the preset convergence condition, so that the training of the traffic information prediction model is finished.
Followed by a prediction phase.
Target city traffic road map G by using trained traffic information prediction model ST-blockTAnd predicting to obtain a traffic information predicted value.
Obtaining a target city traffic road map GTWherein G isT＝(VT，ET，AT) Then utilizing METIS method to make target urban traffic road network GTPartitioning into similarly sized MTTarget subgraphs (each representing a road type). The source subgraph and the target subgraph are similar in size, and model migration is facilitated.
The target subgraph is represented as:
Figure BDA0003286110040000108
the target traffic state information corresponding to each target subgraph comprises target historical traffic state information of the previous H time steps, and is represented as:
Figure BDA0003286110040000109
the set of target traffic status information is represented as:
Figure BDA00032861100400001010
thus, the target input data is represented as: [ G ]Ti，XTi]。
Respectively inputting the target input data into a trained traffic information prediction model ST-block to obtain traffic information prediction values of Q time steps, wherein the traffic information prediction values are represented as follows:
Figure BDA00032861100400001011
referring to fig. 7, the traffic information prediction model ST-block has a structure including a grn and a gated loop unit GRU, an output of the GCN is connected to an input of the GRU, and an output of the GRU is connected to an input of the GRU of the next layer by layer until the last GRU outputs a traffic information prediction value. The GCN captures the spatial information of each pair of input data and inputs the spatial information into the GRU, then the GRU captures the time information of the input data, and a traffic information predicted value is obtained according to the spatial information and the time information.
In one embodiment, in order to improve the prediction accuracy of the traffic information prediction model migrated to the target city with a small data amount, the method further includes: true target traffic status information using predicted time periods (Q time steps)
Figure BDA0003286110040000111
And the above predicted value
Figure BDA0003286110040000112
And adjusting a second loss function according to a second error value, and optimizing parameters of the traffic information prediction model.
In one embodiment, the second loss function includes:
1) mean Absolute Error (MAE), expressed as:
Figure BDA0003286110040000113
2) root mean square error (RMAE), expressed as:
Figure BDA0003286110040000114
3) mean Absolute Percent Error (MAPE), expressed as:
Figure BDA0003286110040000115
and according to the convergence condition, until the second loss value meets the preset convergence condition, completing the optimization of the traffic information prediction model at the moment.
In one embodiment, in the traffic speed prediction task, the traffic information prediction model of the embodiment of the present application is compared with the prediction results of other baseline models, and the comparison result is as follows.
A. Experimental setup:
a) the experimental environment is as follows: all experiments were performed on a Linux server using Intel E5-2620v4 CPU and GeForce RTX 2080Ti CPU. All baseline models and traffic information prediction models of the embodiments of the present application were constructed using Pytorch 1.7.0 and Python 3.8.3.
b) Super parameter setting: future traffic conditions are predicted using H-12 traffic conditions with a short history (60 minutes) and predicted time periods of Q-3/6/9/12 (if one time step is 5 minutes, the corresponding predicted time periods are: 15 minutes/30 minutes/45 minutes/60 minutes, respectively).
An Adam optimizer is used to train the traffic information prediction model. For example, the initial learning rate is set to 0.001, the batch size is 64, and the training time for all models is 500. Specifically, the number of hidden units is set to 32.
The number of subgraph divisions is 8.
B. Data set
Two real network-wide traffic speed datasets are used.
a) Nav-BJ: the data set consisted of average vehicle speeds in beijing, 3 months 1 to 3 months 31 days 2019. The survey data is 1159 nodes (road segments) in Beijing City.
The experimental data contained two main matrices. The first is an adjacency matrix of 1159 x 1159 in size, which describes the spatial relationship between nodes. Each value in the matrix represents whether the node represented by the row is connected to the node represented by the column. The other is a feature matrix, where each row represents the speed of each node at a time. Each column represents the speed of the node from the start time to the end. The timestamp interval is set to 5 minutes.
b) Nav-SH: the data set includes the average velocities in the Shanghai collected simultaneously with the Nav-BJ.
There are 400 nodes (links) in the Shanghai for prediction. It has two matrices like Nav-BJ, the first is a 400X 400 contiguous matrix. The other is a feature matrix with the same format as Nav-BJ.
In this embodiment, the data set from Beijing is treated as one large data set (i.e., the data set of the source city in the migration learning) through the cross-validation process. Meanwhile, Shanghai is the target city to be migrated. Its data is not used for training, but only as real-time input to the traffic information prediction model, and is mapped to range (0,1) using sigmoid activity function after Z-score normalization.
In the optimization process of the traffic information prediction model, the Nav-SH subset of one day is used for retraining the parameters in the traffic information prediction model.
C. Baseline model
The traffic information prediction model of the embodiment of the application is compared with two types of methods: non-migratory learning and migratory learning methods. For the non-transfer learning method, the TGCN model is selected, which is one of the latest techniques for traffic speed prediction.
1) Non-transfer learning:
a) TGCN _ L: the model uses a subset of the target urban traffic data set (Nav-SH) as a training data set. The subset contains 20 days of data and there is a rich source of data that can be used to predict traffic speeds in the future 15-60 minutes.
b) TGCN _ S: the model uses a small portion of the target city traffic data set (Nav-SH) as a training data set. Unlike TGCN-L, this subset contains only one day's data, the same as the data in the traffic information prediction model optimization process. In this case, it is considered that the data set as training is small and it is difficult to predict the traffic speed of 15 to 60 minutes in the future.
2) Transfer learning:
a) traffic information prediction model without sub-graph partitioning: the model predicts the target city using all data of the source city (Nav-BJ) as a training set.
D. Comparison results
As shown in fig. 8, for the performance comparison results of the above models, the prediction time periods are respectively: 15 min/30 min/45 min/60 min, the evaluation indexes include: 1) mean Absolute Error (MAE); 2) root mean square error (RMAE); 3) mean Absolute Percent Error (MAPE).
And (3) performance comparison results:
referring to FIG. 8, a comparison of different methods of predicting 15 min/30 min/45 min/60 min advancement on Nav-SH is shown.
Compared to TGCN _ S, the traffic information prediction model without any data from the target city has been comparable to the performance of TGCN _ S with only a small amount of data, indicating that inter-city data migration is possible.
Compared with the TGCN _ L, the optimized traffic information prediction model has a slightly inferior performance but is superior to the traffic information prediction model. The reason is that the optimized traffic information prediction model uses a small portion of the target city data set. This shows that the migration method can approach the results of models trained using large amounts of data and outperform the computational results of models trained using only a small amount of data (TGCN _ S) using only a small portion of the target city data. This demonstrates the effectiveness of migration.
Compared with the traffic information prediction model without subgraph division, the traffic information prediction model can capture more information in the traffic network graph due to the operation of subgraph division, and the three evaluation indexes are improved compared with the method of direct migration.
As shown in fig. 9, as another performance comparison result of the above models, the number of partitions of the sub-graph partition is: 2. 4, 8 and 16, and the evaluation indexes comprise: 1) mean Absolute Error (MAE); 2) root mean square error (RMAE); 3) mean Absolute Percent Error (MAPE).
And (3) performance comparison results:
in the traffic information prediction model, the traffic network of a target city is divided into several target subgraphs (or clusters) of different sizes. The number of clusters may affect the performance index. In this experiment, the number of clusters was set to 2, 4, 8 and 16, respectively, to see how it affects performance.
With reference to fig. 9, best results are obtained with a number of 8 settings. When the number is 4 and the number is 6, the performance is affected, but not significantly. However, when the number is 2, the performance is much worse. This may be because the two cities do not have exactly the same traffic network and data distribution.
The goal of migration learning is to learn similar graph and time structures from a source city to a target city. After the sub-graph partitioning, each sub-graph is considered to represent one or a class of traffic structures (i.e., intersections, roundabouts, etc. in the road) that exist in each city. When the map is too large, one map may contain various traffic structures, and thus the similarity between the two cities is impaired, so that the effect of migration is poor. When a graph containing a traffic structure or a plurality of similar structures is subjected to subgraph division with proper size, the spatio-temporal information captured by the graph neural network GCN can be better transmitted to a target city, and the prediction effect is better.
Compared with the related technology, the traffic information prediction method includes the steps of obtaining a target city traffic road map, dividing the target city traffic road map into a plurality of target sub-maps, forming target input data by the target sub-maps and corresponding target traffic state information, and inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value, wherein the traffic information prediction model is obtained by training a source city traffic road map in advance, and the number of nodes of the target city traffic road map is less than that of nodes of the source city traffic road network. The traffic information prediction model is trained by using the source urban traffic network data with sufficient data volume, and then applied to the target city with less data volume, so that the traffic information prediction precision and prediction efficiency of the target city can be remarkably improved.
In addition, an embodiment of the present application further provides a traffic information prediction apparatus, and referring to fig. 10, the apparatus includes:
the target city traffic road map acquisition module 101 is configured to acquire a target city traffic road map and divide the target city traffic road map into a plurality of target subgraphs;
the target input data composition module 102 is configured to compose a plurality of target sub-graphs and corresponding target traffic state information into target input data;
and the traffic information prediction module 103 is used for inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value, wherein the traffic information prediction model is obtained by training a source city traffic road map in advance, and the number of nodes of the target city traffic road map is different from that of the source city traffic road network.
The above-described embodiments of the apparatus are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may also be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
It should be noted that the traffic information prediction apparatus in the present embodiment may execute the traffic information prediction method in the embodiment shown in fig. 2. That is, the traffic information prediction apparatus in the present embodiment and the traffic information prediction method in the embodiment shown in fig. 2 both belong to the same inventive concept, and therefore, these embodiments have the same implementation principle and technical effect, and are not described in detail herein.
In addition, an embodiment of the present application further provides a computer device, where the computer device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor.
The processor and memory may be connected by a bus or other means.
The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The non-transitory software programs and instructions required to implement the traffic information prediction method of the above-described embodiment are stored in the memory, and when executed by the processor, perform the traffic information prediction method of the above-described embodiment, for example, perform the above-described method steps S110 to S130 in fig. 2, method steps S111 to S113 in fig. 3, method steps S610 to S612 in fig. 6, and the like.
Furthermore, an embodiment of the present application further provides a computer-readable storage medium, which stores computer-executable instructions, which are executed by a processor or a controller, for example, by a processor in the above-mentioned computer device embodiment, and can make the above-mentioned processor execute the traffic information prediction method in the above-mentioned embodiment, for example, execute the above-mentioned method steps S110 to S130 in fig. 2, method steps S111 to S113 in fig. 3, method steps S610 to S612 in fig. 6, and the like.
For another example, when executed by one of the processors in the above-mentioned computer device embodiment, the processor may be caused to execute the traffic information prediction method in the above-mentioned embodiment, for example, to execute the above-mentioned method steps S110 to S130 in fig. 2, the method steps S111 to S113 in fig. 3, the method steps S610 to S612 in fig. 6, and the like.
One of ordinary skill in the art will appreciate that all or some of the steps, systems, and methods disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
While the preferred embodiments of the present invention have been described in detail, it will be understood, however, that the present invention is not limited to those precise embodiments, and that various changes and modifications may be effected therein by one skilled in the art without departing from the scope of the invention as defined by the appended claims.

Claims (10)

1. A traffic information prediction method, comprising:
acquiring a target city traffic road map, and dividing the target city traffic road map into a plurality of target subgraphs;
forming target input data by the target subgraphs and the corresponding target traffic state information;
and inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value, wherein the traffic information prediction model is obtained by training a source urban traffic road graph in advance, and the number of nodes of the target urban traffic road graph is different from that of the nodes of the source urban traffic road network.
2. The traffic information prediction method of claim 1, wherein the obtaining a target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs comprises:
reducing the size of the target city traffic road map according to different proportions to obtain a plurality of thumbnails in different proportions;
preliminarily dividing the thumbnail with the minimum proportion according to the dividing number to obtain a first target subgraph set;
and dividing the target city traffic road map into corresponding target subgraphs according to the node mapping relation in the first target subgraph set.
3. The traffic information prediction method of claim 2, wherein before obtaining the target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs, the method further comprises:
and when the node number of the target city traffic road graph cannot be divided by the division number, completing the data of the target city traffic road graph by using element zero.
4. The traffic information prediction method of claim 1, wherein the traffic information prediction model comprises a spatial correlation sub-model and a temporal correlation sub-model;
the step of inputting the target input data into a traffic information prediction model trained in advance to obtain a traffic information prediction value comprises the following steps:
acquiring spatial information by using the spatial correlation submodel according to the target input data;
and acquiring time information by using the time correlation sub-model according to the spatial information, and acquiring the traffic information predicted value according to the spatial information and the time information.
5. The traffic information prediction method of claim 4, wherein the spatial correlation sub-model is comprised of a graph neural network model and the temporal correlation sub-model is comprised of a gated cyclic unit model.
6. The traffic information prediction method of claim 1, wherein the training process of the traffic information prediction model comprises:
constructing a source city training data set, the source city training data set comprising: the source input data comprise source subgraphs corresponding to the source city traffic road map and corresponding source traffic state information;
inputting the source city training data set into the traffic information prediction model to obtain source prediction traffic state information;
and optimizing a loss function according to the error value between the source predicted traffic state information and the label, and training to obtain the traffic information prediction model.
7. The traffic information prediction method according to any one of claims 1 to 6, further comprising adjusting a second loss function using a second error value between the predicted value of the traffic information and the real target traffic state information of the predicted time zone to optimize parameters of the traffic information prediction model.
8. A traffic information prediction apparatus, comprising:
the target city traffic road map acquisition module is used for acquiring a target city traffic road map and dividing the target city traffic road map into a plurality of target subgraphs;
the target input data composition module is used for composing a plurality of target subgraphs and corresponding target traffic state information into target input data;
and the traffic information prediction module is used for inputting the target input data into a traffic information prediction model which is trained in advance to obtain a traffic information prediction value, the traffic information prediction model is obtained by training a source urban traffic road graph in advance, and the number of nodes of the target urban traffic road graph is different from that of the source urban traffic road network.
9. A computer device comprising a processor and a memory;
the memory is used for storing programs;
the processor is configured to execute the traffic information prediction method according to any one of claims 1 to 7 according to the program.
10. A computer-readable storage medium storing computer-executable instructions for performing the traffic information prediction method of any one of claims 1 to 7.
CN202111147932.5A 2021-09-29 2021-09-29 Traffic information prediction method, device, equipment and storage medium Pending CN113887704A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111147932.5A CN113887704A (en) 2021-09-29 2021-09-29 Traffic information prediction method, device, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111147932.5A CN113887704A (en) 2021-09-29 2021-09-29 Traffic information prediction method, device, equipment and storage medium

Publications (1)

Publication Number Publication Date
CN113887704A true CN113887704A (en) 2022-01-04

Family

ID=79007726

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111147932.5A Pending CN113887704A (en) 2021-09-29 2021-09-29 Traffic information prediction method, device, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113887704A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114004429A (en) * 2022-01-04 2022-02-01 苏州元澄科技股份有限公司 Data processing method and system for constructing digital city
CN114743379A (en) * 2022-06-13 2022-07-12 广东邦盛北斗科技股份公司 Beidou-based urban large-area road network traffic sensing method and system and cloud platform
CN115034478A (en) * 2022-06-14 2022-09-09 西南交通大学 Traffic flow prediction method based on domain self-adaptation and knowledge migration

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114004429A (en) * 2022-01-04 2022-02-01 苏州元澄科技股份有限公司 Data processing method and system for constructing digital city
CN114743379A (en) * 2022-06-13 2022-07-12 广东邦盛北斗科技股份公司 Beidou-based urban large-area road network traffic sensing method and system and cloud platform
CN115034478A (en) * 2022-06-14 2022-09-09 西南交通大学 Traffic flow prediction method based on domain self-adaptation and knowledge migration

Similar Documents

Publication Publication Date Title
Sun et al. RSOD: Real-time small object detection algorithm in UAV-based traffic monitoring
Sun et al. Predicting citywide crowd flows in irregular regions using multi-view graph convolutional networks
Ali et al. Exploiting dynamic spatio-temporal graph convolutional neural networks for citywide traffic flows prediction
CN110827544B (en) Short-term traffic flow control method based on graph convolution recurrent neural network
Qin et al. A novel combined prediction scheme based on CNN and LSTM for urban PM 2.5 concentration
CN110675623B (en) Short-term traffic flow prediction method, system and device based on hybrid deep learning
Wang et al. DeepSD: Supply-demand prediction for online car-hailing services using deep neural networks
Eslami et al. A real-time hourly ozone prediction system using deep convolutional neural network
Liang et al. Fine-grained urban flow prediction
CN109508360B (en) Geographical multivariate stream data space-time autocorrelation analysis method based on cellular automaton
CN104462190B (en) A kind of online position predicting method excavated based on magnanimity space tracking
CN113887704A (en) Traffic information prediction method, device, equipment and storage medium
CN113705880A (en) Traffic speed prediction method and device based on space-time attention diagram convolutional network
CN110570035B (en) People flow prediction system for simultaneously modeling space-time dependency and daily flow dependency
CN111047078B (en) Traffic characteristic prediction method, system and storage medium
Huang et al. Transfer learning in traffic prediction with graph neural networks
Zhang et al. Multistep speed prediction on traffic networks: A graph convolutional sequence-to-sequence learning approach with attention mechanism
WO2021129668A1 (en) Neural network training method and device
CN114944053A (en) Traffic flow prediction method based on spatio-temporal hypergraph neural network
CN113591380A (en) Traffic flow prediction method, medium and equipment based on graph Gaussian process
EP3685266A1 (en) Power state control of a mobile device
CN115018039A (en) Neural network distillation method, target detection method and device
Xia et al. SW-BiLSTM: a Spark-based weighted BiLSTM model for traffic flow forecasting
Fafoutellis et al. Dilated LSTM networks for short-term traffic forecasting using network-wide vehicle trajectory data
Zheng et al. A deep learning–based approach for moving vehicle counting and short-term traffic prediction from video images

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