CN117131979A - Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism - Google Patents
Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism Download PDFInfo
- Publication number
- CN117131979A CN117131979A CN202311018667.XA CN202311018667A CN117131979A CN 117131979 A CN117131979 A CN 117131979A CN 202311018667 A CN202311018667 A CN 202311018667A CN 117131979 A CN117131979 A CN 117131979A
- Authority
- CN
- China
- Prior art keywords
- traffic flow
- directed
- road
- attention
- flow speed
- 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
- 230000007246 mechanism Effects 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000004927 fusion Effects 0.000 claims abstract description 29
- 239000013598 vector Substances 0.000 claims description 63
- 239000011159 matrix material Substances 0.000 claims description 31
- 238000012549 training Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 14
- 238000001514 detection method Methods 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 7
- 238000003860 storage Methods 0.000 claims description 6
- 230000002776 aggregation Effects 0.000 claims description 5
- 238000004220 aggregation Methods 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 230000004931 aggregating effect Effects 0.000 abstract description 5
- 238000013528 artificial neural network Methods 0.000 abstract description 4
- 230000007547 defect Effects 0.000 abstract description 3
- 230000008034 disappearance Effects 0.000 abstract description 3
- 238000000605 extraction Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 15
- 238000006243 chemical reaction Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000002679 ablation Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- 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
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- 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
-
- 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
- G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- General Business, Economics & Management (AREA)
- Analytical Chemistry (AREA)
- Marketing (AREA)
- Chemical & Material Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a traffic flow speed prediction method and a traffic flow speed prediction system based on a directed hypergraph and a attention mechanism, which belong to the technical field of traffic state prediction and construct the relationship among nodes in a road network based on the directed hypergraph; aggregating the directed superside information to represent complex space-time characteristics of the road network; constructing an encoder and decoder framework; the attention mechanism is fused, the attention module is constructed, and an improved dense connection structure is introduced to improve the accuracy of the method. The method can effectively overcome the defects of the method based on the traditional graph structure, realize extraction and fusion of time sequence and space characteristics, solve the problem of gradient disappearance and the like in the deep neural network, can be suitable for multi-step prediction of traffic flow in a real road network, and can obtain good prediction precision.
Description
Technical Field
The invention relates to the technical field of traffic state prediction, in particular to a traffic flow speed prediction method and system based on a directed hypergraph and a attention mechanism.
Background
Traffic flow speed is the most intuitive parameter that can reflect traffic situation. The accurate prediction of traffic flow speed is the basis for realizing traffic situation deduction, and has great significance for urban traffic management and normal operation. For example, traffic management departments can conduct traffic control in advance based on traffic state prediction results, dredge the traffic jam and reduce potential safety hazards. However, early traffic flow velocity prediction methods mainly model traffic flow based on statistical models, such as historical mean method, integrated moving average autoregressive, and the like. Obviously, the methods are difficult to realize large-scale prediction, can only predict the state of a single sensing node, and have great defects. This is because the space-time relationship in a road traffic system is too complex, and the state of one node can affect the future state of itself and its neighbors at the same time.
Therefore, in recent years, with the development of computing resources and data scale, the mainstream method has been converted into a machine learning method and a neural network method. The neural network model based on the graph has obvious advantages in prediction effect and shows a certain potential. However, it is difficult for the conventional graph to show a complex topological relationship in the actual road network, for example: a complex overpass may connect multiple layers of roads simultaneously, with the same road node being connected to multiple other road nodes. The inability of conventional graphs to accurately and comprehensively model such complex traffic network spatiotemporal relationships greatly limits the performance of conventional graph-based traffic flow speed prediction methods. The hypergraph modeling is used for representing the complex spatial relationship among a plurality of road nodes, has obvious advantages, and accords with the trend of future development. Because of the defects of the traditional graph structure, the complex traffic network space-time relationship is hard to characterize, and the performance of the current traffic flow speed prediction method based on the traditional graph neural network is greatly limited. The traditional traffic flow speed prediction method does not realize the fusion of time dependence and space dependence, and cannot describe the internal correlation of space-time dynamics. The traditional neural network method has the problems that the network is difficult to train due to the fact that the number of network layers is continuously increased, gradient vanishes and the like.
Disclosure of Invention
The invention aims to provide a traffic flow speed prediction method and a traffic flow speed prediction system based on directed hypergraph and attention mechanism, which aim to represent complex spatial relations in real roads and extract and fuse time and spatial information in traffic data more completely, so that accurate and rapid road traffic flow speed prediction can be realized to solve at least one technical problem in the background technology.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
in one aspect, the present invention provides a traffic flow speed prediction method based on directed hypergraph and attention mechanism, comprising:
acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
Optionally, obtaining road network data to obtain a road traffic directional hypergraph g= { V, epsilon, a }, where V represents a road node set, v= { V 1 ,v 2 ,…,v q ,…,v Q Q=1, 2, …, Q being the total number of road nodes, V q Representing a q-th node in the road; epsilon represents the set of directed hyperedges E= (D, T), epsilon= { E 1 ,E 2 ,…,E k ,…,E K K=1, 2, …, K is the total number of directed edges, E k Representing a kth directed superedge in the road, D representing a head of the directed superedge, T representing a tail of the directed superedge, each superedge corresponding to a directed relationship between two sets of nodes (D, T), respectively; a represents an association matrix of a road traffic network,element a ij If 1, the i-th directional overrun header D is represented i Includes the j-th node, if the node is-1, the i-th directed superside tail T is represented i The j-th node is included, if the j-th node is 0, the i-th directed superedge is not associated with the j-th node.
Optionally, aggregating the directed superside information; comprising the following steps:
based on road node speed data X epsilon X at a certain moment in a history step U period, carrying out weighted summation, and respectively extracting traffic flow speed characteristic vectors of the head and the tail in the directed superside;
gating and fusing the extracted traffic flow velocity characteristic vectors of the head and the tail in the directed superside to obtain the corresponding traffic flow velocity characteristic vector of the directed superside;
traffic according to directed overrunExtracting new feature vector of each road node from velocity feature vectorω represents a matrix of output vectors of all road nodes after the aggregation process.
Optionally, defining an encoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism;
matrix H of output vectors (0) H is obtained by outputting through l-1 space-time attention modules in the encoder (l-1) Where L e {1,2,., L }, H (l-1) The spatial attention module output through the first spatial attention module is obtainedThe time attention module output through the first time-space attention module obtains +.>
Will beAnd->Performing gating fusion of space and time characteristics to obtain the output of the first space-time attention moduleThe method comprises the steps of realizing self-adaptive acquisition of the space and time dependence of each traffic node in different time steps;
the output of the encoder is through the full connection layer
Optionally, defining a decoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism;
adaptively selecting correlation characteristics to obtain input H of decoder (L+1) ，H (L+1) H is obtained by outputting through l-1 space-time attention modules in the decoder (L+l) Where L e {1,2,., L }, H (L+l) The spatial attention mechanism output through the first spatiotemporal attention module yields +.>The output of the time attention mechanism via the first spatiotemporal attention module yields +.>
Will beAnd->Performing gating fusion of time and space characteristics to obtain output +.>
The output of the decoder is through the full connection layer
Optionally, the output of the directed hypergraph network, incorporating the attention mechanism, is optimized by the following loss function:
wherein Y is t Is trueThe real value of the real value,for the predicted value, the average absolute error between the predicted value and the true value is minimized.
In a second aspect, the present invention provides a traffic flow speed prediction system based on directed hypergraph and attention mechanism, comprising:
the acquisition module is used for acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
the prediction module is used for processing the acquired traffic flow data by utilizing a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
In a third aspect, the present invention provides a non-transitory computer readable storage medium for storing computer instructions which, when executed by a processor, implement a traffic flow speed prediction method based on directed hypergraph and attention mechanisms as described above.
In a fourth aspect, the present invention provides a computer device comprising a memory and a processor, the processor and the memory being in communication with each other, the memory storing program instructions executable by the processor, the processor invoking the program instructions to perform a traffic flow speed prediction method based on directed hypergraph and attention mechanisms as described above.
In a fifth aspect, the present invention provides an electronic device, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the instructions for implementing the traffic flow speed prediction method based on the directed hypergraph and the attention mechanism as described above.
The invention has the beneficial effects that:
(1) The complex spatial information in the actual road network can be effectively represented. The used directed hypergraph not only can contain all spatial information in the traditional graph structure, but also can more effectively model the directed relationship among a plurality of nodes so as to capture the complex space-time characteristics in the traffic network and finally realize the improvement of accuracy.
(2) The time-space information in the road network can be effectively established. Based on the directed hypergraph and the historical traffic flow data of each node thereof, a deep encoder and decoder architecture, a time-space attention mechanism and a conversion attention mechanism are adopted, so that the study on time-space dependency relationship is enhanced, the extraction and fusion of time sequence and space characteristics are realized, and finally, the traffic flow speed of multiple road nodes in multiple time steps is accurately predicted.
(3) The problem of gradient disappearance in the deep network is overcome. The dense connection structure is introduced, and meanwhile, the concept of a residual error network is used for optimizing the dense connection structure to solve the problem of gradient disappearance of a deep network, avoid parameter redundancy, effectively solve the problem of error accumulation in multi-step prediction and improve the prediction accuracy of the method.
The advantages of additional aspects of the invention will be set forth in part in the description which follows, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an overall traffic flow speed prediction method based on a directed hypergraph and a attention mechanism according to an embodiment of the present invention.
Fig. 2 is a directed hypergraph constructed based on a real road network according to an embodiment of the present invention.
Fig. 3 is a directed hypergraph after directed hyperedge information aggregation according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of a process for aggregating directed superside information according to an embodiment of the present invention.
Fig. 5 is a schematic diagram of a fully-connected layer according to an embodiment of the invention.
FIG. 6 is a schematic diagram of a spatio-temporal attention module according to an embodiment of the present invention.
Fig. 7 is a schematic diagram of a dense connection network according to an embodiment of the present invention.
Fig. 8 is a schematic diagram of a conversion attention module according to an embodiment of the invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements throughout or elements having like or similar functionality. The embodiments described below by way of the drawings are exemplary only and should not be construed as limiting the invention.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless expressly stated otherwise, as understood by those skilled in the art. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, and/or groups thereof.
In order that the invention may be readily understood, a further description of the invention will be rendered by reference to specific embodiments that are illustrated in the appended drawings and are not to be construed as limiting embodiments of the invention.
It will be appreciated by those skilled in the art that the drawings are merely schematic representations of examples and that the elements of the drawings are not necessarily required to practice the invention.
Example 1
In this embodiment 1, there is provided a traffic flow speed prediction system based on directed hypergraph and attention mechanism, including: the acquisition module is used for acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information; the prediction module is used for processing the acquired traffic flow data by utilizing a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
In this embodiment, by using the system, a traffic flow speed prediction method based on a directed hypergraph and a attention mechanism is implemented, including: acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information; processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
Obtaining road network data to obtain a road traffic directional hypergraph G= { V, epsilon, A }, wherein V represents a road node set, and V= { V 1 ,v 2 ,…,v q ,…,v Q }，q=1, 2, …, Q is the total number of road nodes, V q Representing a q-th node in the road; epsilon represents the set of directed hyperedges E= (D, T), epsilon= { E 1 ,E 2 ,…,E k ,…,E K K=1, 2, …, K is the total number of directed edges, E k Representing a kth directed superedge in the road, D representing a head of the directed superedge, T representing a tail of the directed superedge, each superedge corresponding to a directed relationship between two sets of nodes (D, T), respectively; a represents an association matrix of a road traffic network,element a ij If 1, the i-th directional overrun header D is represented i Includes the j-th node, if the node is-1, the i-th directed superside tail T is represented i The j-th node is included, if the j-th node is 0, the i-th directed superedge is not associated with the j-th node.
Aggregating the directed superside information; comprising the following steps:
based on road node speed data X epsilon X at a certain moment in a history step U period, carrying out weighted summation, and respectively extracting traffic flow speed characteristic vectors of the head and the tail in the directed superside;
gating and fusing the extracted traffic flow velocity characteristic vectors of the head and the tail in the directed superside to obtain the corresponding traffic flow velocity characteristic vector of the directed superside;
extracting new feature vectors of all road nodes according to traffic flow velocity feature vectors with directed supersidesω represents a matrix of output vectors of all road nodes after the aggregation process.
Defining an encoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism;
matrix H of output vectors (0) H is obtained by outputting through l-1 space-time attention modules in the encoder (l-1) Where l e {1, 2., -,L}，H (l-1) the spatial attention module output through the first spatial attention module is obtainedThe time attention module output through the first time-space attention module obtains +.>
Will beAnd->Performing gating fusion of space and time characteristics to obtain the output of the first space-time attention moduleThe method comprises the steps of realizing self-adaptive acquisition of the space and time dependence of each traffic node in different time steps;
the output of the encoder is through the full connection layer
Defining a decoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism;
adaptively selecting correlation characteristics to obtain input H of decoder (L+1) ，H (L+1) H is obtained by outputting through l-1 space-time attention modules in the decoder (L+l) Where L e {1,2,., L }, H (L+l) The spatial attention mechanism output through the first spatiotemporal attention module yields +.>The output of the time attention mechanism via the first spatiotemporal attention module yields +.>
Will beAnd->Performing gating fusion of time and space characteristics to obtain output +.>
The output of the decoder is through the full connection layer
The output of the directed hypergraph network, which merges the attention mechanisms, is optimized by the following loss function:
wherein Y is t To be a true value of the value,for the predicted value, the average absolute error between the predicted value and the true value is minimized.
Example 2
With the improvement of the urban level, urban roads and highways are increasingly complicated, and road network structures have high space-time complexity. The same road node may have a directional traffic relationship with a plurality of other nodes. If such spatial information extraction can be fully and efficiently implemented in the mining of traffic flow velocity characteristics, this will improve the accuracy of traffic flow velocity prediction. Thus, in this embodiment 2, a traffic flow speed prediction method based on a directed hypergraph and an attention mechanism is provided, in which a traffic flow speed prediction model is trained.
As shown in fig. 1. Firstly, a road network of a certain area and sensor arrangement thereof are obtained, and traffic flow data collected by a sensor mainly comprises information such as sensor positions, detection time, pulse numbers, vehicle travelling directions, section traffic flow speeds and the like, so that a road traffic directional hypergraph is constructed, namely, step 1.
Step 1: constructing a road traffic directed hypergraph;
as shown in fig. 2, a road traffic directional hypergraph g= { V, epsilon, a } obtained for obtaining road network data, where V represents a road node set, v= { V } 1 ,v 2 ,…,v q ,…,v Q Q=1, 2, …, Q being the total number of road nodes, Q being 74, v in this embodiment q Representing the q-th node in the road, corresponding to a sensor on the real road; epsilon represents the set of directed hyperedges E= (D, T), epsilon= { E 1 ,E 2 ,…,E k ,…,E K K=1, 2, …, K being the total number of directed edges, in this example K is 81, e k Representing the kth directed superedge in the road, D representing the head of the directed superedge, T representing the tail of the directed superedge, each superedge corresponding to a directed relationship between two sets of nodes (D, T), respectively, as shown in FIG. 3, in combination with the road node v in FIG. 2 8 ,v 23 ,v 25 ,v 26 Intuitively, the conceptual illustration of directed superficiality is described;
a represents an association matrix of a road traffic network,the association relation between the head D and the tail T with the directed superside and the road node can be embodied, and the element a is ij If 1, the i-th directional overrun header D is represented i Includes the j-th node, if the node is-1, the i-th directed superside tail T is represented i The method comprises the steps that the jth node is included, if the jth node is 0, the ith directed superside is not associated with the jth node;
step 2: acquiring tracks according to the acquired dataTraffic flow speed of Q road nodes of road traffic directed hypergraph G in historical step U periodThen the q-th road node is at time t u The traffic flow speed data is +.>U is set to 30 in this embodiment;
defining the future traffic flow speed of all Q nodes of the road traffic directional hypergraph G in the period of the prediction step length P asP is set to 12 in the present embodiment;
step 3: FIG. 4 is a schematic diagram showing a process of aggregating directed over-side information;
step 3.1: based on the road node speed data X epsilon X at a certain moment in the historical step U period, carrying out weighted summation according to formulas (1) and (2), and respectively extracting traffic flow speed characteristic vectors of the head and tail in the directed supersideAnd
in the middle ofAnd->Are all learnable parameters, < >>The Hadamard product is indicated;
step 3.2: will omega D And omega T Performing gate control fusion through the formula (3) and the formula (4) to obtain a corresponding traffic flow velocity characteristic vector with a directional overrun
ω DT ＝(1-z 1 )⊙ω D +z 1 ⊙ω T (3)
Z in 1 In order to be able to control the gate, are all learnable parameters, and sigma (·) represents a sigmoid activation function;
step 3.3: according to the traffic flow velocity characteristic vector omega with directed superside DT Extracting new feature vector of each road nodeOmega represents a matrix formed by output vectors of all road nodes after the aggregation process of the step 3.1 and the step 3.2, and the calculation method is shown in a formula (5);
step 4: the traffic flow speed data in the historical step U time period of the Q road nodes are processed in the step 3 to obtainThen the q-th road node is at time t u The new traffic flow velocity feature vector is +.>
Step 5: inputting w into the full connection layer to change the dimension of the matrix, and transforming to obtainWherein D represents the dimension of the matrix as D dimension, and a schematic diagram of a common full-connection layer is shown in FIG. 5;
step 6: defining an encoder;
step 6.1: defining an encoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism, and the specific constitution of the space-time attention module is shown in figure 6;
in order to focus on verifying the influence of L on the performance of the prediction method provided by the invention, in the embodiment, under the same training configuration, four different values of L are respectively discussed by a control variable method to obtain four prediction models respectively representing MD 1 ,MD 2 ,MD 3 ,MD 4 Comparing the results of the four model ablation experiments, and setting L as an L value corresponding to the model with the best performance;
step 6.2: h (0) H is obtained by outputting through l-1 space-time attention modules in the encoder (l-1) Where L e {1,2,., L }, H (l-1) The spatial attention module output through the first spatial attention module is obtainedThe time attention module output through the first time-space attention module obtains +.>
Step 6.3: will beAnd->Gating fusion of spatial and temporal features is performed by equations (6) and (7) to obtain the output +.>The method comprises the steps of realizing self-adaptive acquisition of the space and time dependence of each traffic node in different time steps;
in the middle ofAre all learnable parameters, z 2 Is gating;
step 6.4: the output of the encoder is through the full connection layer
Step 7: introducing a dense connection structure in the encoder, as shown in fig. 7, which is a schematic diagram of the dense connection structure;
step 7.1: a direct connection structure is added among the L series-connected space-time attention modules, so that the input of each space-time attention module comes from the output of all the previous space-time attention modules, and the information transmission and multiplexing among the parts are realized;
step 7.2: optimizing the concatate operation in the dense connection structure into add operation based on the dimension addition idea in the residual error network so as to avoid parameter explosion and realize balance of performance and parameter quantity;
step 8: defining a conversion attention module to realize modeling of a direct relation between each future time step and the historical time step, as shown in fig. 8, which is a schematic diagram of the conversion attention module;
step 8.1: calculating an arbitrary road node v based on the equation (8) and the equation (9) q Is a predicted time step t of (2) m ＝{t U+1 ,t U+2 ,…,t U+P Sum of historic time step t u ＝{t 1 ,t 2 ,…,t U Correlation of };
in the middle ofAnd->Respectively representing different nonlinear mapping functions in the N < th > E {1,2, …, N } attention mechanisms, as shown in a formula (10);
f(x)＝ReLU(xW+b) (10)
wherein W and b are learnable parameters, and ReLU (·) is an activation function;
step 8.2: based on the transition attention score obtained in step 8.1, H (L) Adaptively selecting relevant features in U historical time steps and then converting to decoder inputs via a conversion attention module
Step 9: defining a decoder;
step 9.1: defining a decoder formed by connecting L space-time attention modules in series, wherein one space-time attention module comprises a space attention module, a time attention module and a gating fusion mechanism;
step 9.2: h (L+1) H is obtained by outputting through l-1 space-time attention modules in the decoder (L+l) Where L e {1,2,., L }, H (L+l) The spatial attention mechanism output through the first spatial attention module is obtainedThe output of the time attention mechanism via the first spatiotemporal attention module yields +.>
Step 9.3: will beAnd->The gating fusion of the time and space characteristics is carried out by the formulas (8) and (9) to obtain the output of the first space-time attention module>
In the middle ofAre all learnable parameters, z 3 Is gating;
step 9.4: the output of the decoder is through the full connection layer
Step 10: step 7, introducing a dense connection structure into the decoder;
step 10: defining an output module;
an output module is formed by adopting a full-connection layer,generating prediction results of P prediction steps through an output module>
Step 11: defining a loss function for optimization;
the output of the directed hypergraph network defining the fused attention mechanism is optimized by the following loss function;
wherein Y is t To be a true value of the value,performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing an average absolute error between the predicted value and the true value;
step 12: training the model by adopting an Adam optimizer, setting the initial learning rate to be 0.001, training for 200 rounds, and setting the batch processing size to be 16; for the actual data set, the training set, the validation set and the test set may be divided by using a ratio of 7:1:2, and the accuracy of model training in this embodiment may be quantified using mean squared error (MAE), mean Absolute Percentage Error (MAPE), root Mean Square Error (RMSE) as evaluation indexes.
Example 3
Embodiment 3 provides a non-transitory computer readable storage medium for storing computer instructions that, when executed by a processor, implement a traffic flow speed prediction method based on directed hypergraph and attention mechanisms as described above, the method comprising:
acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
Example 4
Embodiment 4 provides a computer device including a memory and a processor, the processor and the memory being in communication with each other, the memory storing program instructions executable by the processor, the processor invoking the program instructions to perform a traffic flow speed prediction method based on directed hypergraph and attention mechanisms as described above, the method comprising:
acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
Example 5
Embodiment 5 provides an electronic apparatus including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes instructions for implementing the traffic flow speed prediction method based on the directed hypergraph and the attention mechanism as described above, and the method comprises:
acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While the foregoing description of the embodiments of the present invention has been presented in conjunction with the drawings, it should be understood that it is not intended to limit the scope of the invention, but rather, it should be understood that various changes and modifications could be made by one skilled in the art without the need for inventive faculty, which would fall within the scope of the invention.
Claims (10)
1. A traffic flow speed prediction method based on directed hypergraph and attention mechanism, comprising:
acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
processing the acquired traffic flow data by using a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
2. The traffic flow speed prediction method based on the directed hypergraph and the attention mechanism according to claim 1, wherein the road network data is acquired to obtain a road traffic directed hypergraph g= { V, epsilon, a }, wherein V represents a road node set, v= { V 1 ,v 2 ,…,v q ,…,v Q Q=1, 2, …, Q being the total number of road nodes, V q Representing a q-th node in the road; epsilon represents the set of directed hyperedges E= (D, T), epsilon= { E 1 ,E 2 ,…,E k ,…,E K K=1, 2, …, K is the total number of directed edges, E k Representing a kth directed superedge in the road, D representing a head of the directed superedge, T representing a tail of the directed superedge, each superedge corresponding to a directed relationship between two sets of nodes (D, T), respectively; a represents an association matrix of a road traffic network,element a ij If 1, the i-th directional overrun header D is represented i Includes the j-th node, if the node is-1, the i-th directed superside tail T is represented i Includes the j-th node, if 0, the i-th directed superedge is representedIs not associated with the j-th node.
3. The traffic flow speed prediction method based on directed hypergraph and attention mechanism according to claim 2, wherein the directed hyperside information is aggregated; comprising the following steps:
based on road node speed data X epsilon X at a certain moment in a history step U period, carrying out weighted summation, and respectively extracting traffic flow speed characteristic vectors of the head and the tail in the directed superside;
gating and fusing the extracted traffic flow velocity characteristic vectors of the head and the tail in the directed superside to obtain the corresponding traffic flow velocity characteristic vector of the directed superside;
extracting new feature vectors of all road nodes according to traffic flow velocity feature vectors with directed supersidesω represents a matrix of output vectors of all road nodes after the aggregation process.
4. The traffic flow speed prediction method based on directed hypergraph and attention mechanism according to claim 2, wherein an encoder is defined to be composed of L spatiotemporal attention modules in series, one spatiotemporal attention module comprising a spatial attention module, a temporal attention module and a gating fusion mechanism;
matrix H of output vectors (0) H is obtained by outputting through l-1 space-time attention modules in the encoder (l-1) Where L e {1,2,., L }, H (l-1) The spatial attention module output through the first spatial attention module is obtainedThe time attention module output through the first time-space attention module obtains +.>
Will beAnd->Performing gating fusion of space and time characteristics to obtain the output of the first space-time attention moduleThe method comprises the steps of realizing self-adaptive acquisition of the space and time dependence of each traffic node in different time steps;
the output of the encoder is through the full connection layer
5. The traffic flow speed prediction method based on directed hypergraph and attention mechanism according to claim 4, wherein a decoder is defined to be composed of L spatiotemporal attention modules in series, one spatiotemporal attention module comprising a spatial attention module, a temporal attention module and a gating fusion mechanism;
adaptively selecting correlation characteristics to obtain input H of decoder (L+1) ，H (L+1) H is obtained by outputting through l-1 space-time attention modules in the decoder (L+l) Where L e {1,2,., L }, H (L+l) The spatial attention mechanism output through the first spatiotemporal attention module yields +.>The output of the time attention mechanism via the first spatiotemporal attention module yields +.>
Will beAnd->Performing gating fusion of time and space characteristics to obtain output +.>
The output of the decoder is through the full connection layer
6. The traffic flow speed prediction method based on directed hypergraph and attention mechanism according to claim 4, wherein the output of the directed hypergraph network incorporating the attention mechanism is optimized by the following loss function:
wherein Y is t To be a true value of the value,for the predicted value, the average absolute error between the predicted value and the true value is minimized.
7. A traffic flow speed prediction system based on directed hypergraph and attention mechanism, comprising:
the acquisition module is used for acquiring traffic flow data, including sensor position, detection time, pulse number, vehicle travelling direction and section traffic flow velocity information;
the prediction module is used for processing the acquired traffic flow data by utilizing a pre-trained traffic flow speed prediction model to obtain a traffic flow speed prediction result; the training of the pre-trained traffic flow velocity prediction model comprises the following steps: constructing a road traffic directed hypergraph; acquiring the traffic flow speed of a road traffic directional hypergraph road node in a historical step length period; carrying out weighted summation based on road node speed data at a certain moment in a historical step length period, respectively extracting traffic flow speed characteristic vectors of the head and the tail in a road traffic directional hypergraph, carrying out gate control fusion to obtain corresponding traffic flow speed characteristic vectors of the directional hyperedge, extracting new characteristic vectors of each road node, outputting a matrix formed by the vectors, inputting a matrix formed by connecting L space-time attention modules in series to form an encoder, and adaptively acquiring the space and time dependence of each traffic node at different time steps; the input is formed by connecting L space-time attention modules in series to form a decoder, an output module is formed by adopting a full-connection layer, and prediction results of a plurality of prediction step sizes are generated through the output module; and defining a loss function for optimization, and performing end-to-end training on the directed hypergraph network based on an error back propagation mechanism by minimizing the average absolute error between the predicted value and the true value to obtain a traffic flow velocity prediction model.
8. A non-transitory computer readable storage medium storing computer instructions which, when executed by a processor, implement the directional hypergraph and attention mechanism based traffic flow speed prediction method of any one of claims 1-6.
9. A computer device comprising a memory and a processor, the processor and the memory being in communication with each other, the memory storing program instructions executable by the processor, the processor invoking the program instructions to perform the directed hypergraph and attention mechanism based traffic flow speed prediction method according to any of claims 1-6.
10. An electronic device, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and wherein the computer program is stored in the memory, which processor executes the computer program stored in the memory when the electronic device is running, to cause the electronic device to execute instructions implementing the directed hypergraph and attention mechanism based traffic flow speed prediction method according to any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311018667.XA CN117131979A (en) | 2023-08-14 | 2023-08-14 | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311018667.XA CN117131979A (en) | 2023-08-14 | 2023-08-14 | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117131979A true CN117131979A (en) | 2023-11-28 |
Family
ID=88852006
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311018667.XA Pending CN117131979A (en) | 2023-08-14 | 2023-08-14 | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117131979A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117787867A (en) * | 2024-02-27 | 2024-03-29 | 山东财经大学 | Medicine inventory demand analysis method and system |
-
2023
- 2023-08-14 CN CN202311018667.XA patent/CN117131979A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117787867A (en) * | 2024-02-27 | 2024-03-29 | 山东财经大学 | Medicine inventory demand analysis method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110164128B (en) | City-level intelligent traffic simulation system | |
| CN109816095B (en) | Network flow prediction method based on improved gated cyclic neural network | |
| CN113053115B (en) | Traffic prediction method based on multi-scale graph convolution network model | |
| CN113313947B (en) | Road condition evaluation method of short-term traffic prediction graph convolution network | |
| CN111612243B (en) | Traffic speed prediction method, system and storage medium | |
| CN111325976B (en) | Short-term traffic flow prediction method and system | |
| CN114299723B (en) | Traffic flow prediction method | |
| CN113905391A (en) | Ensemble learning network traffic prediction method, system, device, terminal, and medium | |
| CN113361680A (en) | Neural network architecture searching method, device, equipment and medium | |
| CN113762595A (en) | Traffic time prediction model training method, traffic time prediction method and equipment | |
| CN114495500B (en) | Traffic prediction method based on dual dynamic space-time diagram convolution | |
| CN111597175B (en) | Filling method of sensor missing value fusing time-space information | |
| CN111612125A (en) | Novel HTM time pool method and system for online learning | |
| CN117131979A (en) | Traffic flow speed prediction method and system based on directed hypergraph and attention mechanism | |
| CN112766603A (en) | Traffic flow prediction method, system, computer device and storage medium | |
| CN110309537A (en) | A kind of the intelligent health prediction technique and system of aircraft | |
| CN115935796A (en) | Time-space heterogeneous and synchronous graph convolution network traffic flow prediction method | |
| CN113808396A (en) | Traffic speed prediction method and system based on traffic flow data fusion | |
| CN114565187A (en) | Traffic network data prediction method based on graph space-time self-coding network | |
| Huo et al. | Cooperative control for multi-intersection traffic signal based on deep reinforcement learning and imitation learning | |
| CN116307152A (en) | Traffic prediction method for space-time interactive dynamic graph attention network | |
| Chang et al. | Differentiable architecture search with ensemble gumbel-softmax | |
| CN115080795A (en) | Multi-charging-station cooperative load prediction method and device | |
| CN115761654B (en) | Vehicle re-identification method | |
| Jiang et al. | Dynamic adaptive and adversarial graph convolutional network for traffic forecasting |
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 |