CN115186047B - Traffic flow dynamic diagram reconstruction method, related device and computer program product - Google Patents
Traffic flow dynamic diagram reconstruction method, related device and computer program product Download PDFInfo
- Publication number
- CN115186047B CN115186047B CN202210837366.9A CN202210837366A CN115186047B CN 115186047 B CN115186047 B CN 115186047B CN 202210837366 A CN202210837366 A CN 202210837366A CN 115186047 B CN115186047 B CN 115186047B
- Authority
- CN
- China
- Prior art keywords
- flow
- node
- implicit
- auxiliary
- dynamic graph
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000010586 diagram Methods 0.000 title claims description 25
- 238000004590 computer program Methods 0.000 title abstract description 15
- 238000012512 characterization method Methods 0.000 claims abstract description 83
- 230000004927 fusion Effects 0.000 claims abstract description 44
- 239000011159 matrix material Substances 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 32
- 238000009826 distribution Methods 0.000 claims description 29
- 230000008569 process Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000005206 flow analysis Methods 0.000 claims description 9
- 239000004973 liquid crystal related substance Substances 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 abstract description 2
- 239000010410 layer Substances 0.000 description 39
- 230000006870 function Effects 0.000 description 13
- 238000004891 communication Methods 0.000 description 11
- 230000001537 neural effect Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000000295 complement effect Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000001580 bacterial effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000008571 general function Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Quality & Reliability (AREA)
- Remote Sensing (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Traffic Control Systems (AREA)
Abstract
The disclosure provides a traffic flow dynamic graph reconstruction method, a related device and a computer program product, and relates to the technical field of artificial intelligence such as deep learning, dynamic graph reconstruction, variation inference and the like. The method comprises the following steps: obtaining each time snapshot forming a target dynamic graph, wherein the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises target snapshots with partial node information missing; generating an implicit representation of the suction flow and an implicit representation of the sending flow of each node according to the snapshots of each moment; fusing the implicit flow characterization and the nonlinear characteristic of the implicit flow characterization to obtain a fused characteristic; processing the fusion characteristics of each node by using a multi-layer perceptron to obtain a starting point and end point flow estimation matrix; and reconstructing partial node information missing from the target snapshot based on traffic flow information of any starting point and ending point pair in the starting point and ending point flow estimation matrix. By the method, partial missing node information in the dynamic graph can be accurately complemented.
Description
Technical Field
The disclosure relates to the technical field of data processing, in particular to the technical field of artificial intelligence such as deep learning, dynamic graph reconstruction and variation inference, and particularly relates to a traffic flow dynamic graph reconstruction method, a traffic flow dynamic graph reconstruction device, electronic equipment, a computer readable storage medium and a computer program product.
Background
An OD flow matrix (Origin-Destination Matrix Completion, chinese is called as a starting point and end point flow estimation matrix) is a characterization mode for describing flow between different observation nodes of a city (the observation nodes can be towns/communities with coarse granularity, intersections with finer granularity and the like).
Due to uneven quality of sensor data, observation data of partial nodes may be lost, so that an OD flow matrix is required to be completed in a reconstruction mode, and the completed OD flow matrix is further used for situation research and judgment of urban traffic flow.
Disclosure of Invention
The embodiment of the disclosure provides a traffic flow dynamic diagram reconstruction method, a traffic flow dynamic diagram reconstruction device, electronic equipment, a computer readable storage medium and a computer program product.
In a first aspect, an embodiment of the present disclosure provides a traffic flow dynamic map reconstruction method, including: obtaining snapshots of all moments forming a target dynamic graph; the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises a target snapshot with partial node information missing; generating an implicit representation of the suction flow when each node is used as a suction object and an implicit representation of the emission flow when each node is used as an emission object according to the snapshots of each moment; fusing the implicit flow characterization and the nonlinear characteristic of the implicit flow characterization to obtain a fused characteristic; processing the fusion characteristics of each node by using a multi-layer perceptron to obtain a starting point and end point flow estimation matrix; and reconstructing partial node information missing from the target snapshot based on traffic flow information of any starting point and ending point pair in the starting point and ending point flow estimation matrix.
In a second aspect, an embodiment of the present disclosure provides a traffic flow dynamic map reconstruction device, including: a dynamic graph snapshot acquisition unit configured to acquire snapshots of respective times constituting a target dynamic graph; the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises a target snapshot with partial node information missing; the implicit representation generation unit is configured to generate an implicit representation of the suction flow when each node is used as a suction object and an implicit representation of the emission flow when each node is used as an emission object according to the snapshots of each moment; the nonlinear feature fusion unit is configured to fuse the implicit flow characterization and the nonlinear feature for sending the implicit flow characterization to obtain a fusion feature; the multi-layer perceptron processing unit is configured to process the fusion characteristics of each node by utilizing the multi-layer perceptron to obtain a starting point and end point flow estimation matrix; and the missing information reconstruction unit is configured to reconstruct partial node information missing from the target snapshot based on the traffic flow information of any starting point-end point pair in the starting point-end point flow estimation matrix.
In a third aspect, an embodiment of the present disclosure provides an electronic device, including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to implement a traffic flow dynamics map reconstruction method as described in any one of the implementations of the first aspect when executed.
In a fourth aspect, embodiments of the present disclosure provide a non-transitory computer-readable storage medium storing computer instructions for enabling a computer to implement a traffic flow dynamics map reconstruction method as described in any one of the implementations of the first aspect when executed.
In a fifth aspect, embodiments of the present disclosure provide a computer program product comprising a computer program which, when executed by a processor, is capable of implementing the steps of a traffic flow dynamics map reconstruction method as described in any one of the implementations of the first aspect.
According to the traffic flow dynamic graph reconstruction scheme provided by the disclosure, through determining the implicit representation of the attraction flow of each node as the attraction object and the implicit representation of the emission flow of the emission object according to the snapshots of each moment forming the traffic flow dynamic graph, the flow characteristics of each node in a traffic scene can be fully matched, nonlinear factors under actual conditions are fully considered by the fusion characteristics through nonlinear processing and fusion, the fusion characteristics of each node are processed through the multi-layer perceptron, and as the multi-layer perceptron shares own perceptron parameters to all nodes, the characteristics of each node can be better combined, so that the reconstructed starting point and end point flow estimation matrix can contain more accurate missing information, and good complementation of missing parts is further realized.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.
Drawings
Other features, objects and advantages of the present disclosure will become more apparent upon reading of the detailed description of non-limiting embodiments, made with reference to the following drawings:
FIG. 1 is an exemplary system architecture in which the present disclosure may be applied;
fig. 2 is a flowchart of a traffic flow dynamic diagram reconstruction method according to an embodiment of the present disclosure;
FIG. 3 is a flow chart of a method of reconstructing a traffic flow dynamics diagram provided by an embodiment of the present disclosure;
FIG. 4 is a flow chart of another method for reconstructing a traffic flow dynamic diagram provided by an embodiment of the present disclosure;
FIG. 5 is a flow chart of a method of analyzing traffic flow provided by an embodiment of the present disclosure;
fig. 6 is a flow chart of a traffic flow dynamic graph reconstruction method under an application scenario according to an embodiment of the present disclosure;
fig. 7 is a block diagram of a traffic flow dynamic diagram reconstruction device according to an embodiment of the present disclosure;
Fig. 8 is a schematic structural diagram of an electronic device adapted to perform a traffic flow dynamic map reconstruction method according to an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness. It should be noted that, without conflict, the embodiments of the present disclosure and features of the embodiments may be combined with each other.
In the technical scheme of the disclosure, the related processes of collecting, storing, using, processing, transmitting, providing, disclosing and the like of the personal information of the user accord with the regulations of related laws and regulations, and the public order colloquial is not violated.
FIG. 1 illustrates an exemplary system architecture 100 to which embodiments of traffic flow dynamics graph reconstruction methods, apparatus, electronic devices, and computer readable storage media of the present disclosure may be applied.
As shown in fig. 1, a system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various applications for implementing information communication between the terminal devices 101, 102, 103 and the server 105, such as a traffic flow analysis type application, a dynamic graph reconstruction type application, an instant messaging type application, and the like, may be installed on the terminal devices.
The terminal devices 101, 102, 103 and the server 105 may be hardware or software. When the terminal devices 101, 102, 103 are hardware, they may be various electronic devices with display screens, including but not limited to smartphones, tablets, laptop and desktop computers, etc.; when the terminal devices 101, 102, 103 are software, they may be installed in the above-listed electronic devices, which may be implemented as a plurality of software or software modules, or may be implemented as a single software or software module, which is not particularly limited herein. When the server 105 is hardware, it may be implemented as a distributed server cluster formed by a plurality of servers, or may be implemented as a single server; when the server is software, the server may be implemented as a plurality of software or software modules, or may be implemented as a single software or software module, which is not particularly limited herein.
The server 105 may provide various services through various built-in applications, for example, a dynamic graph reconstruction type application that may provide a reconstruction service for a traffic flow dynamic graph, and the server 105 may implement the following effects when running the dynamic graph reconstruction type application: and supplementing the missing part of node information through a dynamic diagram reconstruction scheme according to the received traffic flow dynamic diagram with the missing part of node information.
It should be noted that the traffic flow dynamics map in which the partial node information is missing may be obtained from the terminal devices 101, 102, 103 through the network 104, or may be stored in advance in the server 105 in various ways. Thus, when the server 105 detects that such data has been stored locally (e.g., pending tasks left until processing is initiated), it may choose to retrieve the data directly from the local, in which case the exemplary system architecture 100 may not include the terminal devices 101, 102, 103 and network 104.
Since the dynamic graph reconstruction needs to occupy more operation resources and stronger operation capability, the traffic flow dynamic graph reconstruction method provided in the following embodiments of the present disclosure is generally executed by the server 105 having stronger operation capability and more operation resources, and accordingly, the traffic flow dynamic graph reconstruction device is also generally disposed in the server 105. However, it should be noted that, when the terminal devices 101, 102, 103 also have the required computing capability and computing resources, the terminal devices 101, 102, 103 may also complete each operation performed by the server 105 through the dynamic image reconstruction class application installed thereon, and further output the same result as the server 105. Especially, in the case that there are multiple terminal devices with different computing capabilities at the same time, when the dynamic graph reconstruction application determines that the terminal device where the dynamic graph reconstruction application is located has a stronger computing capability and more computing resources remain, the terminal device can execute the above computation, so that the computing pressure of the server 105 is properly reduced, and correspondingly, the traffic flow dynamic graph reconstruction device can also be set in the terminal devices 101, 102, 103. In this case, the exemplary system architecture 100 may also not include the server 105 and the network 104.
It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring to fig. 2, fig. 2 is a flowchart of a traffic flow dynamic map reconstruction method according to an embodiment of the disclosure, wherein the flowchart 200 includes the following steps:
step 201: obtaining snapshots of all moments forming a target dynamic graph;
the step aims at acquiring, by an execution body (for example, a server 105 shown in fig. 1) of the traffic flow dynamic map reconstruction method, a snapshot of each time constituting a target dynamic map, where the target dynamic map is used for recording traffic flow information, and may also be simply referred to as a traffic flow dynamic map, and each snapshot of each time includes a target snapshot in which part of node information is missing.
In general, a dynamic graph is formed by at least two snapshots at time, and assuming that the target dynamic graph is formed by T snapshots from time 1 to time T, the target snapshot is usually a snapshot at other time except for the first snapshot, and the high probability is the last T snapshot, so that the information recorded in the previous multiple snapshots at time is fully utilized to complement the exact partial node information in the T snapshots.
It should be noted that, since the target dynamic graph is a dynamic graph for recording traffic flow information, and the traffic flow information is described by interactions between traffic objects appearing in the target area, the traffic objects include red and green lights, vehicles, stop boards, intersections, buildings, and the like, that is, each traffic object will be an independent node, and the traffic flow is depicted by the track information between the nodes.
Step 202: generating an implicit representation of the suction flow when each node is used as a suction object and an implicit representation of the emission flow when each node is used as an emission object according to the snapshots of each moment;
on the basis of step 201, this step aims at generating, by the execution body, an implicit characterization of the attraction traffic of each node and an implicit characterization of the emission traffic according to the snapshots of each moment respectively.
The implicit representation of the node attraction flow is the implicit representation of the flow which is presented when the node is used as an attraction object for attracting other nodes to arrive under a traffic scene; conversely, an implicit representation of the node's outgoing traffic is an implicit representation of the traffic that the node presents when the node is in a traffic scenario as an outgoing object that is concurrent to other nodes. In short, the suction flow rate and the discharge flow rate correspond to the flow rates exhibited by the two actions of "in" and "out", respectively.
Implicit characterization of the self-implicit neural representation (Implicit Neural Representations) is a new method of parameterizing various signals. Traditional signal representations are typically discrete, e.g., the image is a discrete grid of pixels, the audio signal is a discrete sample of amplitude, and the three-dimensional shape is typically parameterized as a grid of voxels, point clouds, or grids; in contrast, implicit neural representation parameterizes a signal as a continuous function that maps a signal domain (i.e., coordinates, such as pixel coordinates of an image) to any location at that coordinate (R, G, B color for an image). Of course, these functions are typically not processable, i.e. it is not possible to "write down" the function parameterizing the natural image into a mathematical formula. Therefore, the implicit neural representation approaches the 'natural representation' function through the neural network, and the representation which is as close to the actual situation as possible and is constructed based on the implicit neural representation is the implicit representation.
That is, this step generates an implicit characterization of the suction flow and an implicit characterization of the delivery flow, respectively, which are expected to be better presented by utilizing the implicit neural representation of the characteristics of the suction flow and the delivery flow.
Step 203: fusing the implicit flow characterization and the nonlinear characteristic of the implicit flow characterization to obtain a fused characteristic; based on step 202, the execution subject first uses the nonlinear layer to process the implicit characterization of the attraction flow and the implicit characterization of the emission flow to obtain the nonlinear characteristics of the attraction flow and the nonlinear characteristics of the emission flow, and finally fuses the two nonlinear characteristics of the same node.
Based on step 202, the execution subject first uses the nonlinear layer to process the implicit characterization of the attraction flow and the implicit characterization of the emission flow to obtain the nonlinear characteristics of the attraction flow and the nonlinear characteristics of the emission flow, and finally fuses the two nonlinear characteristics of the same node.
The implicit characterization of the attraction flow and the emission flow is processed by the nonlinear layer, and the traffic flow characteristics of the actual situation are not suitable for fitting by using a linear mode, and are more suitable for fitting by using a nonlinear mode, so that the nonlinear characteristics obtained by processing can be closer to the actual situation.
And the nonlinear characteristics of the suction flow and the emission flow are fused, so that the traffic flow characteristics contained in the fused characteristics are more comprehensive and accurate by combining different characteristics of the same node when the same node is used as the suction object and the emission object respectively.
Step 204: processing the fusion characteristics of each node by using a multi-layer perceptron to obtain a starting point and end point flow estimation matrix;
based on step 203, this step aims at processing, by the execution body, the fusion features of each node by using the multi-layer perceptron to obtain a starting point and ending point flow estimation matrix (i.e. an OD flow matrix).
The basic structure of a multi-layer perceptron (Multilayer Perceptron, MLP) is available based on a biological neuron model, most typically the MLP comprises three layers: input layer, hidden layer (hidden layer) and output layer (i.e. one to a plurality of hidden layers are introduced on the basis of a single-layer neural network, the hidden layers are positioned between the input layer and the output layer), and the different layers of the MLP neural network are fully connected, wherein the fully connected means: any one neuron of the upper layer is connected to all neurons of the lower layer, i.e. based on this property of the multi-layer perceptron, perceptron parameters can be shared to all nodes.
Step 205: and reconstructing partial node information missing from the target snapshot based on traffic flow information of any starting point and ending point pair in the starting point and ending point flow estimation matrix.
Based on step 204, this step aims to reconstruct the partial node information missing from the target snapshot based on the traffic flow information of any starting point-end point pair in the starting point-end point flow estimation matrix by the execution subject.
The arbitrary starting point and end point pair is an OD flow point pair formed by any two nodes in the OD flow matrix, one of the any two nodes is selected as a starting point, the other is selected as an end point, and usually, the selected one node is taken as the starting point, and the selected other node is taken as the end point.
According to the traffic flow dynamic graph reconstruction method provided by the embodiment of the disclosure, through determining the implicit representation of the suction flow of each node as the suction object and the implicit representation of the emission flow of the emission object according to the snapshots of each moment forming the traffic flow dynamic graph, the flow characteristics of each node in a traffic scene can be fully matched, nonlinear factors in actual conditions are fully considered by the fusion characteristics through nonlinear processing and fusion, the fusion characteristics of each node are processed through a multi-layer perceptron, and as the multi-layer perceptron shares own perceptron parameters to all nodes, the characteristics of each node can be better combined, so that the reconstructed starting point and end point flow estimation matrix can contain more accurate missing information, and good completion of missing parts is further realized.
Referring to fig. 3, fig. 3 is a flowchart of a method for reconstructing a traffic flow dynamic diagram according to an embodiment of the present disclosure, in which a specific implementation is provided for step 205 in the flowchart 200 shown in fig. 2, and other steps in the flowchart 200 are not adjusted, so that a new complete embodiment can be obtained by replacing step 205 with the specific implementation provided in the present embodiment. Wherein the process 300 comprises the steps of:
Step 301: constructing a travel event by taking any node in the starting point and end point flow estimation matrix as a starting point and any node as an end point;
the step aims at selecting any node from the starting point and end point flow estimation matrix by the execution main body as a starting point and any node as an end point, so as to construct a travel event between the starting point and the key point. Typically, the start point and the end point are not the same.
Step 302: taking poisson distribution as a desired solving mode of an evidence lower bound mode of a calculation variation inference principle, and fitting target traffic flow information between a starting point and a terminal point in a travel event;
the variational inference (Variational Inference, VI) is a large class of methods in bayesian approximation inference methods, which skillfully transform the posterior inference problem into an optimization problem for solving. In the application of probability models, one central task is to calculate the posterior probability distribution of the latent variable Z given the observed (visible) data variable X, and to calculate the expectations about this probability distribution. For many models in practical applications, it is not feasible to calculate the posterior probability distribution or calculate the expectations about this posterior probability distribution. This may be due to the fact that the dimensions of the potential space are too high to be directly calculated, or because the form of the posterior probability distribution is particularly complex, so that it is desirable to not get an analytical solution.
For the general function f (x), f can be considered as a real operator for x, whose role is to map a real number x to a real number f (x). Then analogize to this mode, assuming that there is a function operator F, which is a function operator for F (x), F (x) can be mapped to a real number F (x)). For F (x), the extremum of F (x) can be found by changing x, which in the variation is replaced by a function y (x), i.e. by changing x, and finally letting F (y (x)) find the extremum.
The variation refers to the variation of the functional, which ultimately seeks extremum functions that cause the functional to take a maximum or minimum value. For example, there are numerous paths from point a to point B, each path is a function, the numerous paths are numerous, the length of each function (path) is a number, and you choose one path from the numerous paths to be shortest or longest, which is the extremum problem of solving the functional.
On this basis, variation inference can be understood as follows; a simple distribution is found to approximate the posterior probability density that cannot be solved in the inference problem. The mathematical language indicates that minimization is desired, but because the posterior probability density in KL divergence (Kullback-Leibler divergence) is also indissolvable, the problem is converted from minimization of KL divergence to maximization by means of the derivation thought of EM (maximization-maximization) (Evidence Lower Bound, where evidence refers to the probability density of data or observable variables).
Based on step 301, this step aims at using poisson distribution as a desired calculation mode of the evidence lower bound mode of the calculation variation inference principle by the execution subject, and fitting to obtain the target traffic flow information between the starting point and the end point in the travel event.
The poisson distribution (Poisson Distribution) is adapted to describe the number of random events occurring per unit time (or space). Such as the number of people that a certain service facility arrives in a certain time, the number of times that a telephone exchange receives a call, the number of waiting guests at a car station, the number of faults that a machine has, the number of times that a natural disaster occurs, the number of defects on a product, the number of bacterial distribution in a unit partition under a microscope, and the like. Therefore, the method is more suitable for describing traffic information among traffic objects in the traffic flow analysis scene aimed at by the present disclosure, and the step is to use the poisson distribution characteristic as a desired solving mode of an evidence lower bound mode so as to obtain the target traffic flow information between the starting point and the end point in the travel event by fitting as accurately as possible.
Step 303: and reconstructing the partial node information missing by the target snapshot according to the target traffic flow information related to the partial node missing by the target snapshot.
Based on step 302, this step aims at reconstructing the partial node information missing from the target snapshot by means of the space-time correlation between the nodes according to the target traffic flow information related to the partial nodes missing from the target snapshot, so that the partial node information obtained by reconstruction is as close to the real information under the actual condition as possible.
According to the embodiment, after any OD flow pair in the OD flow matrix is constructed as a travel event, poisson distribution in a traffic flow scene is used as an expected solving mode of an evidence lower bound mode of a calculation variation inference principle, so that accurate target traffic flow information between a starting point and an ending point in the travel event is fitted as much as possible, and the reconstruction effect is finally as good as possible.
It should be noted that, in this embodiment, the poisson distribution under the evidence lower bound mode under the variation inference principle is used to reconstruct information, different distributions under the same calculation mode under the same principle may be used to perform information fitting, and different calculation modes under the same principle may also be used, so long as similar technical effects can be achieved, and the above embodiment is merely provided as an exemplary example without specific limitation.
Referring to fig. 4, fig. 4 is a flowchart of another traffic flow dynamic map reconstruction method according to an embodiment of the disclosure, and a flow 400 thereof includes the following steps:
step 401: obtaining snapshots of all moments forming a target dynamic graph;
step 402: generating an implicit representation of the suction flow when each node is used as a suction object and an implicit representation of the emission flow when each node is used as an emission object according to the snapshots of each moment;
step 403: fusing the implicit flow characterization and the nonlinear characteristic of the implicit flow characterization to obtain a fused characteristic;
the above steps 401-403 are identical to the steps 201-203 shown in fig. 2, and the same parts are referred to the corresponding parts of the previous embodiment, and will not be described herein again.
Step 404: acquiring an auxiliary dynamic graph with different modes from a target dynamic graph, and acquiring snapshots of all moments forming the auxiliary dynamic graph;
the step aims at acquiring an auxiliary dynamic image which belongs to different modes from the target dynamic image by the execution main body, and further acquiring snapshots of all moments forming the auxiliary dynamic image.
The modes of the target dynamic graph and the auxiliary dynamic graph are different, namely, the acquisition channels and the information expression forms of the two dynamic graphs are different, but traffic flow information is recorded, and the different expression forms of the traffic flow dynamic graph can be simply understood, so that the traffic flow information contained in the other mode is expected to complement part of nodes which are missing from the other angle through the different graph expression forms and the different information acquisition channels, and even the data density of the sparse number can be improved.
It should be noted that the number of the auxiliary dynamic images may be multiple, that is, the auxiliary dynamic images of multiple different modes may be combined at the same time, so as to improve the final reconstruction effect.
Step 405: generating an auxiliary suction flow implicit characterization of each node when the node is used as a suction object and an auxiliary emission flow implicit characterization when the node is used as an emission object according to the snapshots of each moment forming the auxiliary dynamic graph;
specifically, each snapshot at each moment forming the auxiliary dynamic graph can be used for generating an auxiliary suction flow implicit characterization of each node when the auxiliary dynamic graph is used as a suction object and an auxiliary emission flow implicit characterization of each node when the auxiliary dynamic graph is used as an emission object through a bayesian graph feature learning method, so that the characteristic capacity of the auxiliary dynamic graph matched with the dynamic graph aimed at by the present disclosure is fully utilized.
Step 406: fusing the auxiliary suction flow implicit characterization and the nonlinear characteristic of the auxiliary emission flow implicit characterization to obtain auxiliary fusion characteristics;
steps 405-405 are similar to steps 402-403 except that one of the sources is a snapshot of each moment that constitutes the target dynamic graph and one of the sources is a snapshot of each moment that constitutes the auxiliary dynamic graph, both for generating the implicit characterization of the attractive flow and for issuing the implicit characterization of the flow, and further for merging the nonlinear features of the two implicit characterizations thereof.
Step 407: processing the splicing characteristics of each node by using a multi-layer perceptron to obtain a starting point and end point flow estimation matrix, wherein the splicing characteristics are obtained by splicing fusion characteristics and auxiliary fusion characteristics;
based on the step 403 and the step 406, the step aims to firstly splice the fusion features of each node from the target dynamic graph and the auxiliary fusion features of each node from the auxiliary dynamic graph by the execution main body, so as to obtain the splice features fusing different modes, and then process the splice features of each node by using a multi-layer perceptron, so as to finally obtain the optimized starting point and end point flow estimation matrix combining different mode information.
Step 408: and reconstructing partial node information missing from the target snapshot based on traffic flow information of any starting point and ending point pair in the starting point and ending point flow estimation matrix.
Step 408 is the same as step 205, and will not be described in detail herein, and the same parts will be described in detail below in step 205.
In contrast to the implementation scheme shown in fig. 2, the present embodiment further adds a technical scheme of combining auxiliary dynamic graphs of other modes through steps 404-406, so as to reconstruct partial node information with real target time better by additionally combining traffic flow information in the auxiliary dynamic graphs of different modes.
Further, considering that the implicit characterization corresponding to the auxiliary dynamic map may follow any prior distribution, to reduce the difference between cross-modalities, the prior distributions of the first implicit characterization, which is the implicit characterization corresponding to the target dynamic map (i.e., the attractive flow implicit characterization and the outgoing flow implicit characterization in steps 202 and 402), and the second implicit characterization, which is the implicit characterization corresponding to the auxiliary dynamic map (i.e., the auxiliary attractive flow implicit characterization and the auxiliary outgoing flow implicit characterization in step 405), may also be aligned.
Furthermore, in order to prevent the information in the auxiliary dynamic graph of a different modality from the target dynamic graph from interfering with the prior distribution of the implicit characterization from the target dynamic graph, the gradient information of the second implicit characterization may be discarded during the back propagation of the training.
On the basis of the above embodiment, the present embodiment further shows a flow chart of a method for analyzing traffic flow through fig. 5, so as to fully consider the actual usage scenario and usage of the completed dynamic diagram, and the flow 500 includes the following steps:
step 501: generating a dynamic graph after completion according to the partial node information missing from the target snapshot obtained by reconstruction;
Step 502: and generating a traffic flow analysis result based on the traffic flow information of each node recorded in the dynamic graph after completion.
According to the embodiment, firstly, a generated post-completion dynamic diagram is obtained according to partial node information missing from the target snapshot obtained through reconstruction, then, based on traffic flow information of each node recorded in the post-completion dynamic diagram of the completion information, a traffic flow analysis result corresponding to actual requirements is generated by combining corresponding traffic flow control areas, and then, based on the analysis result, each traffic control parameter is adjusted or corrected.
For further understanding, the disclosure further provides a specific implementation scheme in combination with a specific application scenario, please refer to a flow chart shown in fig. 6:
1. given a dynamic diagramWherein a snapshot of the dynamic graph at time is recorded as，/>And->Representation->A set of directed edges and a set of nodes. Furthermore, the->Node attribute of->The adjacency matrix is marked->；
2. Obtaining time of day via an encoderImplicit characterization of->Let->Then-> And->Implicit characterization of the attraction flow (in-flow) and the occurrence flow (out-flow), respectively, and further input to the nonlinear layer +_, respectively>And- >；
3. And an end-to-end characterization fusion process is realized by using the multi-layer perceptron to realize the reconstruction of the OD flow matrix. Specifically, after the in-flow and out-flow characterization of each node is obtained, it is vector stitched and fed to an MLP. All nodes here share the parameters of the MLP;
4. by the steps, the flow value of any OD pair can be generated, so that the reconstruction of an OD flow matrix is realized, and the decoding process is recorded as；
Specifically, one trip between any two nodes can be regarded as one eventThe flow values of the OD pairs can be fitted by poisson distribution. Thus, it is possible to obtain the informationThe following posterior distribution:
。
furthermore, the OD flow matrix reconstruction is realized by utilizing the principle of variation inference and utilizing end-to-end learning.
The embodiment can realize the directed graph depth reconstruction method aiming at the special application scene of the urban OD flow matrix complement by using the decoder constructed based on the multi-layer perceptron and utilizing Poisson distribution estimation, thereby effectively improving the complement effect of the missing information in the OD flow matrix.
With further reference to fig. 7, as an implementation of the method shown in the foregoing figures, the present disclosure provides an embodiment of a traffic flow dynamic map reconstruction apparatus, where the embodiment of the apparatus corresponds to the embodiment of the method shown in fig. 2, and the apparatus may be specifically applied to various electronic devices.
As shown in fig. 7, the traffic flow dynamic map reconstruction device 700 of the present embodiment may include: the device comprises a dynamic diagram snapshot acquisition unit 701, an implicit representation generation unit 702, a nonlinear feature fusion unit 703, a multi-layer perceptron processing unit 704 and a missing information reconstruction unit 705. Wherein, the dynamic graph snapshot obtaining unit 701 is configured to obtain snapshots of each time constituting the target dynamic graph; the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises a target snapshot with partial node information missing; an implicit representation generating unit 702 configured to generate an implicit representation of the suction flow rate when each node is the suction object and an implicit representation of the emission flow rate when each node is the emission object according to the snapshots of each time; a nonlinear feature fusion unit 703 configured to fuse the implicit flow characterization and the nonlinear feature that emits the implicit flow characterization to obtain a fused feature; a multi-layer perceptron processing unit 704 configured to process the fusion features of each node by using the multi-layer perceptron to obtain a starting point and end point flow estimation matrix; the missing information reconstructing unit 705 is configured to reconstruct partial node information missing from the target snapshot based on traffic flow information of any of the start-point-end-point pairs in the start-point-end-point traffic flow estimation matrix.
In this embodiment, in the traffic flow dynamic map reconstruction device 700: the specific processing and the technical effects of the dynamic graph snapshot obtaining unit 701, the implicit characteristic generating unit 702, the nonlinear characteristic fusion unit 703, the multi-layer perceptron processing unit 704, and the missing information reconstructing unit 705 may refer to the relevant descriptions of steps 201-205 in the corresponding embodiment of fig. 2, and are not repeated herein.
In some optional implementations of the present embodiment, the missing information reconstruction unit 705 may be further configured to:
constructing a travel event by taking any node in the starting point and end point flow estimation matrix as a starting point and any node as an end point;
taking poisson distribution as a desired solving mode of an evidence lower bound mode of a calculation variation inference principle, and fitting target traffic flow information between a starting point and a terminal point in a travel event;
and reconstructing the partial node information missing by the target snapshot according to the target traffic flow information related to the partial node missing by the target snapshot.
In some optional implementations of the present embodiment, the traffic flow dynamic map reconstruction apparatus 700 may further include:
an auxiliary dynamic image snapshot acquisition unit configured to acquire an auxiliary dynamic image having a different modality from the target dynamic image, and acquire snapshots of respective times constituting the auxiliary dynamic image;
An auxiliary implicit characterization generating unit configured to generate an auxiliary suction flow implicit characterization when the auxiliary dynamic graph is a suction object and an auxiliary emission flow implicit characterization when the auxiliary dynamic graph is an emission object, according to each time snapshot constituting the auxiliary dynamic graph;
the auxiliary nonlinear feature fusion unit is configured to assist in attracting the flow implicit characterization and assist in sending out the nonlinear features of the flow implicit characterization to obtain auxiliary fusion features;
correspondingly, the multi-layer perceptron processing unit is further configured to:
processing the splicing characteristics of each node by using a multi-layer perceptron, and starting a point and end point flow estimation matrix; the splicing features are obtained by splicing fusion features and auxiliary fusion features.
In some optional implementations of the present embodiment, the auxiliary implicit token generation unit may be further configured to:
and generating an auxiliary suction flow implicit characterization of each node when the auxiliary dynamic graph is used as a suction object and an auxiliary emission flow implicit characterization of each node when the auxiliary dynamic graph is used as an emission object by a Bayesian chart feature learning method.
In some optional implementations of the present embodiment, the traffic flow dynamic map reconstruction apparatus 700 may further include:
A prior distribution alignment unit configured to align prior distributions of the first implicit characterization and the second implicit characterization; the first implicit representation is an implicit representation corresponding to the target dynamic graph, and the second implicit representation is an implicit representation corresponding to the auxiliary dynamic graph.
In some optional implementations of the present embodiment, the traffic flow dynamic map reconstruction apparatus 700 may further include:
and a gradient information discarding unit configured to discard the gradient information of the second implicit characterization during the back propagation.
In some optional implementations of the present embodiment, the traffic flow dynamic map reconstruction apparatus 700 may further include:
the post-completion dynamic diagram generating unit is configured to generate a post-completion dynamic diagram according to the partial node information missing by the target snapshot obtained through reconstruction;
and the traffic flow analysis unit is configured to generate a traffic flow analysis result based on the traffic flow information of each node recorded in the completed dynamic graph.
The embodiment exists as an embodiment of a device corresponding to the embodiment of the method, and the device for reconstructing the traffic flow dynamic graph provided by the embodiment determines the implicit representation of the suction flow and the implicit representation of the emission flow of the emission object of each node respectively as the suction object according to the snapshots at each moment forming the traffic flow dynamic graph, so that the flow characteristics of each node in a traffic scene can be fully matched, nonlinear factors in actual conditions are fully considered by the fusion characteristics through nonlinear processing and fusion, and the fusion characteristics of each node are processed through a multi-layer perceptron.
According to an embodiment of the present disclosure, the present disclosure further provides an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; the memory stores instructions executable by the at least one processor to enable the at least one processor to implement the traffic flow dynamics map reconstruction method described in any of the embodiments above when executed.
According to an embodiment of the disclosure, the disclosure further provides a readable storage medium storing computer instructions for enabling a computer to implement the traffic flow dynamics map reconstruction method described in any of the above embodiments when executed.
According to an embodiment of the present disclosure, the present disclosure further provides a computer program product, which, when executed by a processor, is capable of implementing the traffic flow dynamics map reconstruction method described in any of the above embodiments.
Fig. 8 illustrates a schematic block diagram of an example electronic device 800 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 8, the apparatus 800 includes a computing unit 801 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 802 or a computer program loaded from a storage unit 808 into a Random Access Memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other by a bus 804. An input/output (I/O) interface 805 is also connected to the bus 804.
Various components in device 800 are connected to I/O interface 805, including: an input unit 806 such as a keyboard, mouse, etc.; an output unit 807 such as various types of displays, speakers, and the like; a storage unit 808, such as a magnetic disk, optical disk, etc.; and a communication unit 809, such as a network card, modem, wireless communication transceiver, or the like. The communication unit 809 allows the device 800 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.
The computing unit 801 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 801 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the respective methods and processes described above, such as a traffic flow map reconstruction method. For example, in some embodiments, the traffic flow dynamics reconstruction method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 808. In some embodiments, part or all of the computer program may be loaded and/or installed onto device 800 via ROM 802 and/or communication unit 809. When the computer program is loaded into RAM 803 and executed by computing unit 801, one or more steps of the traffic flow dynamics map reconstruction method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the traffic flow dynamics map reconstruction method by any other suitable means (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so as to solve the defects of large management difficulty and weak service expansibility in the traditional physical host and virtual private server (VPS, virtual Private Server) service.
According to the technical scheme of the embodiment of the disclosure, by determining the implicit representation of the attraction flow of each node as the attraction object and the implicit representation of the emission flow of the emission object according to the snapshots of each moment forming the traffic flow dynamic graph, the flow characteristics of each node in a traffic scene can be fully matched, the nonlinear factors in actual conditions are fully considered by the fusion characteristics through nonlinear processing and fusion, and the fusion characteristics of each node are processed through the multi-layer perceptron.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.
The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.
Claims (14)
1. A traffic flow dynamic graph reconstruction method, comprising:
obtaining snapshots of all moments forming a target dynamic graph; the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises a target snapshot with partial node information missing;
generating an implicit representation of the suction flow when each node is used as a suction object and an implicit representation of the emission flow when each node is used as an emission object according to each time snapshot;
fusing the implicit characterization of the suction flow and the nonlinear characteristic of the implicit characterization of the emission flow to obtain a fused characteristic;
processing the fusion characteristics of each node by using a multi-layer perceptron to obtain a starting point and end point flow estimation matrix;
constructing a travel event by taking any node in the starting point and end point flow estimation matrix as a starting point and any node as an end point;
taking poisson distribution as a desired solving mode of an evidence lower bound mode of a calculation variation inference principle, and fitting target traffic flow information between a starting point and an ending point in the travel event;
reconstructing the partial node information missing by the target snapshot according to the target traffic flow information related to the partial node missing by the target snapshot.
2. The method of claim 1, further comprising:
acquiring an auxiliary dynamic graph with different modes from the target dynamic graph, and acquiring snapshots of all moments forming the auxiliary dynamic graph;
generating an auxiliary suction flow implicit characterization of each node when the node is used as a suction object and an auxiliary emission flow implicit characterization when the node is used as an emission object according to the snapshots of each moment forming the auxiliary dynamic graph;
fusing the auxiliary suction flow implicit characterization and the nonlinear characteristic of the auxiliary emission flow implicit characterization to obtain an auxiliary fusion characteristic;
correspondingly, the processing the fusion characteristic of each node by using the multi-layer perceptron to obtain a starting point and end point flow estimation matrix comprises the following steps:
processing the splicing characteristics of each node by using a multi-layer perceptron to obtain the starting point and ending point flow estimation matrix; the splicing features are spliced by the fusion features and the auxiliary fusion features.
3. The method of claim 2, wherein generating an auxiliary attractive flow implicit characterization for each node as an attractive object and an auxiliary outgoing flow implicit characterization for each node as an outgoing object from the time snapshots comprising the auxiliary dynamic graph comprises:
And generating auxiliary suction flow implicit characterization of each node when the auxiliary dynamic graph is used as a suction object and auxiliary emission flow implicit characterization of each node when the auxiliary dynamic graph is used as an emission object by a Bayesian chart feature learning method.
4. The method of claim 2, further comprising:
aligning a priori distributions of the first implicit characterization and the second implicit characterization; wherein the first implicit representation is an implicit representation corresponding to the target dynamic graph and the second implicit representation is an implicit representation corresponding to the auxiliary dynamic graph.
5. The method of claim 4, further comprising:
the gradient information of the second implicit characterization is discarded during back propagation.
6. The method of any of claims 1-5, further comprising:
generating a dynamic graph after completion according to the partial node information missing from the target snapshot obtained by reconstruction;
and generating a traffic flow analysis result based on the traffic flow information of each node recorded in the dynamic graph after completion.
7. A traffic flow dynamics map reconstruction apparatus comprising:
a dynamic graph snapshot acquisition unit configured to acquire snapshots of respective times constituting a target dynamic graph; the target dynamic graph is used for recording traffic flow information, and each time snapshot comprises a target snapshot with partial node information missing;
An implicit representation generating unit configured to generate an implicit representation of the suction flow rate when each node is the suction object and an implicit representation of the emission flow rate when each node is the emission object according to each time snapshot;
the nonlinear feature fusion unit is configured to fuse the nonlinear features of the implicit characterization of the suction flow and the implicit characterization of the emission flow to obtain fusion features;
the multi-layer perceptron processing unit is configured to process the fusion characteristics of each node by utilizing the multi-layer perceptron to obtain a starting point and end point flow estimation matrix;
the missing information reconstruction unit is configured to construct a travel event by taking any node in the starting point and end point flow estimation matrix as a starting point and any node as an end point; taking poisson distribution as a desired solving mode of an evidence lower bound mode of a calculation variation inference principle, and fitting target traffic flow information between a starting point and an ending point in the travel event; reconstructing the partial node information missing by the target snapshot according to the target traffic flow information related to the partial node missing by the target snapshot.
8. The apparatus of claim 7, further comprising:
an auxiliary dynamic graph snapshot acquisition unit configured to acquire an auxiliary dynamic graph having a different modality from the target dynamic graph, and acquire snapshots of respective times constituting the auxiliary dynamic graph;
An auxiliary implicit characterization generating unit configured to generate an auxiliary suction flow implicit characterization when each node is a suction object and an auxiliary emission flow implicit characterization when each node is an emission object according to each time snapshot constituting the auxiliary dynamic graph;
the auxiliary nonlinear feature fusion unit is configured to obtain auxiliary fusion features according to the auxiliary suction flow implicit characterization and the nonlinear features of the auxiliary emission flow implicit characterization;
correspondingly, the multi-layer perceptron processing unit is further configured to:
processing the splicing characteristics of each node by using a multi-layer perceptron to obtain the starting point and ending point flow estimation matrix; the splicing features are spliced by the fusion features and the auxiliary fusion features.
9. The apparatus of claim 8, wherein the auxiliary implicit characterization generation unit is further configured to:
and generating auxiliary suction flow implicit characterization of each node when the auxiliary dynamic graph is used as a suction object and auxiliary emission flow implicit characterization of each node when the auxiliary dynamic graph is used as an emission object by a Bayesian chart feature learning method.
10. The apparatus of claim 8, further comprising:
A prior distribution alignment unit configured to align prior distributions of the first implicit characterization and the second implicit characterization; wherein the first implicit representation is an implicit representation corresponding to the target dynamic graph and the second implicit representation is an implicit representation corresponding to the auxiliary dynamic graph.
11. The apparatus of claim 10, further comprising:
and a gradient information discarding unit configured to discard the gradient information of the second implicit characterization during back propagation.
12. The apparatus of any of claims 7-11, further comprising:
the post-completion dynamic diagram generating unit is configured to generate a post-completion dynamic diagram according to the partial node information missing by the target snapshot obtained through reconstruction;
and the traffic flow analysis unit is configured to generate a traffic flow analysis result based on the traffic flow information of each node recorded in the completed dynamic diagram.
13. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein, the liquid crystal display device comprises a liquid crystal display device,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the traffic flow dynamics map reconstruction method according to any one of claims 1-6.
14. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the traffic flow dynamics map reconstruction method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210837366.9A CN115186047B (en) | 2022-07-15 | 2022-07-15 | Traffic flow dynamic diagram reconstruction method, related device and computer program product |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210837366.9A CN115186047B (en) | 2022-07-15 | 2022-07-15 | Traffic flow dynamic diagram reconstruction method, related device and computer program product |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115186047A CN115186047A (en) | 2022-10-14 |
| CN115186047B true CN115186047B (en) | 2023-07-18 |
Family
ID=83519878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210837366.9A Active CN115186047B (en) | 2022-07-15 | 2022-07-15 | Traffic flow dynamic diagram reconstruction method, related device and computer program product |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115186047B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111063194A (en) * | 2020-01-13 | 2020-04-24 | 兰州理工大学 | Traffic flow prediction method |
| CN114445252A (en) * | 2021-11-15 | 2022-05-06 | 南方科技大学 | Data completion method and device, electronic equipment and storage medium |
| CN114611798A (en) * | 2022-03-06 | 2022-06-10 | 北京工业大学 | OD passenger flow prediction method based on dynamic hypergraph convolutional neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TR201514432T1 (en) * | 2013-06-21 | 2016-11-21 | Aselsan Elektronik Sanayi Ve Ticaret Anonim Sirketi | Method for pseudo-recurrent processing of data using a feedforward neural network architecture |
-
2022
- 2022-07-15 CN CN202210837366.9A patent/CN115186047B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111063194A (en) * | 2020-01-13 | 2020-04-24 | 兰州理工大学 | Traffic flow prediction method |
| CN114445252A (en) * | 2021-11-15 | 2022-05-06 | 南方科技大学 | Data completion method and device, electronic equipment and storage medium |
| CN114611798A (en) * | 2022-03-06 | 2022-06-10 | 北京工业大学 | OD passenger flow prediction method based on dynamic hypergraph convolutional neural network |
Non-Patent Citations (1)
| Title |
|---|
| 基于Optima的实时在线交通流预测方法研究;李颖宏等;交通运输系统工程与信息;17(02);119-125 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115186047A (en) | 2022-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112862877B (en) | Method and apparatus for training an image processing network and image processing | |
| US20140240310A1 (en) | Efficient approach to estimate disparity map | |
| CN113901909B (en) | Video-based target detection method and device, electronic equipment and storage medium | |
| CN113538235B (en) | Training method and device for image processing model, electronic equipment and storage medium | |
| CN114092673B (en) | Image processing method and device, electronic equipment and storage medium | |
| CN114627239B (en) | Bounding box generation method, device, equipment and storage medium | |
| CN110633716A (en) | Target object detection method and device | |
| CN114792355A (en) | Virtual image generation method and device, electronic equipment and storage medium | |
| CN115170815A (en) | Method, device and medium for processing visual task and training model | |
| CN114677350A (en) | Connection point extraction method and device, computer equipment and storage medium | |
| CN113870439A (en) | Method, apparatus, device and storage medium for processing image | |
| CN115186047B (en) | Traffic flow dynamic diagram reconstruction method, related device and computer program product | |
| CN113052962A (en) | Model training method, information output method, device, equipment and storage medium | |
| EP4123605A2 (en) | Method of transferring image, and method and apparatus of training image transfer model | |
| CN113781653B (en) | Object model generation method and device, electronic equipment and storage medium | |
| CN113514053B (en) | Method and device for generating sample image pair and method for updating high-precision map | |
| CN115866229A (en) | Method, apparatus, device and medium for converting view angle of multi-view image | |
| CN115688917A (en) | Neural network model training method and device, electronic equipment and storage medium | |
| CN113591580B (en) | Image annotation method and device, electronic equipment and storage medium | |
| CN113505834A (en) | Method for training detection model, determining image updating information and updating high-precision map | |
| CN114187318A (en) | Image segmentation method and device, electronic equipment and storage medium | |
| CN114266937A (en) | Model training method, image processing method, device, equipment and storage medium | |
| CN115187698A (en) | Dynamic graph reconstruction method, device, equipment, readable storage medium and program product | |
| CN115049895B (en) | Image attribute identification method, attribute identification model training method and device | |
| CN115661449B (en) | Image segmentation and training method and device for image segmentation model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |