CN115345354A - Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network - Google Patents
Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network Download PDFInfo
- Publication number
- CN115345354A CN115345354A CN202210916015.7A CN202210916015A CN115345354A CN 115345354 A CN115345354 A CN 115345354A CN 202210916015 A CN202210916015 A CN 202210916015A CN 115345354 A CN115345354 A CN 115345354A
- Authority
- CN
- China
- Prior art keywords
- traffic
- network
- mode
- hypergraph
- subway
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000008878 coupling Effects 0.000 claims abstract description 13
- 238000010168 coupling process Methods 0.000 claims abstract description 13
- 238000005859 coupling reaction Methods 0.000 claims abstract description 13
- 238000013136 deep learning model Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 57
- 239000000126 substance Substances 0.000 claims description 28
- 230000006870 function Effects 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 3
- 238000000547 structure data Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 abstract description 2
- 238000012549 training Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 230000029305 taxis Effects 0.000 description 3
- 101150104646 SET4 gene Proteins 0.000 description 1
- 101100421299 Schizosaccharomyces pombe (strain 972 / ATCC 24843) set7 gene Proteins 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Educational Administration (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method for predicting urban multi-mode traffic super-network situation based on a super-map deep network, which comprises the steps of constructing a traffic map network of each traffic mode according to the spatial topological relation of subway stations, bus stations and road sections in urban traffic; analyzing the high-order semantic relevance of each traffic mode, and constructing traffic hypergraph networks of different traffic modes; analyzing the coupling relation among different traffic modes, and constructing a traffic hypergraph network fusing urban multi-mode traffic; establishing a hypergraph deep learning model facing urban multi-mode traffic hypergraph network collaborative prediction through a hypergraph neural network, and training the model by using historical data; and finally, predicting the multi-mode traffic passenger flow situation in the future time period by using the trained hypergraph deep learning model. The method provided by the invention can reflect the operation rule of the urban multi-mode traffic system more objectively and comprehensively, and improve the prediction accuracy of the multi-mode traffic passenger flow situation.
Description
Technical Field
The invention belongs to the field of traffic passenger flow prediction, and particularly relates to a urban multi-mode traffic super-network situation prediction method based on a super-graph deep network.
Background
With the construction and continuous promotion of urban comprehensive transportation systems, the urban transportation passenger flow prediction has important significance for the operation and management of the urban transportation systems, and has attracted wide attention in recent years. Graph Neural Networks (GNNs) show remarkable advantages in traffic prediction problems due to the strong relational data modeling capability, and are widely applied. However, the traditional graph network model models the correlation according to the spatial topological relation, and only a binary connection relation between network nodes can be constructed, that is, one edge connects two nodes, and high-order correlation between nodes cannot be constructed, for example, a certain bus route. Meanwhile, in the urban integrated traffic system, a complex large system is adopted, and obvious coupling exists among multiple traffic modes, such as a bus station and a subway station on the same road or the same name of the bus station and the subway station (usually located in a similar place), while the traditional graph network model cannot model the coupling relation among the multiple modes of traffic. In order to accurately analyze the multi-mode traffic situation of the urban integrated traffic system, a more comprehensive and efficient modeling method is urgently needed to accurately predict the passenger flow states and development trends of different traffic modes.
Disclosure of Invention
In view of the above, the present invention provides a method for predicting a situation of an urban multi-mode traffic hyper-network based on a hyper-graph deep network, so as to solve the problem that high-order correlation of nodes and multi-mode traffic coupling cannot be modeled in multi-mode traffic prediction.
The technical scheme for realizing the purpose of the invention is as follows: a city multi-mode traffic hyper-network situation prediction method based on a hyper-map deep network comprises the following specific steps:
step 1: analyzing the traffic network to obtain multi-mode traffic network data;
step 2: constructing a traffic map network of each traffic mode by using the spatial topological relation;
and step 3: analyzing the high-order semantic correlation of each traffic mode, and constructing traffic hypergraph networks of different traffic modes;
and 4, step 4: analyzing the coupling relation among different traffic modes, and constructing a traffic hypergraph network fusing urban multi-mode traffic;
and 5: establishing a hypergraph deep network facing to urban multi-mode traffic hypergraph collaborative prediction;
and 6: and predicting the multi-mode traffic passenger flow situation in the future time period by using a hypergraph deep network oriented to urban multi-mode traffic hypergraph collaborative prediction.
Preferably, the traffic network is analyzed in step 1, and the specific method for obtaining the multimode traffic network data comprises the following steps:
step 1.1: analyzing the multi-mode traffic network of the subway, the public transport and the taxi to obtain node data of each traffic mode map network;
step 1.2: analyzing the topological structure of the multi-mode traffic network of the subway, the public bus and the taxi to obtain the spatial topological structure data of each traffic mode.
Preferably, for the subway traffic mode, the subway stations are extracted as the nodes of the subway traffic mode map network to form a subway traffic map network node set
I (m) The number of subway stations; aiming at the public transportation mode, bus stops are extracted to be used as nodes of a public transportation mode graph network to form a public transportation mode graph network node set
I (b) The number of bus stops; for taxi traffic modes, extracting sections where taxies can pass through as nodes of a taxi traffic mode graph network to form a taxi traffic graph network node set
I (r) Is the number of road segments.
Preferably, the specific method for constructing the traffic map network of each transportation mode by using the spatial topological relation comprises the following steps:
aiming at the subway traffic mode, a network node V of a subway traffic map is utilized (m) Adjacency matrix A with subway traffic mode (m) Traffic map network G for constructing subway traffic mode (m) ＝(V (m) ,A (m) )；
Aiming at the public transport mode, the network node V of the public transport map is utilized (b) Adjacent matrix A with public transport mode (b) Traffic map network G for constructing public transport mode (b) ＝(V (b) ,A (b) )；
Aiming at the taxi traffic mode, a network node V of a taxi traffic map is utilized (r) Adjacency matrix A with taxi traffic mode (r) Traffic map network G for constructing public transport mode (r) ＝(V (r) ,A (r) )。
Preferably, the specific steps of analyzing the high-order semantic relevance of each traffic mode and constructing the traffic hypergraph network of different traffic modes are as follows:
step 3.1: aiming at the subway traffic mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (m) Building subway traffic hypergraph network
The method comprises the following specific steps:
step 3.1.1: aiming at the subway traffic mode, the semantic relation that the subway stations belong to the same subway line is utilized to establish the overedge
J (m) Is a super edge
The number of (2);
step 3.1.2: by usingSet of hyper-edges epsilon (m) Construction of subway traffic hypergraph network
Wherein the content of the first and second substances,
is a node set of the subway traffic hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in the subway traffic hypergraph network (m) And J (m) Respectively the number of nodes and the number of excess edges, h (m) (i, j) is
Row i and column j elements of (1);
step 3.2: aiming at the public transportation mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (b) Building public transport traffic hypergraph network
The method specifically comprises the following steps:
step 3.2.1: aiming at the public transportation mode, the semantic relation that the public transportation stations belong to the same public transportation line is utilized to establish the transfinite
J (b) Is a super edge
The number of (c);
step 3.2.2: using super-edge sets epsilon (b) Constructing public transport traffic hypergraph network
Wherein, the first and the second end of the pipe are connected with each other,
is a node set of the public transportation hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in the public transportation super map network (b) And J (b) Respectively the number of nodes and the number of excess edges, h (b) (i, j) is
Row i and column j.
Preferably, the specific steps of analyzing the coupling relationship between different transportation modes and constructing the traffic hypergraph network fusing urban multi-mode transportation are as follows:
step 4.1: analyzing semantic relations among multi-mode traffic, establishing a super edge between every two sub-networks, and constructing a traffic hypergraph network between every two sub-networks;
step 4.2: analyzing the coupling relation among three sub-networks of subway, bus and taxi, and utilizing semantic relationStation names are the same, and the superedges are built on the same road section, and a traffic supergraph network integrating three traffic modes of subway, bus and taxi is built according to the superedges
Wherein
In the form of a set of nodes, the nodes,
is a node set of the subway traffic hypergraph network,
is a node set of the public transportation hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a subway-bus-taxi traffic hypergraph network (m,b,r) ＝I (m) +I (b) +I (r) And J (m,b,r) Respectively the number of nodes and the number of excess edges, h (m ,b,r) (i, j) is
Row i and column j.
Preferably, the specific method for analyzing the semantic relationship between the multimode traffic, establishing the super edges between every two sub-networks and constructing the traffic hypergraph network between every two sub-networks comprises the following steps:
step 4.1.1: aiming at a subway-bus network, a super edge is established by utilizing the semantic relation that the station names of a subway station and a bus station are the same
Wherein
And constructing subway-bus traffic hypergraph network by using the same
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in the subway-bus transit hypergraph network (m,b) ＝I (m) +I (b) And J (m,b) Respectively the number of nodes and the number of excess edges, h (m,b) (i, j) is
Row i and column j elements of (1);
step 4.1.2: aiming at a subway-taxi network, a semantic relation of 'subway station on a certain road section' is utilized to establish a super edge
Wherein
And constructing subway-taxi traffic hypergraph network by using the same
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a subway-taxi traffic hypergraph network (m,r) ＝I (m) +I (r) And J (m,r) Respectively the number of nodes and the number of excess edges, h (m,r) (i, j) is
Row i and column j elements of (1);
step 4.1.3: aiming at a public transport-taxi network, a semantic relation of 'a bus station on a certain road segment' is utilized to establish a super edge
Wherein
And construct public transportation-leasing by thisVehicle traffic hypergraph network
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a public transport-taxi traffic hypergraph network (b,r) ＝I (b) +I (r) And J (b,r) Respectively the number of nodes and the number of excess edges, h (b,r) (i, j) is
Row i and column j.
Preferably, the specific method for establishing the supergraph deep network facing the urban multi-mode traffic supernetwork collaborative prediction comprises the following steps:
step 5.1: construction of traffic hypergraph network fusing subway-bus-taxi multi-mode traffic
The method specifically comprises the following steps:
step 5.1.1: the graph network G constructed in the step 2 to the step 4 (m) 、G (b) 、G (r) And hypergraph network
The fusion is carried out, the super edge of the fusion subway-bus-taxi multi-mode traffic hypergraph network is established, as shown in the following,
ε＝{e j |j＝1,2,…,J}＝ε (1) ∪ε (2) ∪ε (3) ∪ε (m) ∪ε (b) ∪ε (m,b) ∪ε (m,r) ∪ε (b,r) ∪ε (m,b,r)
wherein, the first and the second end of the pipe are connected with each other,
wherein A is (m) (i,j)、A (b) (i, j) and A (r) (i, j) are each matrix A (m) 、A (b) And A (r) Row i and column j elements of (1);
step 5.1.2: constructing a traffic hypergraph network fusing subway-bus-taxi multi-mode traffic by utilizing a hyperedge set epsilon
Wherein
Is a collection of nodes that are to be grouped together,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and e j Epsilon is respectively the ith node and the jth super edge in the multi-mode traffic hypergraph network, and I = I (m) +I (b) +I (r) The number of nodes of the hypergraph network; j = I (m) +I (b) +I (r) +J (m) +J (b) +J (m,b) +J (m,r) +J (b,r) +J (m,b,r) The number of the super edges of the hypergraph network is;
step 5.2: constructing a feature matrix X (t) and a prediction label y (t) by using historical data;
step 5.3: establishing a hypergraph deep learning model f oriented to multi-mode traffic cooperative prediction, wherein an optimization objective function of the hypergraph deep learning model is as follows,
wherein the content of the first and second substances,
is an empirical loss function, Ω reg (. H) is a regularization function;
step 5.4: inputting sample data into a hypergraph depth network model to train the sample data to obtain a multi-mode traffic collaborative prediction model
Preferably, the method for constructing the feature matrix X (t) and the prediction label y (t) by using the historical data specifically comprises the following steps:
step 5.2.1: for the time period T, the hypergraph network in the current time period T and T-1 historical time periods is utilized
Traffic passenger flow data on each nodeHypergraph network feature matrix as current time period
Wherein
The traffic passenger flow of each node of the hypergraph network in the time period t forms a characteristic vector;
step 5.2.2: for the time period t, the traffic passenger flow on each node of the hypergraph network in the next time period t +1 is used for forming the prediction label of the current time period
Preferably, the specific method for predicting the multi-mode traffic passenger flow situation in the future time period by using the hypergraph deep network oriented to the urban multi-mode traffic hyper-network collaborative prediction comprises the following steps:
step 6.1: for any time period T ', the hypergraph network in the current time period T' and T-1 historical time periods is utilized
Traffic passenger flow data on each node is used as a hypergraph network characteristic matrix of the current time period
Wherein
A characteristic vector is formed by the traffic passenger flow of each node of the hypergraph network in the time period t';
step 6.2: the feature matrix X (t') and the hypergraph correlation matrix constructed in the step 5.1 are combined
Inputting the multi-mode traffic cooperative prediction model obtained in the step 5.4
For future timeThe multi-mode traffic flow within the segment is predicted and, as shown below,
wherein the content of the first and second substances,
the predicted value of the traffic passenger flow of each node of the hypergraph network in the next time period t' +1 is obtained.
Compared with the prior art, the invention has the following remarkable advantages: according to the method, the high-order correlation between the traffic network nodes and the coupling relation between the multi-mode traffic are simultaneously modeled in a unified mode, and the multi-mode traffic passenger flow situation is cooperatively predicted, so that the operation rule of the urban multi-mode traffic system can be reflected more objectively and comprehensively, and the accuracy of the multi-mode traffic passenger flow situation prediction is improved.
The present invention is described in further detail below with reference to the attached drawings.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a schematic diagram of a subway and bus hypergraph network in the invention.
FIG. 3 is a schematic diagram of a hypergraph network integrating multi-mode traffic of subways, buses and taxis.
FIG. 4 is a schematic diagram of a fusion method of an adjacency matrix and an association matrix in the present invention.
Detailed Description
Referring to fig. 1, the invention provides a method for predicting urban multi-mode traffic hyper-network situation based on a hyper-graph deep network, which comprises the following specific steps:
step 1: analyzing the traffic network in the research area to obtain multi-mode traffic network data;
step 2: constructing a traffic map network of each traffic mode by using the spatial topological relation;
and 3, step 3: analyzing the high-order semantic relevance of each traffic mode, and constructing traffic hypergraph networks of different traffic modes;
and 4, step 4: analyzing the coupling relation among different traffic modes, and constructing a traffic hypergraph network fusing urban multi-mode traffic;
and 5: establishing a hypergraph deep network facing to urban multi-mode traffic hypergraph collaborative prediction;
step 6: and predicting the multi-mode traffic passenger flow situation in the future time period.
In this embodiment, the step 1 specifically includes:
step 1.1: analyzing the multi-mode traffic networks of subways, buses and taxis in the area near the new street intersection in Nanjing city to obtain node data V of each traffic mode map network (m) 、V (b) And V (r) ；
Step 1.2: analyzing the multi-mode traffic network topological structure of subways, buses and taxis to obtain space topological structure data A of all traffic modes (m) 、A (b) And A (r) 。
In this embodiment, the step 1.1 specifically includes:
step 1.1.1: aiming at the subway traffic mode, subway stations are extracted to be used as nodes of the traffic mode graph network to form a subway traffic graph network node set
4 is the number of subway stations;
step 1.1.2: aiming at the public transportation mode, bus stops are extracted as nodes of the transportation mode graph network to form a public transportation mode graph network node set
7 is the number of bus stops;
step 1.1.3: aiming at the taxi traffic mode, the passable road sections are extracted to be used as nodes of the traffic mode graph network to form a taxi traffic graph network node set
16 is the number of taxi-driven road segments in the area under study.
In this embodiment, the step 1.2 specifically includes:
step 1.2.1: aiming at subway traffic modes, an adjacency matrix of the traffic mode map network is constructed by utilizing the spatial topological structure relationship between subway stations
Step 1.2.2: aiming at the public transportation mode, the spatial topological structure relationship between the bus stops is utilized to construct the adjacency matrix of the transportation mode graph network
Step 1.2.3: aiming at the taxi traffic mode, constructing an adjacency matrix of the traffic mode graph network by utilizing the space topological structure relationship among road sections
In this embodiment, the step 2 specifically includes:
step 2.1: aiming at the subway traffic mode, a network node V is utilized (m) And the adjacent matrix A (m) Constructing a traffic map network G of the traffic mode (m) ＝(V (m) ,A (m) )；
Step 2.2: aiming at the public transportation mode, a network node V is utilized (b) And the adjacent matrix A (b) Constructing a traffic map network G of the traffic mode (b) ＝(V (b) ,A (b) )；
Step 2.3: aiming at taxi traffic mode, network node V is utilized (r) And the adjacent matrix A (r) Constructing a traffic map network G of the traffic mode (r) ＝(V (r) ,A (r) )。
In this embodiment, the step 3 specifically includes:
step 3.1: aiming at the subway traffic mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (m) Building subway traffic hypergraph network
Step 3.2: aiming at the public transportation mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (b) Building public transport traffic hypergraph network
In this embodiment, the step 3.1 specifically includes:
step 3.1.1: aiming at the subway traffic mode, the method establishes the overlimit by utilizing the semantic relation that the subway stations belong to the same subway line
Step 3.1.2: using super-edge sets epsilon (m) Construction of subway traffic hypergraph network
Wherein, the first and the second end of the pipe are connected with each other,
is a node set of the subway traffic hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and with
The ith node and the jth super edge in the subway traffic super map network are respectively.
In this embodiment, the step 3.2 specifically includes:
step 3.2.1: aiming at the public transportation mode, the semantic relation that the public transportation stations belong to the same public transportation line is utilized to establish the transfinite
Step 3.2.2: using super-edge sets epsilon (b) Constructing public transport traffic hypergraph network
Wherein the content of the first and second substances,
is a node set of the public transportation hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and with
The ith node and the jth super edge in the bus transit super map network are respectively.
In this embodiment, the step 4 specifically includes:
step 4.1: analyzing semantic relation between multi-mode traffic and establishing hyper-edge epsilon between every two sub-networks (m,b) 、ε (m,r) And epsilon (b,r) Constructing a traffic hypergraph network between two subnetworks
And with
And 4.2: analyzing the coupling relation among three sub-networks of subway, bus and taxi, and establishing a super edge by utilizing the semantic relation that the station names of the subway station and the bus station are the same and are on the same road section
Wherein
And constructing a traffic hypergraph network fusing three traffic modes of subway, bus and taxi
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
the ith node and the jth super edge in the subway-bus-taxi traffic super map network are respectively.
In this embodiment, the step 4.1 specifically includes:
step 4.1.1: aiming at a subway-bus network, a super edge is established by utilizing the semantic relation that the station names of a subway station and a bus station are the same
Wherein
And constructing subway-bus traffic hypergraph network by using the same
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge in the subway-bus transit hypergraph network;
step 4.1.2: aiming at a subway-taxi network, a semantic relation of 'subway station on a certain road section' is utilized to establish a super edge
Wherein
And constructing subway-taxi traffic hypergraph network by using the same
Wherein
Is a section ofThe set of points is then set to a point,
is a correlation matrix, which is defined as follows,
wherein, the first and the second end of the pipe are connected with each other,
and with
Respectively an ith node and a jth super edge in the subway-taxi traffic hypergraph network;
step 4.1.3: aiming at a public transport-taxi network, a semantic relation of 'a bus station on a certain road segment' is utilized to establish a super edge
Wherein
And a public transport-taxi traffic hypergraph network is constructed by the method
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein, the first and the second end of the pipe are connected with each other,
and
respectively an ith node and a jth super edge in the public transport-taxi traffic super map network;
in this embodiment, the step 5 specifically includes:
step 5.1: construction of traffic hypergraph network fusing subway-bus-taxi multi-mode traffic
Step 5.2: constructing a feature matrix X (t) and a prediction label y (t) by using historical data;
step 5.3: establishing a hypergraph deep learning model f for multi-mode traffic cooperative prediction;
step 5.4: inputting sample data into a hypergraph depth network model to train the sample data to obtain a multi-mode traffic collaborative prediction model
In this embodiment, the step 5.1 specifically includes:
step 5.1.1: the graph network G constructed in the step 2 to the step 4 (m) 、G (b) 、G (r) And hypergraph network
The fusion is carried out, the super edge of the fusion subway-bus-taxi multi-mode traffic hypergraph network is established, as shown in the following,
ε＝{e j |j＝1,2,…,J}＝ε (1) ∪ε (2) ∪ε (3) ∪ε (m) ∪ε (b) ∪ε (m,b) ∪ε (m,r) ∪ε (b,r) ∪ε (m,b,r)
wherein the content of the first and second substances,
wherein A is (m) (i,j)、A (b) (i, j) and A (r) (i, j) are each matrix A (m) 、A (b) And A (r) Row i and column j.
Step 5.1.2: constructing a traffic hypergraph network fusing subway-bus-taxi multi-mode traffic by utilizing a hyperedge set epsilon
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and e j Epsilon is respectively the ith node and the jth super edge in the multi-mode traffic hypergraph network, and I = I (m) +I (b) +I (r) =27 and J = I (m) +I (b) +I (r) +J (m) +J (b) +J (m,b) +J (m,r) +J (b,r) +J (m,b,r) =49 sections of hypergraph networks respectivelyThe number of points and the number of super edges.
In this embodiment, the step 5.2 specifically includes:
step 5.2.1: for the time period t, the hypergraph network in the current time period t and 5 historical time periods is utilized
Traffic passenger flow data on each node is used as a hypergraph network characteristic matrix of the current time period
Wherein t is taken to be 15 minutes,
the traffic passenger flow of each node of the hypergraph network in the current 15 minutes forms a feature vector;
step 5.2.2: for the time period t, the traffic passenger flow on each node of the hypergraph network in the next time period t +1 (namely the next 15 minutes) is used for forming the prediction label of the current time period
In this embodiment, the step 5.3 is specifically to establish a hypergraph deep learning model f for multimodal transportation collaborative prediction by using a hypergraph convolutional neural network model, and an optimization objective function of the hypergraph deep learning model f is as follows,
wherein the content of the first and second substances,
is an empirical loss function, Ω reg (. Cndot.) is a regularization function.
In this embodiment, the step 5.4 specifically includes the feature matrix X (t) and the prediction label y (t) constructed in the step 5.2 and the hypergraph correlation matrix constructed in the step 5.1
Inputting the hypergraph deep learning model f established in the step 5.3, and training the model by using a gradient descent method to obtain an optimized multi-mode traffic cooperative prediction model
In this embodiment, the step 6 specifically includes:
step 6.1: for any time period t ', the hypergraph network in the current time period t' and 5 historical time periods is utilized
Traffic passenger flow data on each node is used as a hypergraph network characteristic matrix of the current time period
Wherein
A characteristic vector formed by traffic passenger flow of each node of the hypergraph network in the time period t';
step 6.2: the feature matrix X (t') and the hypergraph incidence matrix constructed in the step 5.1 are combined
Inputting the multi-mode traffic cooperative prediction model obtained in the step 5.4
The prediction of multi-mode traffic flow over a future time period is performed, as shown below,
wherein, the first and the second end of the pipe are connected with each other,
for the next time period t' +1 internal hypergraph networkAnd (4) a predicted value of the traffic passenger flow of each node.
The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.
Claims (10)
1. A city multi-mode traffic hyper-network situation prediction method based on a hyper-map deep network is characterized by comprising the following specific steps:
step 1: analyzing the traffic network to obtain multi-mode traffic network data;
step 2: constructing a traffic map network of each traffic mode by using the spatial topological relation;
and step 3: analyzing the high-order semantic relevance of each traffic mode, and constructing traffic hypergraph networks of different traffic modes;
and 4, step 4: analyzing the coupling relation among different traffic modes, and constructing a traffic hypergraph network fusing urban multi-mode traffic;
and 5: establishing a hypergraph deep network facing to urban multi-mode traffic hypergraph collaborative prediction;
step 6: and predicting the multi-mode traffic passenger flow situation in the future time period by using a hypergraph deep network oriented to urban multi-mode traffic hypergraph collaborative prediction.
2. The method for predicting the urban multi-mode traffic hyper-network situation based on the hyper-map deep network according to claim 1, wherein the traffic network is analyzed in step 1, and the specific method for obtaining the multi-mode traffic network data comprises the following steps:
step 1.1: analyzing the multi-mode traffic network of the subway, the bus and the taxi to obtain node data of each traffic mode map network;
step 1.2: analyzing the topological structure of the multi-mode traffic network of the subway, the bus and the taxi to obtain the spatial topological structure data of each traffic mode.
3. The hypergraph-based depth mesh of claim 2The method for predicting the situation of the urban multi-mode traffic super-network is characterized in that subway stations are extracted as nodes of a subway traffic mode map network aiming at a subway traffic mode to form a subway traffic map network node set
I (m) The number of subway stations; aiming at the public transportation mode, bus stops are extracted to be used as nodes of a public transportation mode graph network to form a public transportation mode graph network node set
I (b) The number of bus stops; for taxi traffic modes, extracting sections where taxies can pass through as nodes of a taxi traffic mode graph network to form a taxi traffic graph network node set
I (r) Is the number of road segments.
4. The method for predicting the situation of the urban multi-mode traffic hyper-network based on the hyper-map deep network according to claim 1, wherein the specific method for constructing the traffic map network of each traffic mode by using the spatial topological relation comprises the following steps:
aiming at the subway traffic mode, a network node V of a subway traffic map is utilized (m) Adjacent matrix A of subway traffic mode (m) Traffic map network G for constructing subway traffic mode (m) ＝(V (m) ,A (m) )；
Aiming at the public transport mode, the network node V of the public transport map is utilized (b) Adjacent matrix A with public transport mode (b) Traffic map network G for constructing public transport mode (b) ＝(V (b) ,A (b) )；
Aiming at the taxi traffic mode, a network node V of a taxi traffic map is utilized (r) Adjacency matrix A of taxi traffic mode (r) Traffic map network G for constructing public transport mode (r) ＝(V (r) ,A (r) )。
5. The method for predicting the situation of the urban multi-mode traffic hyper-network based on the hyper-map deep network according to claim 1, wherein the concrete steps of analyzing the high-order semantic correlation of each traffic mode and constructing the traffic hyper-map networks of different traffic modes are as follows:
step 3.1: aiming at the subway traffic mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (m) Building subway traffic hypergraph network
The method specifically comprises the following steps:
step 3.1.1: aiming at the subway traffic mode, the semantic relation that the subway stations belong to the same subway line is utilized to establish the overedge
J (m) Is a super edge
The number of (2);
step 3.1.2: using super-edge sets epsilon (m) Constructing subway traffic hypergraph network
Wherein the content of the first and second substances,
is a node set of the subway traffic hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in the subway traffic hypergraph network (m) And J (m) Respectively the number of nodes and the number of excess edges, h (m) (i, j) is
Row i and column j elements of (1);
step 3.2: aiming at the public transportation mode, the high-order semantic relation is utilized to establish the hyper-edge epsilon (b) And constructing a public transport traffic hypergraph network
The method comprises the following specific steps:
step 3.2.1: aiming at the public transportation mode, the semantic relation that the public transportation stations belong to the same public transportation line is utilized to establish the transfinite
J (b) Is a super edge
The number of (2);
step 3.2.2: using super-edge sets epsilon (b) Constructing public transport traffic hypergraph network
Wherein the content of the first and second substances,
is a node set of the public transportation hypergraph network,
is a correlation matrix, which is defined as follows,
wherein, the first and the second end of the pipe are connected with each other,
and
respectively an ith node and a jth super edge, I in the public transport hypergraph network (b) And J (b) Respectively the number of nodes and the number of excess edges, h (b) (i, j) is
Row ith and column jth elements of (1).
6. The method for predicting the situation of the urban multi-mode traffic hyper-network based on the hyper-map deep network according to claim 1, wherein the concrete steps of analyzing the coupling relationship among different traffic modes and constructing the traffic hyper-map network fusing urban multi-mode traffic are as follows:
step 4.1: analyzing semantic relations among multi-mode traffic, establishing a super edge between every two sub-networks, and constructing a traffic hypergraph network between every two sub-networks;
step 4.2: analyzing the coupling relation among three sub-networks of subway, bus and taxi, establishing a super edge by utilizing the semantic relation that the station names of the subway station and the bus station are the same and are on the same road section, and constructing a traffic super map network fusing the three traffic modes of subway, bus and taxi according to the super edge
Wherein
In the form of a set of nodes, the nodes,
is a node set of the subway traffic hypergraph network,
is a node set of the public transportation hypergraph network,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a subway-bus-taxi traffic hypergraph network (m,b,r) ＝I (m) +I (b) +I (r) And J (m,b,r) Respectively the number of nodes and the number of excess edges, h (m,b,r) (i, j) is
Row i and column j.
7. The method for predicting the situation of the urban multi-mode traffic hyper-network based on the hyper-map deep network as claimed in claim 6, wherein the concrete method for analyzing the semantic relationship between multi-mode traffic, establishing the hyper-edge between every two sub-networks and constructing the traffic hyper-map network between every two sub-networks comprises the following steps:
step 4.1.1: aiming at the subway-bus network, the construction of the subway station and the bus station with the same station name by utilizing the semantic relationVertical super edge
Wherein
And constructing subway-bus traffic hypergraph network by using the same
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein, the first and the second end of the pipe are connected with each other,
and with
Respectively an ith node and a jth super edge, I in the subway-bus traffic hypergraph network (m,b) ＝I (m) +I (b) And J (m,b) Respectively the number of nodes and the number of excess edges, h (m,b) (i, j) is
Row ith and column jth elements of (1);
step 4.1.2: aiming at a subway-taxi network, a semantic relation of 'subway station on a certain road section' is utilized to establish a super edge
Wherein
And constructing subway-taxi traffic hypergraph network by using the same
Wherein
Is a collection of nodes that are to be grouped together,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a subway-taxi traffic hypergraph network (m,r) ＝I (m) +I (r) And J (m,r) Respectively the number of nodes and the number of excess edges, h (m,r) (i, j) is
Row i and column j elements of (1);
step 4.1.3: aiming at a public transport-taxi network, a semantic relation of 'a bus station on a certain road segment' is utilized to establish a super edge
Wherein
And a public transport-taxi traffic hypergraph network is constructed by the method
Wherein
In the form of a set of nodes, the nodes,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and
respectively an ith node and a jth super edge, I in a public transport-taxi traffic hypergraph network (b,r) ＝I (b) +I (r) And J (b,r) Respectively the number of nodes and the number of excess edges, h (b,r) (i, j) is
Row i and column j.
8. The method for predicting urban multi-mode traffic hyper-network situation based on the hyper-map deep network as claimed in claim 1, wherein the concrete method for establishing the hyper-map deep network facing urban multi-mode traffic hyper-network cooperative prediction is as follows:
step 5.1: constructing traffic hypergraph network fusing subway-bus-taxi multi-mode traffic
The method specifically comprises the following steps:
step 5.1.1: the graph network G constructed in the step 2 to the step 4 (m) 、G (b) 、G (r) And hypergraph network
The fusion is carried out, the super edge of the fusion subway-bus-taxi multi-mode traffic hypergraph network is established, as shown in the following,
ε＝{e j |j＝1,2,…,J}＝ε (1) ∪ε (2) ∪ε (3) ∪ε (m) ∪ε (b) ∪ε (m,b) ∪ε (m,r) ∪ε (b,r) ∪ε (m,b,r)
wherein the content of the first and second substances,
wherein A is (m) (i,j)、A (b) (i, j) and A (r) (i, j) are each matrix A (m) 、A (b) And A (r) Row i and column j elements of (1);
step 5.1.2: constructing a traffic hypergraph network fusing subway-bus-taxi multi-mode traffic by utilizing a hyperedge set epsilon
Wherein
Is a collection of nodes that are to be grouped together,
is a correlation matrix, which is defined as follows,
wherein the content of the first and second substances,
and e j Belongs to epsilon and is respectively the ith node and the jth super edge in the multimode traffic hypergraph network, and I = I (m) +I (b) +I (r) The number of nodes of the hypergraph network; j = I (m) +I (b) +I (r) +J (m) +J (b) +J (m,b) +J (m,r) +J (b,r) +J (m,b,r) The number of the super edges of the hypergraph network is;
step 5.2: constructing a feature matrix X (t) and a prediction label y (t) by using historical data;
step 5.3: establishing a hypergraph deep learning model f oriented to multi-mode traffic cooperative prediction, wherein an optimization objective function of the hypergraph deep learning model is as shown in the specification,
wherein the content of the first and second substances,
is an empirical loss function, Ω reg (. H) is a regularization function;
step 5.4: inputting sample data into a hypergraph depth network model to train the sample data to obtain a multi-mode traffic collaborative prediction model
9. The method for predicting the urban multi-mode traffic hyper-network situation based on the hyper-map deep network as claimed in claim 8, wherein the method for constructing the feature matrix X (t) and the prediction label y (t) by using historical data specifically comprises the following steps:
step 5.2.1: for the time period T, the hypergraph network in the current time period T and T-1 historical time periods is utilized
Traffic passenger flow data on each node is used as a hypergraph network characteristic matrix of the current time period
Wherein
A characteristic vector is formed by the traffic passenger flow of each node of the hypergraph network in the time period t;
step 5.2.2: for the time period t, the traffic passenger flow on each node of the hypergraph network in the next time period t +1 is used for forming the prediction label of the current time period
10. The method for predicting the urban multi-mode traffic hyper-network situation based on the hyper-map deep network as claimed in claim 9, wherein the specific method for predicting the multi-mode traffic passenger flow situation in the future time period by using the hyper-map deep network oriented to the urban multi-mode traffic hyper-network collaborative prediction is as follows:
step 6.1: for any time period T ', the hypergraph network in the current time period T' and T-1 historical time periods is utilized
Traffic passenger flow data on each node is used as a hypergraph network characteristic matrix of the current time period
Wherein
A characteristic vector is formed by the traffic passenger flow of each node of the hypergraph network in the time period t';
step 6.2: the feature matrix X (t') and the hypergraph correlation matrix constructed in the step 5.1 are combined
Inputting the multi-mode traffic cooperative prediction model obtained in the step 5.4
The prediction of multi-mode traffic flow over a future time period is performed, as shown below,
wherein the content of the first and second substances,
the predicted value of the traffic passenger flow of each node of the hypergraph network in the next time period t' +1 is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210916015.7A CN115345354A (en) | 2022-08-01 | 2022-08-01 | Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210916015.7A CN115345354A (en) | 2022-08-01 | 2022-08-01 | Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115345354A true CN115345354A (en) | 2022-11-15 |
Family
ID=83950067
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210916015.7A Pending CN115345354A (en) | 2022-08-01 | 2022-08-01 | Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115345354A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117131938A (en) * | 2023-10-26 | 2023-11-28 | 合肥工业大学 | Dynamic implicit relation mining method and system based on graph deep learning |
-
2022
- 2022-08-01 CN CN202210916015.7A patent/CN115345354A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117131938A (en) * | 2023-10-26 | 2023-11-28 | 合肥工业大学 | Dynamic implicit relation mining method and system based on graph deep learning |
| CN117131938B (en) * | 2023-10-26 | 2024-01-19 | 合肥工业大学 | Dynamic implicit relation mining method and system based on graph deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103530694B (en) | The city underground dynamic bus traveler assignment method built based on time-space network | |
| CN113326974B (en) | Multi-source traffic flow prediction method based on multi-task hypergraph convolutional network | |
| CN115062873B (en) | Traffic travel mode prediction method and device, storage medium and electronic device | |
| CN111931998B (en) | Individual travel mode prediction method and system based on mobile positioning data | |
| Song et al. | Simultaneous optimization of 3D alignments and station locations for dedicated high‐speed railways | |
| CN107656987A (en) | A kind of subway station function method for digging based on LDA models | |
| CN109325614B (en) | Bus stop site selection method based on GIS | |
| CN115345354A (en) | Urban multi-mode traffic hyper-network situation prediction method based on hyper-map deep network | |
| CN111191817B (en) | Bus network topology division method based on transfer passenger flow | |
| Huang et al. | A multi-mode traffic flow prediction method with clustering based attention convolution LSTM | |
| CN113159371B (en) | Unknown target feature modeling and demand prediction method based on cross-modal data fusion | |
| CN116913088A (en) | Intelligent flow prediction method for expressway | |
| Ma et al. | An interpretable analytic framework of the relationship between carsharing station development patterns and built environment for sustainable urban transportation | |
| CN106547862A (en) | Traffic big data dimension-reduction treatment method based on manifold learning | |
| CN113641749A (en) | Population mobility network estimation method and system based on bipartite graph | |
| CN116665448A (en) | Traffic speed real-time prediction method and system based on graph convolution network | |
| CN108960684B (en) | Public transport passenger closed travel chain mining method based on incomplete track segment | |
| Miao et al. | A queue hybrid neural network with weather weighted factor for traffic flow prediction | |
| CN115423286A (en) | Urban area rail transit suburb section station attraction range dividing method | |
| Li et al. | Aggregate location recommendation in dynamic transportation networks | |
| CN116894609B (en) | Bus track connection evaluation method based on semi-supervised graph neural network | |
| Bai et al. | Identifying the hotspots in urban areas using taxi GPS trajectories | |
| Yang et al. | A model of bus-metro bi-level weighted network with travel efficiency | |
| SHE et al. | Aggregate Tour-Based Model Framework of Urban Transportation | |
| Xu et al. | Analysis of public transport system efficiency based on multi-layer network |
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 |