Road network key section identification method

Through the method based on graph attention network, the key sections of the road network are identified, and the problems of computing complexity and the limitations of single indicator evaluation in the existing technology are solved, and effective processing of multi-section failure scenarios and road network repair decision support are achieved, thereby improving the Identification accuracy and adaptability.

CN119007146APending Publication Date: 2024-11-22BEIJING JIAOTONG UNIV

Patent Information

Application Number
CN202411029642.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The prior art calculates complexity limitations in the optimization of post-hoc road network repair strategy, and has limitations in the evaluation of section importance based on a single indicator, and lacks the synergy of complex failure scenarios for multiple sections that are damaged simultaneously.

Method used

The critical section recognition method of road network based on graph attention network is adopted. By obtaining the section training samples in the failure scenario, converting them into dual networks, extracting the adjacency matrix and node feature vectors, training the graph attention network model, and capturing the interaction between road segments and Network connectivity relationship, adaptive allocation of weights, and identification of critical road segments.

Benefits of technology

It improves the accuracy and adaptability of identification of critical sections of the road network, can process large-scale transportation network data, reduce real-time computing resource requirements, provide more accurate repair target positioning, and meet real-time decision-making needs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007146A_ABST
    Figure CN119007146A_ABST
Patent Text Reader

Abstract

The invention provides a road network key road section identification method. The method comprises the following steps: obtaining a road section training sample in a failure scene; converting an original road network corresponding to the road section training sample into a dual network, and extracting an adjacent matrix and a node feature vector from the dual network; inputting the road section training sample, the adjacent matrix and the node feature vector into a road network key road section recognition model based on a graph attention network for training so as to obtain a trained road network key road section recognition model; and using the trained road network key road section identification model to identify the to-be-identified road section data in the failure scene to obtain a key road section. The method can gradually capture and understand the global information of the road network and the complex relationship between the invalid road section and the residual network performance, thereby improving the recognition accuracy, efficiently recognizing the key road section to quickly make a repair decision, ensuring that the calculation time is reduced, obtaining a better road network repair effect, and meeting the demand of real-time decision.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of neural network technology, and in particular to a method for identifying key sections of a road network. Background Art

[0002] The road network is an important part of the city's comprehensive transportation system and an important lifeline to support the operation of the city. However, the road network is susceptible to interference from external factors and damage from emergencies, such as natural disasters and deliberate attacks. After a destructive event, part of the structure or function of the road network is destroyed, which may lead to large-scale traffic congestion, unmet travel needs, or disconnection between some areas, seriously affecting residents' lives, social production, and the smooth implementation of disaster relief work. When the road network is damaged, managers need to quickly determine which network components need to be repaired first and how to optimize the repair sequence of damaged components in order to restore the connectivity and service functions of the road network as soon as possible and reduce the impact on disaster relief, material distribution, and personnel transfer. Therefore, it is particularly important to optimize the post-event road network repair strategy. By accurately identifying key sections and effectively repairing them, the normal operation of the transportation system can be restored to the greatest extent, thereby reducing interference with urban traffic and economic losses. Especially in the context of the current accelerated urbanization process, the scale and complexity of urban transportation systems are increasing, and the importance of road networks is becoming increasingly prominent.

[0003] In the prior art, for the optimization problem of road network repair sequence, mixed integer programming models, dynamic programming models, two-layer optimization models, etc. are established to optimize the post-repair sequence and provide decision support for managers. For the post-disaster highway bridge network restoration optimization problem, the restoration trajectory skew is introduced to comprehensively evaluate the network connectivity restoration capacity and economic losses, and the multi-objective restoration optimization model is solved based on the non-dominated sorting genetic algorithm II (NSGA-II) to obtain the optimal restoration sequence. Alternatively, the priority repair target is optimized through a mixed integer programming model to maximize network connectivity, and the search process for large-scale networks is simplified by introducing a community discovery algorithm, effectively reducing the computational burden.

[0004] For the problem of identifying key sections of road networks, on the one hand, the criticality of nodes can be evaluated based on their topological indicators or functional indicators. Among them, topological indicators are quantitative indicators used to describe the overall and local topological characteristics of the network, such as degree centrality, betweenness centrality, eigenvector centrality, nuclearity, closeness centrality, LeaderRank and PageRank. In addition, multiple centrality metrics can be comprehensively considered to evaluate the criticality of nodes, and topological entropy can be used to determine the influence weights of these different metrics. On the other hand, key nodes can also be identified based on road network vulnerability experiments. Specifically, this method causes nodes to lose their normal functions one by one, including reducing the functional attributes of nodes or disconnecting the edges between nodes and other nodes, and by comparing the degree of decline in network performance after each node fails, the criticality of the node in maintaining the normal function of the complex network is evaluated. In addition, different centrality metrics are used to sort the nodes in descending order and delete them one by one, and the performance of different node criticality indicators is compared based on the change curve of the maximum community size of the network.

[0005] However, the existing technologies have the following problems: (1) In the process of optimizing post-event road network repair strategies, existing mathematical optimization models such as mixed integer programming models and dynamic programming models can provide decision support to a certain extent. However, these models are often limited by computational complexity and cannot meet the needs of rapid response and efficient decision-making when faced with large-scale transportation networks.

[0006] (2) Although the method of evaluating the importance of road sections based on topological indicators or functional indicators is simple and intuitive, it has certain limitations. The measurement method of a single indicator cannot fully reflect the actual situation of road sections in a complex road network, and the allocation of weights when integrating multiple indicators lacks a scientific basis.

[0007] (3) The method of assessing the importance of road sections based on road network vulnerability mainly focuses on the failure of a single road section, lacks research on complex failure scenarios where multiple road sections are damaged at the same time, and ignores the synergy and mutual influence between multiple road sections.

[0008] In summary, there is an urgent need for a method to identify critical sections of road networks that can effectively repair damaged road networks to restore their connectivity and service functions during the emergency recovery phase. Summary of the invention

[0009] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for identifying key sections of a road network.

[0010] In order to achieve the above object, the present invention adopts the following technical scheme.

[0011] In a first aspect, the present invention provides a method for identifying key road sections in a road network, comprising:

[0012] Obtain road section training samples under failure scenarios;

[0013] Converting the original road network corresponding to the road segment training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network;

[0014] Inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model;

[0015] The trained road network key section identification model is used to identify the road section data to be identified in the failure scenario to obtain the key section.

[0016] Furthermore, the road network key section identification model based on the graph attention network includes a plurality of graph attention network modules, a fully connected layer and an output layer connected in sequence.

[0017] Furthermore, the step of inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section identification model based on a graph attention network for training includes:

[0018] Using the multiple graph attention network modules to perform feature extraction on the road segment training samples, adjacency matrix and node feature vectors;

[0019] Inputting the extracted multiple feature vectors into the fully connected layer for processing, and inputting the processed feature vectors into the output layer to obtain a road section criticality value vector;

[0020] The road network key section identification model based on the graph attention network is trained and updated according to the road section criticality value vector and the loss function to obtain a trained road network key section identification model.

[0021] Furthermore, the using the multiple graph attention network modules to extract features from the road segment training samples, the adjacency matrix and the node feature vectors includes:

[0022] The following formula is used to extract features from the road segment training samples, adjacency matrix and node feature vectors:

[0023] X (l+1) =σ(GATConv(X (l) ,A)+Linear(X (l) )),l=1,2,...,n

[0024] In the formula, GATConv(X,A) represents the feature vector X of the current layer in the input graph attention network module (l)Perform graph attention convolution operation with adjacency matrix A, Linear(X) represents the feature vector X (l) Perform linear transformation; n is the total number of graph attention network modules in the road network key section recognition model; the graph attention network module passes the feature vector aggregated after graph attention convolution and linear transformation to the next layer through the activation function σ(·).

[0025] Furthermore, the step of converting the original road network corresponding to the road segment training sample into a dual network and extracting an adjacency matrix and a node feature vector from the dual network includes:

[0026] Abstracting the road segments in the original road network as nodes in the dual network, and abstracting the upstream and downstream relationships between the road segments as edges in the dual network;

[0027] Feature extraction is performed based on the nodes and edges in the dual network to obtain the adjacency matrix and node feature vector.

[0028] Furthermore, the node feature vector includes multiple types of road capacity, degree centrality, clustering coefficient, feature vector centrality, nuclearity, PageRank value, weighted betweenness centrality and weighted closeness centrality.

[0029] Furthermore, the road section training samples are obtained by:

[0030] Multiple road sections are randomly removed from a normal road network to simulate multi-section failure scenarios, and the network efficiency of the road network after removal is calculated. Multiple road section failure scenarios are repeatedly generated, and the road section data under extreme failure scenarios are screened out according to the calculated network efficiency as road section training samples.

[0031] In a second aspect, the present invention further provides a road network key section identification system, comprising:

[0032] A sample acquisition module is used to obtain road section training samples under failure scenarios;

[0033] A dual extraction module, used for converting the original road network corresponding to the road section training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network;

[0034] A training module, used for inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model;

[0035] The identification module is used to use the trained road network key section identification model to identify the road section data to be identified in the failure scenario to obtain the key section.

[0036] In a third aspect, the present invention also provides an electronic device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the above-mentioned road network key section identification method.

[0037] In a fourth aspect, the present invention further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned method for identifying key road sections in a road network when the computer program is executed by a processor.

[0038] Beneficial effects of the present invention: The road network key section identification method provided by the present invention performs key section identification through a road network key section identification model based on a graph attention network. The model uses a multi-section failure scenario as a training sample set for training, and stacks multiple graph attention modules to capture the relationship between the interaction between the sections and the remaining network connectivity. At the same time, the attention mechanism is introduced to adaptively assign different weights to different complex traffic features during the back propagation process, so that the model can gradually capture and understand the global information of the road network and the complex relationship between the failed sections and the remaining network performance, thereby improving the recognition accuracy and providing more accurate target positioning for subsequent road network repair. Among them, using a multi-section failure scenario that has a significant impact on the road network connectivity as training data can enhance the adaptability and accuracy of the model when dealing with complex traffic conditions. The method provided by the present application can process large-scale traffic network data, and by combining offline learning with online applications, it greatly reduces the demand for real-time computing resources. Therefore, in the actual road network repair process, the pre-trained model can be used to efficiently identify key sections to quickly make repair decisions, ensuring that while reducing the calculation time, a better road network repair effect is obtained to meet the demand for real-time decision-making.

[0039] Additional aspects and advantages of the present invention will be given in part in the following description, which will become obvious from the following description, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0041] Figure 1 A schematic diagram of a process flow of a method for identifying key road sections in a road network provided by an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a road network key section identification model training process provided by an embodiment of the present invention;

[0043] Figure 3 A schematic diagram of a process for obtaining road segment training samples provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0045] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "an" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or coupling. The term "and / or" used herein includes any unit and all combinations of one or more associated listed items.

[0046] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0047] To facilitate understanding of the embodiments of the present invention, several specific embodiments will be further explained below with reference to the accompanying drawings, and each embodiment does not constitute a limitation on the embodiments of the present invention.

[0048] Example 1

[0049] See also Figure 1 and Figure 2 , a method for identifying key road sections in a road network, comprising the following steps:

[0050] S101, obtaining road section training samples under failure scenarios.

[0051] The road section training samples are obtained in the following manner:

[0052] Multiple road sections are randomly removed from a normal road network to simulate multi-section failure scenarios, and the network efficiency of the road network after removal is calculated. Multiple road section failure scenarios are repeatedly generated, and the road section data under extreme failure scenarios are screened out according to the calculated network efficiency as road section training samples.

[0053] like Figure 3 As shown in the figure, several sections are randomly selected from the original normal road network and made to fail at the same time to simulate the failure scenario of the road network after a destructive event. Then the remaining performance of the road network after these sections fail is calculated. By repeating this sampling process many times, a large number of failure scenarios are obtained, and then the extreme failure scenarios with extremely poor remaining road network performance are screened out to construct a training sample set.

[0054] This step finally outputs the set of failed road sections and their corresponding key feature sets for each training sample. If a road section fails in a specific sample, its corresponding key feature value is 1; if the road section is normal, its corresponding key feature value is 0.

[0055] It should be noted that the selection of extreme failure scenarios is based on network efficiency. The specific approach is as follows: Assume that the shortest path length between node i and node j is d ij , then the transmission efficiency between node i and node j is If multiple road sections lose their normal functions due to emergencies, resulting in a disconnection between nodes i and j, that is, there is no path that can transmit traffic from node i to node j, then d ij =∞, the corresponding transmission efficiency between node i and node j is 0. The network efficiency E of the entire road network is obtained by calculating the average transmission efficiency between all node pairs:

[0056]

[0057] Where N represents the total number of nodes in the road network.

[0058] Assume that in a failure scenario, the sample whose remaining network efficiency is lower than the abnormal threshold can be regarded as an extreme failure scenario. an Then it is calculated by the lower edge of the box plot:

[0059] E an =E 25% -1.5(E 75% -E 25% );

[0060] Among them, E 25% and E 75% They respectively represent the 25th and 75th percentiles of the network efficiency of the remaining road network for several failure scenarios sampled.

[0061] S102, converting the original road network corresponding to the road segment training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network.

[0062] Specifically, the road sections in the original road network are abstracted as nodes in the dual network, and the upstream and downstream relationships between the road sections (specifically reflected as the flow direction of the intersection) are abstracted as edges of the dual network.

[0063] Feature extraction is performed based on the nodes and edges in the dual network to obtain the adjacency matrix and node feature vector.

[0064] The node feature vectors include multiple types of road capacity, degree centrality, clustering coefficient, feature vector centrality, nuclearity, PageRank value, weighted betweenness centrality and weighted closeness centrality.

[0065] The purpose of this step is to more effectively extract and express the characteristics of road sections in the road network, which are mainly obtained by analyzing the node characteristics of the dual network.

[0066] S103, inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model.

[0067] In this step, the graph neural network model is trained using the road segment training samples in S101 and the adjacency matrix and node feature vectors in S102. During the training process, the adjacency matrix reflecting the topological structure of the road network and the feature vector representing the characteristics of the road segment are used as input data, and the set of key features of the road segment of each sample is used as training labels. The model aggregates the feature information of the road segment in the neighborhood through multiple graph attention modules, and then extracts the features of the key road segments. Finally, through the processing of a fully connected layer and an output layer, the model can accurately output the critical values ​​of each road segment. By sorting these critical values ​​from high to low, the key sections in the urban road network can be accurately identified.

[0068] Among them, the road network key section identification model based on graph attention network includes multiple graph attention network modules, a fully connected layer and an output layer connected in sequence.

[0069] Specifically, the multiple graph attention network modules are used to extract features from the road segment training samples, adjacency matrix and node feature vectors.

[0070] The extracted multiple feature vectors are input into the fully connected layer for processing, and the processed feature vectors are input into the output layer to obtain a road section criticality value vector.

[0071] The following formula is used to extract features from the road segment training samples, adjacency matrix and node feature vectors:

[0072] X (l+1) =σ(GATConv(X (l) ,A)+Linear(X (l) )),l=1,2,...,n

[0073] In the formula, GATConv(X,A) represents the feature vector X of the current layer in the input graph attention network module (l) Perform graph attention convolution operation with adjacency matrix A, Linear(X) represents the feature vector X (l) Perform linear transformation; n is the total number of graph attention network modules in the road network key section recognition model; the graph attention network module passes the feature vector aggregated after graph attention convolution and linear transformation to the next layer through the activation function σ(·).

[0074] In addition, by stacking multiple graph attention network modules, the model can aggregate the information of multi-hop adjacent nodes layer by layer. This multi-level feature extraction method helps the model better capture the inherent laws and patterns in graph structure data, thereby further improving the prediction accuracy of the model.

[0075] The fully connected layer is responsible for further integrating and transforming the feature information processed by multiple graph attention network modules. It consists of several linear layers and activation function ReLU, which can effectively transform multi-dimensional feature vectors into one-dimensional feature vectors, aiming to integrate all local features to obtain global feature expression.

[0076] X (l+1) =σ(FCon(X (l) )),l=1,2,...,m

[0077] Where FCon(X) represents the feature vector X of the current layer (l) A linear transformation operation is performed and then passed to the next layer through the activation function σ(·). In this study, the activation function ReLU is used in this layer; m is the total number of linear layers in the fully connected layer.

[0078] The last part of the model is the output layer, which uses the activation function Softmax. Its function is to normalize the criticality values ​​of each road section so that all values ​​fall within the range of [0,1]. The purpose of this is to facilitate intuitive comparison and analysis of the criticality of different road sections.

[0079] Y = Softmax(X (FCn) )

[0080] Among them, X(FCn) It represents the one-dimensional feature vector output by the fully connected layer. After the activation function Softmax, it outputs the critical value vector Y of each road section.

[0081] The road network key section identification model based on the graph attention network is trained and updated according to the road section criticality value vector and the loss function to obtain a trained road network key section identification model.

[0082] It should be noted that the graph attention network module is the core component of the road network key section identification model based on the graph attention network, and carries the main calculation and learning tasks of the model. Each module consists of a graph attention layer, a linear layer, and an activation function ReLU. Together, these components form a powerful learning unit that enables the model to deeply learn and understand graph structured data. The graph attention layer is a key part of the graph attention network module, which can dynamically calculate the attention weights between nodes, allowing the model to adaptively aggregate the information of adjacent nodes and assign unique weights to each node based on the characteristics of each node and the information of its neighboring nodes.

[0083] S104, using the trained road network key section identification model to identify the road section data to be identified in the failure scenario, to obtain the key section.

[0084] According to the method for identifying key sections of a road network provided by an embodiment of the present invention, a key section identification model for a road network based on a graph attention network is used to identify key sections. The model uses a multi-section failure scenario as a training sample set for training, and stacks multiple graph attention modules to capture the relationship between the interaction between sections and the remaining network connectivity. At the same time, an attention mechanism is introduced to adaptively assign different weights to different complex traffic features during the back propagation process, so that the model can gradually capture and understand the global information of the road network and the complex relationship between the failed sections and the remaining network performance, thereby improving the recognition accuracy and providing more accurate target positioning for subsequent road network repair. Among them, using a multi-section failure scenario that has a significant impact on the road network connectivity as training data can enhance the adaptability and accuracy of the model when dealing with complex traffic conditions. The method provided in the present application can process large-scale traffic network data, and by combining offline learning with online applications, it greatly reduces the demand for real-time computing resources. Therefore, in the actual road network repair process, the pre-trained model can be used to efficiently identify key sections to quickly make repair decisions, ensuring that while reducing the calculation time, a better road network repair effect is obtained to meet the demand for real-time decision-making.

[0085] Example 2

[0086] On the basis of Example 1, this Example 2 provides a road network key section identification system, which corresponds to the above-mentioned road network key section identification, and specifically includes:

[0087] A sample acquisition module is used to obtain road section training samples under failure scenarios;

[0088] A dual extraction module, used for converting the original road network corresponding to the road section training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network;

[0089] A training module, used for inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model;

[0090] The identification module is used to use the trained road network key section identification model to identify the road section data to be identified in the failure scenario to obtain the key section.

[0091] For specific details, please refer to the description of the road network key section identification method part, which will not be repeated here.

[0092] Example 3

[0093] Embodiment 3 of the present invention provides an electronic device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, the processor calls the program instructions to execute a method for identifying key road sections in a road network, and the method includes the following process steps:

[0094] Obtain road section training samples under failure scenarios;

[0095] Converting the original road network corresponding to the road segment training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network;

[0096] Inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model;

[0097] The trained road network key section identification model is used to identify the road section data to be identified in the failure scenario to obtain the key section.

[0098] Example 4

[0099] Embodiment 4 of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, a method for identifying key sections of a road network is implemented. The method includes the following steps:

[0100] Obtain road section training samples under failure scenarios;

[0101] Converting the original road network corresponding to the road segment training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network;

[0102] Inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model;

[0103] The trained road network key section identification model is used to identify the road section data to be identified in the failure scenario to obtain the key section.

[0104] In summary, the road network key section identification method provided by the embodiment of the present invention is to identify key sections through a road network key section identification model based on a graph attention network. The model uses a multi-section failure scenario as a training sample set for training, and stacks multiple graph attention modules to capture the relationship between the interaction between sections and the remaining network connectivity. At the same time, the attention mechanism is introduced to adaptively assign different weights to different complex traffic features during the back propagation process, so that the model can gradually capture and understand the global information of the road network and the complex relationship between the failed sections and the remaining network performance, thereby improving the recognition accuracy and providing more accurate target positioning for subsequent road network repair. Among them, using a multi-section failure scenario that has a significant impact on the road network connectivity as training data can enhance the adaptability and accuracy of the model when dealing with complex traffic conditions. The method provided by the present application can process large-scale traffic network data, and by combining offline learning with online applications, it greatly reduces the demand for real-time computing resources. Therefore, in the actual road network repair process, the pre-trained model can be used to efficiently identify key sections to quickly make repair decisions, ensuring that while reducing the calculation time, a better road network repair effect is obtained to meet the demand for real-time decision-making.

[0105] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0106] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the method or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The method and system embodiments described above are merely schematic, in which the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0107] The above are only preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for identifying key road sections in a road network, characterized in that: include: Obtain road section training samples under failure scenarios; Converting the original road network corresponding to the road segment training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network; Inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model; The trained road network key section identification model is used to identify the road section data to be identified in the failure scenario to obtain the key section.

2. The method according to claim 1, characterized in that The road network key section identification model based on graph attention network includes multiple graph attention network modules, a fully connected layer and an output layer connected in sequence.

3. The method according to claim 2, characterized in that The step of inputting the road section training sample, the adjacency matrix and the node feature vector into a road network key section identification model based on a graph attention network for training comprises: Using the multiple graph attention network modules to perform feature extraction on the road segment training samples, adjacency matrix and node feature vectors; Inputting the extracted multiple feature vectors into the fully connected layer for processing, and inputting the processed feature vectors into the output layer to obtain a road section criticality value vector; The road network key section identification model based on the graph attention network is trained and updated according to the road section criticality value vector and the loss function to obtain a trained road network key section identification model.

4. The method according to claim 3, characterized in that The using the multiple graph attention network modules to extract features from the road segment training samples, the adjacency matrix and the node feature vectors includes: The following formula is used to extract features from the road segment training samples, adjacency matrix and node feature vectors: X (l+1) =σ(GATConv(X (l) ,A)+Linear(Z (l) )),l=1,2,...,n In the formula, GATConv(X,A) represents the feature vector X of the current layer in the input graph attention network module (l) Perform graph attention convolution operation with adjacency matrix A, Linear(X) represents the feature vector X (l) Perform linear transformation; n is the total number of graph attention network modules in the road network key section recognition model; the graph attention network module passes the feature vector aggregated after graph attention convolution and linear transformation to the next layer through the activation function σ(·).

5. The method according to claim 1, characterized in that The step of converting the original road network corresponding to the road segment training sample into a dual network and extracting an adjacency matrix and a node feature vector from the dual network includes: Abstracting the road segments in the original road network as nodes in the dual network, and abstracting the upstream and downstream relationships between the road segments as edges in the dual network; Feature extraction is performed based on the nodes and edges in the dual network to obtain the adjacency matrix and node feature vector.

6. The method according to claim 5, characterized in that The node feature vectors include multiple types of road capacity, degree centrality, clustering coefficient, feature vector centrality, nuclearity, PageRank value, weighted betweenness centrality and weighted closeness centrality.

7. The method according to claim 1, characterized in that The road section training samples are obtained in the following manner: Multiple road sections are randomly removed from a normal road network to simulate multi-section failure scenarios, and the network efficiency of the road network after removal is calculated. Multiple road section failure scenarios are repeatedly generated, and the road section data under extreme failure scenarios are screened out according to the calculated network efficiency as road section training samples.

8. A road network key section identification system, characterized in that: include: A sample acquisition module is used to obtain road section training samples under failure scenarios; A dual extraction module, used for converting the original road network corresponding to the road section training sample into a dual network, and extracting an adjacency matrix and a node feature vector from the dual network; A training module, used for inputting the road section training samples, the adjacency matrix and the node feature vector into a road network key section recognition model based on a graph attention network for training, so as to obtain a trained road network key section recognition model; The identification module is used to use the trained road network key section identification model to identify the road section data to be identified in the failure scenario to obtain the key section.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The device stores a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Traffic state prediction method for urban road network based on key road sections

    CN109285346A

  • Complex network key node identification method based on graph attention neural network

    CN114925815A

  • Power grid transient stability key feature screening method, device and system

    CN115392772A

  • Medium voltage distribution network topology identification method and system based on graph attention network

    CN117039889A

Cited By

  • Method and system for identifying key road sections in traffic road network

    CN119763332A

  • Dynamic traffic flow direction migration method for traffic key road identification

    CN120279705A

  • A dynamic traffic flow direction walking method for traffic key road identification

    CN120279705B

  • Key road section identification method based on road network dynamic feature sampling aggregation

    CN120452190A

  • A key road section identification method based on road network dynamic feature sampling and aggregation

    CN120452190B