A knowledge graph-based city resilience dynamic early warning method and system

Abstract

The application discloses a kind of city resilience dynamic early warning method and system based on knowledge graph, belong to knowledge representation and reasoning technical field, specifically as follows: obtaining the crowd distribution data of carrying multiple personnel identity tags during target activity, and the operation state data of target facility;The above data is respectively loaded to the service layer and the physical layer of the preset knowledge graph, obtain the initial state value of each node, and determine the dependence intensity coefficient of mapping relationship from the personnel identity tag of load node in the support scheme;Receive target disturbance event and determine disturbance node;Disturbance node, initial state value and dependence intensity coefficient are input into pre-trained graph neural network model, with dependence intensity coefficient as cross-layer edge weight, perform alternating state update between physical layer and service layer until the preset convergence condition is met, to generate city resilience early warning signal.Therefore, by implementing the present application, the accuracy of city resilience early warning under major activity support scenario can be improved.

Description

Technical Field

[0001] This invention relates to the field of knowledge representation and reasoning technology, and in particular to a dynamic early warning method and system for urban resilience based on knowledge graphs. Background Technology

[0002] With the increasing frequency of major events such as large-scale urban sporting events, expos, and celebrations, ensuring smooth operation during these events has become a crucial task for urban resilience building and emergency management. During major events, areas surrounding event venues experience sudden, high-density crowds within specific time windows. Simultaneously, to ensure the smooth running of the events, urban management departments develop contingency plans in advance, including crowd segmentation, traffic control, and capacity allocation. This combination of "instantaneous pulsed crowd flow" and "differentiated contingency plans for different crowd groups" places urban infrastructure systems under unconventional operating conditions completely different from daily operations, fundamentally altering their vulnerability characteristics.

[0003] Existing urban resilience assessment methods are mostly based on normal topology and homogeneous pedestrian flow models, which have the following technical shortcomings: First, pedestrian flow data does not distinguish between different types of people, and cannot depict the differentiated travel rules of different customer groups as pre-defined in the major event security plan. Different customer groups have fundamentally different degrees of dependence on infrastructure, and the results of homogeneous model extrapolation are seriously out of sync with the security plan. Second, modeling physical facilities and pedestrian flow areas as a single-layer homogeneous network and using a synchronous update mechanism for extrapolation cannot depict the alternating temporal causal chain of facility failure and pedestrian response in the scenario of major event discontinuation. Third, model weights rely on end-to-end learning of historical data, which cannot be applied to the event-specific topology formed by the major event security plan. Historical data cannot cover the specific layout of each event, the generalization ability of the learned weights is poor, and it cannot reflect the preset rules of the security plan. Fourth, the termination conditions of extrapolation often use a fixed number of iteration layers, which is out of sync with the actual state of the system and cannot dynamically match the strict evacuation time limits preset in the major event, resulting in a distorted output time window. Summary of the Invention

[0004] This invention provides a knowledge graph-based dynamic early warning method and system for urban resilience, which can improve the accuracy of urban resilience early warning in scenarios involving major events.

[0005] This invention provides a knowledge graph-based dynamic early warning method for urban resilience, comprising:

[0006] Acquire pedestrian flow distribution data during the target activity, as well as operational status data of the target facility during the target activity; wherein, the pedestrian flow distribution data includes multiple personnel identification tags; The pedestrian flow distribution data and the operational status data are loaded into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node. Receive the target disturbance event, and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node; The perturbation node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients are input into a pre-trained graph neural network model. The graph neural network model starts with the perturbation node, uses the initial load value and the initial operating state value as the initial state, and uses the dependency strength coefficient as the cross-layer edge weight. It then performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node. Each of the aforementioned operating status values ​​is compared with the corresponding preset threshold, and an urban resilience early warning signal is generated based on the comparison result.

[0007] This invention, through acquiring pedestrian distribution data carrying multiple personnel identification tags, can incorporate the identity information of different customer groups into the deduction process, providing a data foundation for subsequent deduction based on customer group differentiation rules. By loading pedestrian distribution data and operational status data into the service and physical layers of the knowledge graph respectively to obtain initial state values, and determining the dependency strength coefficient of the mapping relationship from the assurance scheme based on the personnel identification tags of the load nodes, the cross-layer mapping relationship can obtain weight values ​​reflecting the preset rules of the assurance scheme. By receiving target disturbance events and using the node corresponding to the event's location as the disturbance node, the starting node of the deduction can be determined. This invention provides a starting point for subsequent alternating state updates. By inputting perturbation nodes, initial state values, and dependency strength coefficients into a pre-trained graph neural network model, and using the dependency strength coefficients as cross-layer edge weights, alternating state updates are performed between the physical and service layers. This process is repeated until the change in the operational state of each facility node meets a preset convergence condition, at which point the operational state value is output. This simulates the alternating temporal sequence of facilities and pedestrian flow, and dynamically terminates the simulation based on the actual convergence. By comparing the operational state values ​​of each facility node with corresponding preset thresholds and generating early warning signals based on the comparison results, this invention can output resilience early warning information based on the current simulation results. Compared to existing technologies that do not consider the coupling effect between facilities and pedestrian flow in major event scenarios, this application can improve the accuracy of urban resilience early warning in major event scenarios.

[0008] Furthermore, acquiring the pedestrian flow distribution data during the target activity includes: Obtain ticketing data for the target event, as well as gate access data for the target event venue; Associate each passage record in the gate passage data with the ticketing data to obtain the personnel identity tag corresponding to each passage record; The number of personnel at each load node is determined based on the number of passage records within the coverage area of ​​each load node. The personnel identity tags corresponding to each passage record and the number of personnel at each load node are used as the pedestrian flow distribution data.

[0009] This invention provides a method to obtain personnel identification tags by associating gate access data with ticketing data, and to determine the number of personnel based on the number of access records within the coverage area of ​​the load node. This method can generate pedestrian flow distribution data carrying personnel identification tags and the number of personnel by fusing data from multiple sources.

[0010] Furthermore, before loading the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, the method further includes: Obtain a protection plan for the target activity; wherein, the protection plan includes the hierarchical structure of each load node and the type classification of each facility node; Based on the hierarchical structure in the aforementioned protection scheme, a service layer of the knowledge graph is constructed; wherein, the service layer includes multiple load nodes, and each load node has a corresponding coverage area; Based on the type classification in the aforementioned protection scheme, the physical layer of the knowledge graph is constructed; wherein, the physical layer includes multiple facility nodes; According to the aforementioned protection scheme, a mapping relationship is established between load nodes and facility nodes that have a service support relationship.

[0011] The embodiments of the present invention obtain a protection scheme and construct a service layer and a physical layer according to the hierarchical structure and type division therein, which can establish a mapping relationship between load nodes and facility nodes with service support relationship.

[0012] Further, the step of loading the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, includes: The pedestrian flow distribution data is loaded into the service layer, and the initial load value of each load node is determined based on the number of people at each load node. The operational status data is loaded into the physical layer, and the access capacity occupancy rate of each facility node is extracted from the operational status data. The initial operational status value of each facility node is determined based on the extraction results.

[0013] This invention, by determining the initial load value based on the number of personnel and extracting the traffic capacity occupancy rate to determine the initial operating state value, can transform multi-source data into the initial state values ​​of each node required by the inference model.

[0014] Further, determining the dependency strength coefficient of the mapping relationship from the protection scheme of the target activity based on the personnel identity tags of each load node includes: Obtain the travel modes planned for each person's identity tag in the aforementioned protection plan; Based on the travel mode, determine the correspondence between each person's identity tag and each facility node; For any load node, based on the personnel identity tag corresponding to the load node and the corresponding relationship, determine the facility node that has a mapping relationship with the load node; The dependency strength coefficient between the load node and the facility node is calculated based on the number of personnel with the personnel identity tags corresponding to the load node.

[0015] This invention provides an embodiment of the invention that obtains the travel mode in the protection plan and determines the correspondence between identity tags and facility nodes. Then, it calculates the dependency strength coefficient based on the personnel identity tags and the number of personnel at the load node. This enables the cross-layer edge weight to be jointly determined by the preset rules of the protection plan and the actual personnel composition.

[0016] Further, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node satisfies the preset convergence condition, and outputting the operating state value of each facility node, includes: When the disturbance node is a facility node, the operating status value of the disturbance node is updated according to the target disturbance event; The change in the operating state of the disturbed node is transmitted along the edge of the physical layer to other facility nodes in the physical layer, and the operating state values ​​of the other facility nodes are updated. Based on the cross-layer edge weights, the changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer for load value updates. The updated load change of each load node is transmitted to the corresponding facility node in the physical layer through the cross-layer edge weight for the next round of operation status value update. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0017] This invention, when the disturbance node is a facility node, first updates the disturbance node, then transmits the change in operating state along the physical layer edge, then updates the load value through the cross-layer edge weight, and then transmits the load change back to the physical layer for the next round of updates. This enables alternating deduction and output of the final state value when the disturbance origin is a facility node.

[0018] Furthermore, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node satisfies the preset convergence condition, and outputting the operating state value of each facility node, also includes: When the disturbance node is a load node, the load value of the disturbance node is updated according to the target disturbance event; The load change of the disturbed node is transmitted to other load nodes in the service layer along the edges of the service layer, and the load values ​​of the other load nodes are updated. Based on the cross-layer edge weights, the load change of each load node is transmitted to the corresponding facility node in the physical layer for updating the operating status value. The updated changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer through the cross-layer edge weights for the next round of load value updates. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0019] This invention, when the disturbance node is a load node, first updates the disturbance node, then transmits the load change along the service layer edge, then updates the facility operation status value through cross-layer edge weights, and then transmits the facility status change back to the service layer for the next round of updates. This enables alternating deduction and output of the final status value when the disturbance origin is a load node.

[0020] Another embodiment of the present invention provides a knowledge graph-based urban resilience dynamic early warning system, comprising: a data acquisition module, a data loading module, a node positioning module, a status update module, and a signal generation module; The data acquisition module is used to acquire pedestrian flow distribution data during the target activity and operational status data of the target facility during the target activity; wherein, the pedestrian flow distribution data includes multiple personnel identity tags; The data loading module is used to load the pedestrian flow distribution data and the operation status data into the service layer and physical layer of the pre-built knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operation status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node. The node positioning module is used to receive target disturbance events and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node. The state update module is used to input the disturbance node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients into a pre-trained graph neural network model, so that the graph neural network model takes the disturbance node as the starting point, the initial load value and the initial operating state value as the initial state, and the dependency strength coefficient as the cross-layer edge weight, and performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node; The signal generation module is used to compare each of the operating status values ​​with the corresponding preset thresholds, and generate an urban resilience early warning signal based on the comparison results.

[0021] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps of the knowledge graph-based dynamic early warning method for urban resilience as described in the present invention.

[0022] Another embodiment of the present invention also provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform steps such as the knowledge graph-based dynamic early warning method for urban resilience of the present invention. Attached Figure Description

[0023] Figure 1 A flowchart illustrating an embodiment of the knowledge graph-based dynamic early warning method for urban resilience provided by the present invention; Figure 2 This is a schematic diagram of the structure of an embodiment of the knowledge graph-based dynamic early warning system for urban resilience provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0026] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0029] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0030] See Figure 1To address the issue that existing technologies do not consider the coupling effect between facilities and pedestrian flow during major events, an embodiment of the present invention provides a dynamic early warning method for urban resilience based on knowledge graphs, including steps S101 to S105: Step S101: Obtain pedestrian flow distribution data during the target activity period, and operational status data of the target facility during the target activity period; wherein, the pedestrian flow distribution data includes multiple personnel identification tags.

[0031] It should be noted that obtaining crowd distribution data during the target event and operational status data of the target facilities during the target event means that, each time an early warning task is executed, two types of dynamic information need to be collected simultaneously: First, crowd distribution data including various personnel identity tags such as audience members, performers, and guests, used to distinguish the gathering and movement of different customer groups in the area surrounding the venue; second, real-time operational status data of the target facilities (such as subway stations, bus transfer points, or VIP parking lots) specified in the target event guarantee plan during the event, used to reflect the current service capacity of these facilities.

[0032] Preferably, acquiring the pedestrian flow distribution data during the target activity includes: Obtain ticketing data for the target event, as well as gate access data for the target event venue; Associate each passage record in the gate passage data with the ticketing data to obtain the personnel identity tag corresponding to each passage record; The number of personnel at each load node is determined based on the number of passage records within the coverage area of ​​each load node. The personnel identity tags corresponding to each passage record and the number of personnel at each load node are used as the pedestrian flow distribution data.

[0033] In one embodiment, assuming the target event is the opening ceremony of a large-scale sports event, it is necessary to obtain the pedestrian flow distribution data around the venue during the opening ceremony. The specific steps are as follows: First, obtain the ticketing data for the opening ceremony and the gate access data at each entrance of the venue. The ticketing data records the personnel identity tags corresponding to each ticket, such as spectators, performers, guests, and staff; the gate access data records the gate number, access timestamp, and associated ticket number for each passage. Second, associate each passage record in the gate access data with the ticketing data. Specifically, using the ticket number as the association key, match each passage record with the corresponding ticket record in the ticketing data to obtain the personnel identity tag corresponding to that passage. For example, if a gate access record has a ticket number of A001, and the personnel identity tag corresponding to this ticket is found to be "spectator" in the ticketing data, then the personnel identity tag for that passage is "spectator." Then, based on the coverage range of each load node in the preset knowledge graph, count the number of passage records within the coverage range of each load node. The coverage area of ​​a load node is a spatial region pre-defined according to the security plan when constructing the knowledge graph. For example, the south exit plaza of the venue covers five turnstiles numbered G01 to G05. The total number of passage records for these five turnstiles during the event needs to be counted as the number of people at that load node. Finally, the personnel identity tags corresponding to each passage record, along with the number of people at that load node, are used together as the pedestrian flow distribution data. This pedestrian flow distribution data carries various personnel identity tags and personnel quantity information, providing a data foundation for calculating the dependency strength coefficient and determining the initial load value in subsequent steps.

[0034] In the above embodiments, it is also necessary to obtain the operational status data of the target facilities (such as subway stations, bus transfer points, and VIP parking lots) specified in the opening ceremony support plan during the event. The operational status data can be obtained from the real-time monitoring system of the urban traffic management department, and may include indicators such as the occupancy rate of the entry capacity of each subway station, the occupancy rate of the waiting area of ​​each bus transfer point, and the remaining number of parking spaces in each VIP parking lot, thereby reflecting the actual service status of these target facilities during the target event and providing a data basis for determining the initial operational status value in subsequent steps.

[0035] Step S102: Load the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node.

[0036] It should be noted that loading the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, means that after acquiring multi-source data, the pedestrian distribution data carrying personnel identification tags is loaded into the service layer of the knowledge graph, and an initial load value is assigned to each load node based on the number of people within the coverage area of ​​each load node. At the same time, the operational status data of the target facility is loaded into the physical layer of the knowledge graph, and an initial operational status value is assigned to each facility node based on the real-time access capacity occupancy rate of each facility node. This transforms the external multi-source data into a standardized initial state that can be directly calculated by the graph neural network.

[0037] Preferably, before loading the pedestrian flow distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, and obtaining the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, the method further includes: Obtain a protection plan for the target activity; wherein, the protection plan includes the hierarchical structure of each load node and the type classification of each facility node; Based on the hierarchical structure in the aforementioned protection scheme, a service layer of the knowledge graph is constructed; wherein, the service layer includes multiple load nodes, and each load node has a corresponding coverage area; Based on the type classification in the aforementioned protection scheme, the physical layer of the knowledge graph is constructed; wherein, the physical layer includes multiple facility nodes; According to the aforementioned protection scheme, a mapping relationship is established between load nodes and facility nodes that have a service support relationship.

[0038] In the above embodiments, the construction process of the preset knowledge graph is based on the pre-completed support plan for the target activity. The specific steps are as follows: First, the support plan for the opening ceremony is obtained. This plan clarifies the hierarchical structure of the load nodes, including remote assembly points and near-end venue nodes, as well as the classification of facility nodes by type, such as subway stations, bus transfer points, and VIP parking lots. Then, according to the hierarchical structure in the support plan, multiple load nodes are constructed in the service layer of the knowledge graph, and a corresponding coverage area is set for each load node. Simultaneously, according to the type classification in the support plan, multiple facility nodes are constructed in the physical layer of the knowledge graph. Finally, according to the service support relationships specified in the support plan, mapping relationships are established between load nodes and facility nodes that have evacuation or connection relationships. For example, if the audience at the south exit plaza mainly relies on subway station B and bus transfer point A for evacuation, then mapping relationships are established between this load node and these two facility nodes respectively.

[0039] Preferably, the step of loading the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, includes: The pedestrian flow distribution data is loaded into the service layer, and the initial load value of each load node is determined based on the number of people at each load node. The operational status data is loaded into the physical layer, and the access capacity occupancy rate of each facility node is extracted from the operational status data. The initial operational status value of each facility node is determined based on the extraction results.

[0040] In the above embodiment, it is necessary to load the pedestrian flow distribution data carrying personnel identification tags into the service layer. For example, for the load node of the south exit plaza, according to statistics, there are 3200 turnstile passage records within the coverage area of ​​this load node, so the initial load value of this load node is determined to be 3200. At the same time, it is necessary to load the operational status data of the target facility into the physical layer. The entry capacity occupancy rate of the subway station B facility node is extracted from the real-time monitoring system as 85%, and based on this, the initial operational status value of this facility node is determined to be 0.85, thereby completing the conversion of multi-source data into the initial status values ​​of each node.

[0041] Preferably, determining the dependency strength coefficient of the mapping relationship from the protection scheme of the target activity based on the personnel identity tags of each load node includes: Obtain the travel modes planned for each person's identity tag in the aforementioned protection plan; Based on the travel mode, determine the correspondence between each person's identity tag and each facility node; For any load node, based on the personnel identity tag corresponding to the load node and the corresponding relationship, determine the facility node that has a mapping relationship with the load node; The dependency strength coefficient between the load node and the facility node is calculated based on the number of personnel with the personnel identity tags corresponding to the load node.

[0042] In the above embodiments, it is necessary to obtain the travel modes planned for each personnel's identity tags in the opening ceremony support plan. For example, audience members are planned to take the subway and buses, guests are planned to take private cars, and performers are planned to take buses. Based on the above travel modes, the correspondence between each personnel's identity tag and each facility node is determined, that is, audience members correspond to subway stations and bus transfer points, guests correspond to VIP parking lots, and performers correspond to bus transfer points. Taking the south exit plaza as a load node, based on the personnel identity tags and correspondences it contains, the facility nodes with mapping relationships with this load node are determined to be subway station B, bus transfer point A, and bus transfer point B. When calculating the dependency strength coefficient of each mapping relationship, the ratio of the number of people corresponding to a certain type of personnel identity tag in the load node to the total number of people in the load node is converted according to the allocation ratio of the travel mode corresponding to that personnel identity tag, thereby transforming the pre-set differentiated travel rules of the customer groups in the support plan into quantitative weight values ​​of each mapping relationship.

[0043] Step S103: Receive the target disturbance event and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node.

[0044] It should be noted that receiving a target disturbance event and using the load node or facility node corresponding to the location of the target disturbance event as the disturbance node means that during the inference process, a target disturbance event (e.g., "a subway station entrance gate malfunction" or "crowd congestion at a venue exit") is received, and then located in the constructed knowledge graph based on the location information contained in the target disturbance event. If the location corresponds to a physical facility (such as a subway station), the facility node is determined as the disturbance node; if the location corresponds to a crowd gathering area (such as a venue exit plaza), the load node is determined as the disturbance node. This node is the starting point for subsequent graph neural network inference.

[0045] In the above embodiments, when the received target disturbance event is "failure of the entrance gate at Metro Station B", the location of this event corresponds to the facility node "Metro Station B" in the physical layer. This facility node is matched in the physical layer of the knowledge graph and marked as a disturbance node. When the received target disturbance event is "crowd congestion at the south exit plaza", the location of this event corresponds to the load node "south exit plaza" in the service layer. This load node is matched in the service layer of the knowledge graph and marked as a disturbance node.

[0046] Step S104: Input the disturbance node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients into the pre-trained graph neural network model, so that the graph neural network model takes the disturbance node as the starting point, the initial load value and the initial operating state value as the initial state, and the dependency strength coefficient as the cross-layer edge weight, and performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node.

[0047] It should be noted that the perturbation node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients are input into the pre-trained graph neural network model. This allows the graph neural network model to perform alternating state updates between the physical layer and the service layer, starting with the perturbation node, using the initial load values ​​and the initial operating state values ​​as initial states, and using the dependency strength coefficients as cross-layer edge weights, until the change in the operating state of each facility node meets the preset convergence condition. Outputting the operating state values ​​of each facility node means that after completing the calculation of the initial state values ​​and dependency strength coefficients, all the input data required for the deduction (perturbation node, initial load value, initial operating state value, and dependency strength coefficient) are fed into the pre-trained graph neural network model. This graph neural network model starts with a perturbation node, using initial load and initial operating state values ​​as the baseline for the initial derivation. Dependency strength coefficients are used as edge weights for cross-layer propagation between the physical and service layers. Alternating state updates are then performed between the physical and service layers: if the perturbation node is a facility node, updates begin at the physical layer, propagating the facility node's state change across layers to the load node in the service layer; if the perturbation node is a load node, updates begin at the service layer, propagating the load node's load change across layers to the facility node in the physical layer. After each round of cross-layer propagation, the receiving layer completes its state update and propagates the new change back to the other layer, repeating this process. When the operating state changes of each facility node meet the preset convergence criteria, the model outputs the final operating state value of each facility node.

[0048] Preferably, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node satisfies a preset convergence condition, and outputting the operating state value of each facility node, includes: When the disturbance node is a facility node, the operating status value of the disturbance node is updated according to the target disturbance event; The change in the operating state of the disturbed node is transmitted along the edge of the physical layer to other facility nodes in the physical layer, and the operating state values ​​of the other facility nodes are updated. Based on the cross-layer edge weights, the changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer for load value updates. The updated load change of each load node is transmitted to the corresponding facility node in the physical layer through the cross-layer edge weight for the next round of operation status value update. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0049] In the above embodiment, the target disturbance event is set as "failure of the entrance gate of Metro Station B". First, the operating status value of the facility node of Metro Station B is directly updated according to the target disturbance event. Then, along the functional dependency edge of the physical layer, the change in the operating status of Metro Station B is transmitted to other facility nodes connected to it, such as the adjacent bus connection point A, and the operating status value of bus connection point A is updated. Next, according to the cross-layer edge weight, the change in the operating status of each facility node is transmitted to the corresponding load node in the service layer, such as the change in Metro Station B is transmitted to the load node of the south exit plaza, and its load value is updated. After that, the updated load change of each load node is transmitted back to the corresponding facility node in the physical layer through the cross-layer edge weight for the next round of operating status value update. This alternating update is performed in a loop until the change in the operating status of each facility node meets the preset convergence condition, and the final operating status value of each facility node is output.

[0050] Preferably, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node meets a preset convergence condition, and outputting the operating state value of each facility node, further includes: When the disturbance node is a load node, the load value of the disturbance node is updated according to the target disturbance event; The load change of the disturbed node is transmitted to other load nodes in the service layer along the edges of the service layer, and the load values ​​of the other load nodes are updated. Based on the cross-layer edge weights, the load change of each load node is transmitted to the corresponding facility node in the physical layer for updating the operating status value. The updated changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer through the cross-layer edge weights for the next round of load value updates. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0051] In the above embodiment, the target disturbance event is set as "crowd congestion at the south exit plaza". First, the load value of the load node of the south exit plaza is directly updated based on the target disturbance event. Then, along the personnel flow path edge of the service layer, the load change of the south exit plaza is transmitted to other adjacent load nodes, such as the north exit plaza, to update the load value of the north exit plaza. Next, according to the cross-layer edge weight, the load change of each load node is transmitted to the corresponding facility node in the physical layer. For example, the change of the south exit plaza is transmitted to subway station B according to the dependency strength coefficient to update the operating status value of subway station B. After that, the updated operating status change of each facility node is transmitted back to the corresponding load node in the service layer through the cross-layer edge weight for the next round of load value update. This alternating update is performed in a loop until the operating status change of each facility node meets the preset convergence condition, and then the final operating status value of each facility node is output.

[0052] Step S105: Compare each of the described operating status values ​​with the corresponding preset thresholds, and generate an urban resilience early warning signal based on the comparison results.

[0053] It should be noted that comparing each operational status value with its corresponding preset threshold and generating an urban resilience early warning signal based on the comparison result means that after the graph neural network model outputs the final operational status value of each facility node, the operational status value of each facility node is compared with a preset threshold for that node. These thresholds are preset based on the design capacity, safety standards, or emergency requirements of each facility in the major event protection plan. When the operational status value of a facility node exceeds its corresponding preset threshold, it indicates that the facility has become a weak link under the current disturbance. An urban resilience early warning signal is then generated based on the comparison result. This signal at least includes the location information and warning level of the weak facility node, providing a basis for emergency decision-making.

[0054] In the above embodiment, after the graph neural network model outputs the final operating status value of each facility node, it compares the operating status value of subway station B (120%) with preset multi-level thresholds (e.g., 80% for yellow warning, 100% for orange warning, and 120% for red warning). Since the operating status value of subway station B reaches 120%, the red warning threshold is triggered, marking it as a vulnerable facility node. A city resilience warning signal carrying the location information of subway station B and the "red" warning level is generated, prompting the emergency management department to immediately take intervention measures such as limiting passenger flow or increasing transport capacity at the facility.

[0055] This invention, through acquiring pedestrian distribution data carrying multiple personnel identification tags, can incorporate the identity information of different customer groups into the deduction process, providing a data foundation for subsequent deduction based on customer group differentiation rules. By loading pedestrian distribution data and operational status data into the service and physical layers of the knowledge graph respectively to obtain initial state values, and determining the dependency strength coefficient of the mapping relationship from the assurance scheme based on the personnel identification tags of the load nodes, the cross-layer mapping relationship can obtain weight values ​​reflecting the preset rules of the assurance scheme. By receiving target disturbance events and using the node corresponding to the event's location as the disturbance node, the starting node of the deduction can be determined. This invention provides a starting point for subsequent alternating state updates. By inputting perturbation nodes, initial state values, and dependency strength coefficients into a pre-trained graph neural network model, and using the dependency strength coefficients as cross-layer edge weights, alternating state updates are performed between the physical and service layers. This process is repeated until the change in the operational state of each facility node meets a preset convergence condition, at which point the operational state value is output. This simulates the alternating temporal sequence of facilities and pedestrian flow, and dynamically terminates the simulation based on the actual convergence. By comparing the operational state values ​​of each facility node with corresponding preset thresholds and generating early warning signals based on the comparison results, this invention can output resilience early warning information based on the current simulation results. Compared to existing technologies that do not consider the coupling effect between facilities and pedestrian flow in major event scenarios, this application can improve the accuracy of urban resilience early warning in major event scenarios.

[0056] Optionally, in this embodiment of the invention, obtaining the pedestrian flow distribution data during the target activity includes: Obtain ticketing data for the target event, as well as gate access data for the target event venue; Associate each passage record in the gate passage data with the ticketing data to obtain the personnel identity tag corresponding to each passage record; The number of personnel at each load node is determined based on the number of passage records within the coverage area of ​​each load node. The personnel identity tags corresponding to each passage record and the number of personnel at each load node are used as the pedestrian flow distribution data.

[0057] This invention provides a method to obtain personnel identification tags by associating gate access data with ticketing data, and to determine the number of personnel based on the number of access records within the coverage area of ​​the load node. This method can generate pedestrian flow distribution data carrying personnel identification tags and the number of personnel by fusing data from multiple sources.

[0058] Optionally, in this embodiment of the invention, before loading the pedestrian flow distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, and obtaining the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, the method further includes: Obtain a protection plan for the target activity; wherein, the protection plan includes the hierarchical structure of each load node and the type classification of each facility node; Based on the hierarchical structure in the aforementioned protection scheme, a service layer of the knowledge graph is constructed; wherein, the service layer includes multiple load nodes, and each load node has a corresponding coverage area; Based on the type classification in the aforementioned protection scheme, the physical layer of the knowledge graph is constructed; wherein, the physical layer includes multiple facility nodes; According to the aforementioned protection scheme, a mapping relationship is established between load nodes and facility nodes that have a service support relationship.

[0059] The embodiments of the present invention obtain a protection scheme and construct a service layer and a physical layer according to the hierarchical structure and type division therein, which can establish a mapping relationship between load nodes and facility nodes with service support relationship.

[0060] Optionally, in this embodiment of the invention, loading the pedestrian distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, includes: The pedestrian flow distribution data is loaded into the service layer, and the initial load value of each load node is determined based on the number of people at each load node. The operational status data is loaded into the physical layer, and the access capacity occupancy rate of each facility node is extracted from the operational status data. The initial operational status value of each facility node is determined based on the extraction results.

[0061] This invention, by determining the initial load value based on the number of personnel and extracting the traffic capacity occupancy rate to determine the initial operating state value, can transform multi-source data into the initial state values ​​of each node required by the inference model.

[0062] Optionally, in this embodiment of the invention, determining the dependency strength coefficient of the mapping relationship from the protection scheme of the target activity based on the personnel identity tags of each load node includes: Obtain the travel modes planned for each person's identity tag in the aforementioned protection plan; Based on the travel mode, determine the correspondence between each person's identity tag and each facility node; For any load node, based on the personnel identity tag corresponding to the load node and the corresponding relationship, determine the facility node that has a mapping relationship with the load node; The dependency strength coefficient between the load node and the facility node is calculated based on the number of personnel with the personnel identity tags corresponding to the load node.

[0063] This invention provides an embodiment of the invention that obtains the travel mode in the protection plan and determines the correspondence between identity tags and facility nodes. Then, it calculates the dependency strength coefficient based on the personnel identity tags and the number of personnel at the load node. This enables the cross-layer edge weight to be jointly determined by the preset rules of the protection plan and the actual personnel composition.

[0064] Optionally, in this embodiment of the invention, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node satisfies a preset convergence condition, and outputting the operating state value of each facility node, includes: When the disturbance node is a facility node, the operating status value of the disturbance node is updated according to the target disturbance event; The change in the operating state of the disturbed node is transmitted along the edge of the physical layer to other facility nodes in the physical layer, and the operating state values ​​of the other facility nodes are updated. Based on the cross-layer edge weights, the changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer for load value updates. The updated load change of each load node is transmitted to the corresponding facility node in the physical layer through the cross-layer edge weight for the next round of operation status value update. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0065] This invention, when the disturbance node is a facility node, first updates the disturbance node, then transmits the change in operating state along the physical layer edge, then updates the load value through the cross-layer edge weight, and then transmits the load change back to the physical layer for the next round of updates. This enables alternating deduction and output of the final state value when the disturbance origin is a facility node.

[0066] Optionally, in this embodiment of the invention, the step of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node satisfies a preset convergence condition, and outputting the operating state value of each facility node, further includes: When the disturbance node is a load node, the load value of the disturbance node is updated according to the target disturbance event; The load change of the disturbed node is transmitted to other load nodes in the service layer along the edges of the service layer, and the load values ​​of the other load nodes are updated. Based on the cross-layer edge weights, the load change of each load node is transmitted to the corresponding facility node in the physical layer for updating the operating status value. The updated changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer through the cross-layer edge weights for the next round of load value updates. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

[0067] This invention, when the disturbance node is a load node, first updates the disturbance node, then transmits the load change along the service layer edge, then updates the facility operation status value through cross-layer edge weights, and then transmits the facility status change back to the service layer for the next round of updates. This enables alternating deduction and output of the final status value when the disturbance origin is a load node.

[0068] like Figure 2 As shown, based on the above method embodiments, corresponding system embodiments are provided; One embodiment of the present invention provides a dynamic early warning system for urban resilience based on knowledge graph, including: a data acquisition module 201, a data loading module 202, a node positioning module 203, a status update module 204, and a signal generation module 205; The data acquisition module 201 is used to acquire pedestrian flow distribution data during the target activity and operational status data of the target facility during the target activity; wherein, the pedestrian flow distribution data includes multiple personnel identity tags; The data loading module 202 is used to load the pedestrian flow distribution data and the operation status data into the service layer and physical layer of the pre-built knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operation status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node; The node positioning module 203 is used to receive the target disturbance event and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node. The state update module 204 is used to input the disturbance node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients into a pre-trained graph neural network model, so that the graph neural network model takes the disturbance node as the starting point, the initial load value and the initial operating state value as the initial state, and the dependency strength coefficient as the cross-layer edge weight, and performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node; The signal generation module 205 is used to compare each of the operating status values ​​with the corresponding preset thresholds, and generate an urban resilience early warning signal based on the comparison results.

[0069] Optionally, in this embodiment of the invention, the data acquisition module 201 includes: a data acquisition submodule, a data association submodule, a data statistics submodule, and a data merging submodule; The data acquisition submodule is used to acquire ticketing data for the target event and gate access data for the target event venue. The data association submodule is used to associate each passage record in the gate passage data with the ticketing data to obtain the personnel identity tag corresponding to each passage record; The data statistics submodule is used to determine the number of personnel at each load node based on the number of passage records within the coverage area of ​​each load node. The data merging submodule is used to use the personnel identity tags corresponding to each passage record and the number of personnel at each load node as pedestrian flow distribution data.

[0070] This invention provides a method to obtain personnel identification tags by associating gate access data with ticketing data, and to determine the number of personnel based on the number of access records within the coverage area of ​​the load node. This method can generate pedestrian flow distribution data carrying personnel identification tags and the number of personnel by fusing data from multiple sources.

[0071] Optionally, in this embodiment of the invention, before the data loading module 202, it further includes: a scheme acquisition submodule, a service layer construction submodule, a physical layer construction submodule, and a mapping establishment submodule; The scheme acquisition submodule is used to acquire the guarantee scheme for the target activity; wherein, the guarantee scheme includes the hierarchical structure of each load node and the type classification of each facility node; The service layer construction submodule is used to construct the service layer of the knowledge graph according to the hierarchical structure in the protection scheme; wherein, the service layer includes multiple load nodes, and each load node has a corresponding coverage area; The physical layer construction submodule is used to construct the physical layer of the knowledge graph according to the type classification in the protection scheme; wherein, the physical layer includes multiple facility nodes; The mapping establishment submodule is used to establish a mapping relationship between load nodes and facility nodes that have a service support relationship, according to the protection scheme.

[0072] The embodiments of the present invention obtain a protection scheme and construct a service layer and a physical layer according to the hierarchical structure and type division therein, which can establish a mapping relationship between load nodes and facility nodes with service support relationship.

[0073] Optionally, in this embodiment of the invention, the data loading module 202 includes: a service layer data loading submodule and a physical layer data loading submodule; The service layer data loading submodule is used to load the people flow distribution data into the service layer and determine the initial load value of each load node based on the number of people at each load node. The physical layer data loading submodule is used to load the operating status data into the physical layer, extract the access capacity occupancy rate of each facility node from the operating status data, and determine the initial operating status value of each facility node based on the extraction result.

[0074] This invention, by determining the initial load value based on the number of personnel and extracting the traffic capacity occupancy rate to determine the initial operating state value, can transform multi-source data into the initial state values ​​of each node required by the inference model.

[0075] Optionally, in this embodiment of the invention, the data loading module 202 further includes: a travel mode acquisition submodule, a correspondence determination submodule, a mapping node determination submodule, and a dependency strength coefficient calculation submodule; The travel mode acquisition submodule is used to acquire the travel modes planned for each person's identity tag in the protection plan; The correspondence determination submodule is used to determine the correspondence between each of the personnel identity tags and each of the facility nodes based on the travel mode; The mapping node determination submodule is used to determine, for any load node, the facility node that has a mapping relationship with the load node based on the personnel identity tag corresponding to the load node and the corresponding relationship; The dependency strength coefficient calculation submodule is used to calculate the dependency strength coefficient between the load node and the facility node based on the number of personnel with the personnel identity tags corresponding to the load node.

[0076] This invention provides an embodiment of the invention that obtains the travel mode in the protection plan and determines the correspondence between identity tags and facility nodes. Then, it calculates the dependency strength coefficient based on the personnel identity tags and the number of personnel at the load node. This enables the cross-layer edge weight to be jointly determined by the preset rules of the protection plan and the actual personnel composition.

[0077] Optionally, in this embodiment of the invention, the state update module 204 includes: a first running state value update submodule, a second running state value update submodule, a first cross-layer update submodule, a second cross-layer update submodule, and a first result output submodule; The first operating status value update submodule is used to update the operating status value of the disturbance node according to the target disturbance event when the disturbance node is a facility node; The second operating status value update submodule is used to transmit the change in operating status of the disturbed node to other facility nodes in the physical layer along the edge of the physical layer, and update the operating status values ​​of the other facility nodes. The first cross-layer update submodule is used to transmit the change in the operating status of each facility node to the corresponding load node in the service layer according to the cross-layer edge weight, so as to update the load value. The second cross-layer update submodule is used to transmit the updated load change of each load node to the corresponding facility node in the physical layer through the cross-layer edge weight, so as to perform the next round of operation status value update. The first result output submodule is used to output the operating status value of each facility node when the change in the operating status of each facility node meets the preset convergence condition.

[0078] This invention, when the disturbance node is a facility node, first updates the disturbance node, then transmits the change in operating state along the physical layer edge, then updates the load value through the cross-layer edge weight, and then transmits the load change back to the physical layer for the next round of updates. This enables alternating deduction and output of the final state value when the disturbance origin is a facility node.

[0079] Optionally, in this embodiment of the invention, the state update module 204 further includes: a first load value update submodule, a second load value update submodule, a third cross-layer update submodule, a fourth cross-layer update submodule, and a second result output submodule; The first load value update submodule is used to update the load value of the disturbance node according to the target disturbance event when the disturbance node is a load node; The second load value update submodule is used to transmit the load change of the disturbed node to other load nodes in the service layer along the edge of the service layer, and update the load value of the other load nodes; The third cross-layer update submodule is used to transmit the load change of each load node to the corresponding facility node in the physical layer according to the cross-layer edge weight, so as to update the running status value. The fourth cross-layer update submodule is used to transmit the updated changes in the operating status of each facility node to the corresponding load node in the service layer through the cross-layer edge weight, so as to perform the next round of load value update. The second result output submodule is used to output the operating status value of each facility node when the change in the operating status of each facility node meets the preset convergence condition.

[0080] This invention, when the disturbance node is a load node, first updates the disturbance node, then transmits the load change along the service layer edge, then updates the facility operation status value through cross-layer edge weights, and then transmits the facility status change back to the service layer for the next round of updates. This enables alternating deduction and output of the final status value when the disturbance origin is a load node.

[0081] It is understood that the above system item embodiments correspond to the method item embodiments of the present invention, and can implement the knowledge graph-based dynamic early warning method for urban resilience provided by any of the above method item embodiments of the present invention.

[0082] This invention, through a data acquisition module 201, acquires pedestrian distribution data carrying multiple personnel identity tags, enabling the introduction of identity information from different customer groups into the deduction process and providing a data foundation for subsequent deduction based on customer group differentiation rules. The data loading module 202 loads the pedestrian distribution data and operational status data into the service and physical layers of the knowledge graph, respectively, to obtain initial state values. Based on the personnel identity tags of the load nodes, it determines the dependency strength coefficient of the mapping relationship from the guarantee scheme, enabling the cross-layer mapping relationship to obtain weight values ​​reflecting the preset rules of the guarantee scheme. The node positioning module 203 receives target disturbance events and uses the node corresponding to the event's location as the disturbance node, thus determining the starting point of the deduction. The initial node provides a starting point for subsequent alternating state updates. The state update module 204 inputs the disturbed node, initial state value, and dependency strength coefficient into a pre-trained graph neural network model. Using the dependency strength coefficient as the cross-layer edge weight, alternating state updates are performed between the physical layer and the service layer. This process is repeated until the change in the operating state of each facility node meets the preset convergence condition, at which point the operating state value is output. This simulates the alternating temporal sequence of facilities and pedestrian flow, and dynamically terminates the simulation based on the actual convergence. The signal generation module 205 compares the operating state value of each facility node with the corresponding preset threshold and generates an early warning signal based on the comparison result, outputting resilience early warning information based on the current simulation results. Compared to existing technologies that do not consider the coupling effect between facilities and pedestrian flow in major event scenarios, this application can improve the accuracy of urban resilience early warning in major event scenarios.

[0083] It should be noted that the system embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0084] Based on the above embodiment of a knowledge graph-based dynamic early warning method for urban resilience, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a knowledge graph-based dynamic early warning method for urban resilience according to any embodiment of the present invention.

[0085] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0086] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0087] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0088] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the knowledge graph-based dynamic early warning method for urban resilience described in any of the above-described method embodiments of the present invention.

[0089] The modules / units integrated into the system / terminal device, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0090] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A dynamic early warning method for urban resilience based on knowledge graphs, characterized in that, include: Acquire pedestrian flow distribution data during the target activity, as well as operational status data of the target facility during the target activity; wherein, the pedestrian flow distribution data includes multiple personnel identification tags; The pedestrian flow distribution data and the operational status data are loaded into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node. Receive the target disturbance event, and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node; The perturbation node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients are input into a pre-trained graph neural network model. The graph neural network model starts with the perturbation node, uses the initial load value and the initial operating state value as the initial state, and uses the dependency strength coefficient as the cross-layer edge weight. It then performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node. Each of the aforementioned operating status values ​​is compared with the corresponding preset threshold, and an urban resilience early warning signal is generated based on the comparison result.

2. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 1, characterized in that, The acquisition of pedestrian flow distribution data during the target event includes: Obtain ticketing data for the target event, as well as gate access data for the target event venue; Associate each passage record in the gate passage data with the ticketing data to obtain the personnel identity tag corresponding to each passage record; The number of personnel at each load node is determined based on the number of passage records within the coverage area of ​​each load node. The personnel identity tags corresponding to each passage record and the number of personnel at each load node are used as the pedestrian flow distribution data.

3. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 1, characterized in that, Before loading the pedestrian flow distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, and obtaining the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, the method further includes: Obtain a protection plan for the target activity; wherein, the protection plan includes the hierarchical structure of each load node and the type classification of each facility node; Based on the hierarchical structure in the aforementioned protection scheme, a service layer of the knowledge graph is constructed; wherein, the service layer includes multiple load nodes, and each load node has a corresponding coverage area; Based on the type classification in the aforementioned protection scheme, the physical layer of the knowledge graph is constructed; wherein, the physical layer includes multiple facility nodes; According to the aforementioned protection scheme, a mapping relationship is established between load nodes and facility nodes that have a service support relationship.

4. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 1, characterized in that, The step of loading the pedestrian flow distribution data and the operational status data into the service layer and physical layer of the pre-constructed knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operational status value of each facility node in the physical layer, includes: The pedestrian flow distribution data is loaded into the service layer, and the initial load value of each load node is determined based on the number of people at each load node. The operational status data is loaded into the physical layer, and the access capacity occupancy rate of each facility node is extracted from the operational status data. The initial operational status value of each facility node is determined based on the extraction results.

5. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 3, characterized in that, The step of determining the dependency strength coefficient of the mapping relationship from the protection scheme of the target activity based on the personnel identity tags of each load node includes: Obtain the travel modes planned for each person's identity tag in the aforementioned protection plan; Based on the travel mode, determine the correspondence between each person's identity tag and each facility node; For any load node, based on the personnel identity tag corresponding to the load node and the corresponding relationship, determine the facility node that has a mapping relationship with the load node; The dependency strength coefficient between the load node and the facility node is calculated based on the number of personnel with the personnel identity tags corresponding to the load node.

6. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 1, characterized in that, The process involves taking the disturbed node as the starting point, the initial load value and the initial operating state value as the initial state, and the dependency strength coefficient as the cross-layer edge weight, performing alternating state updates between the physical layer and the service layer until the operating state changes of each facility node meet a preset convergence condition, and outputting the operating state value of each facility node, including: When the disturbance node is a facility node, the operating status value of the disturbance node is updated according to the target disturbance event; The change in the operating state of the disturbed node is transmitted along the edge of the physical layer to other facility nodes in the physical layer, and the operating state values ​​of the other facility nodes are updated. Based on the cross-layer edge weights, the changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer for load value updates. The updated load change of each load node is transmitted to the corresponding facility node in the physical layer through the cross-layer edge weight for the next round of operation status value update. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

7. The knowledge graph-based dynamic early warning method for urban resilience as described in claim 1, characterized in that, The process of performing alternating state updates between the physical layer and the service layer, starting from the disturbed node, using the initial load value and the initial operating state value as the initial state, and using the dependency strength coefficient as the cross-layer edge weight, until the change in the operating state of each facility node meets the preset convergence condition, and outputting the operating state value of each facility node, further includes: When the disturbance node is a load node, the load value of the disturbance node is updated according to the target disturbance event; The load change of the disturbed node is transmitted to other load nodes in the service layer along the edges of the service layer, and the load values ​​of the other load nodes are updated. Based on the cross-layer edge weights, the load change of each load node is transmitted to the corresponding facility node in the physical layer for updating the operating status value. The updated changes in the operating status of each facility node are transmitted to the corresponding load node in the service layer through the cross-layer edge weights for the next round of load value updates. When the change in the operating status of each facility node meets the preset convergence condition, the operating status value of each facility node is output.

8. A knowledge graph-based dynamic early warning system for urban resilience, characterized in that, include: The module includes a data acquisition module, a data loading module, a node positioning module, a status update module, and a signal generation module. The data acquisition module is used to acquire pedestrian flow distribution data during the target activity and operational status data of the target facility during the target activity; wherein, the pedestrian flow distribution data includes multiple personnel identification tags; The data loading module is used to load the pedestrian flow distribution data and the operation status data into the service layer and physical layer of the pre-built knowledge graph, respectively, to obtain the initial load value of each load node in the service layer and the initial operation status value of each facility node in the physical layer; wherein, for any load node and facility node with a mapping relationship, the dependency strength coefficient of the mapping relationship is determined from the guarantee scheme of the target activity based on the personnel identity tags of each load node. The node positioning module is used to receive target disturbance events and use the load node or facility node corresponding to the location where the target disturbance event occurs as the disturbance node. The state update module is used to input the disturbance node, each of the initial load values, each of the initial operating state values, and each of the dependency strength coefficients into a pre-trained graph neural network model, so that the graph neural network model takes the disturbance node as the starting point, the initial load value and the initial operating state value as the initial state, and the dependency strength coefficient as the cross-layer edge weight, and performs alternating state updates between the physical layer and the service layer until the change in the operating state of each facility node meets the preset convergence condition, and outputs the operating state value of each facility node; The signal generation module is used to compare each of the operating status values ​​with the corresponding preset thresholds, and generate an urban resilience early warning signal based on the comparison results.

9. A terminal device, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements a knowledge graph-based dynamic early warning method for urban resilience as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a knowledge graph-based dynamic early warning method for urban resilience as described in any one of claims 1-7.

Images