Urban toughness evaluation method based on multi-dimensional data analysis

Abstract

The invention discloses a city toughness evaluation method based on multi-dimensional data analysis, and relates to the technical field of city toughness evaluation.The city toughness evaluation method comprises the steps that geographic information and multi-source heterogeneous data of an Internet of Things sensor are integrated, and a city subsystem network diagram is constructed through standardization and vectorization processing; a graph neural network is utilized to learn a dependency relationship between nodes, a quantitative interaction matrix is generated, and high-risk links and vulnerabilities are identified; combining real-time sensor data and an attention mechanism to dynamically update interaction strength, simulating a crisis propagation path and calculating a system toughness score; and finally, a city toughness diagnosis report is generated through a visual tool. According to the method, effective fusion of multi-source data and quantitative analysis of interaction between subsystems are realized, and the accuracy and real-time performance of urban toughness evaluation are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of urban resilience assessment, and particularly relates to a method for assessing urban resilience based on multi-dimensional data analysis. BACKGROUND

[0002] With the accelerating process of urbanization, urban systems are increasingly facing various complex challenges such as natural disasters, public health crises, and social unrest. Urban resilience has become a key indicator of measuring the level of sustainable development and safety of a city, and its assessment not only involves the stability of infrastructure, but also covers the coordinated response capabilities in multiple fields such as economy, society, ecology and governance. Scientific assessment of urban resilience helps decision-makers identify system vulnerabilities and optimize resource allocation, thereby enhancing the city's ability to adapt and recover in crisis, and has become a core issue in urban planning and risk management.

[0003] However, the existing methods for assessing urban resilience still have obvious limitations. Many methods rely too much on single-field indicators and fail to fully consider the deep connections between various subsystems in the urban system, making it difficult for the assessment results to fully reflect the city's real ability to respond to crises. In addition, current methods often fail to effectively integrate information from different sources and formats when dealing with multi-source data, such as the differences in format, time and scale between static spatial data provided by geographic information systems and real-time dynamic data generated by Internet of Things sensors, resulting in a lack of comprehensiveness and real-time response capability in the assessment. More critically, existing methods often ignore the interaction mechanisms between urban subsystems, such as the impact of infrastructure failure on social order and economic activity, which cannot be quantitatively analyzed, and this isolated analysis limits the understanding of the overall behavior of the urban system.

[0004] In practical application scenarios, the above problems are particularly prominent. For example, in a rainstorm-induced urban flooding event, managers need to quickly determine which areas of the drainage system may fail and how that failure will affect traffic operation, resident life and emergency rescue efficiency. However, due to heterogeneous data sources, static pipe network data cannot be effectively integrated with real-time rainfall data or social media information, resulting in lagging or inaccurate assessment results. At the same time, the interaction between the transportation system and the drainage system cannot be quantitatively identified, which may lead to underestimation of the vulnerability of some areas, thereby affecting the rational allocation of rescue resources. The lack of data integration and system correlation analysis capabilities makes it difficult for the city to respond to dynamic crises and make optimal decisions in a timely manner.

[0005] Therefore, how to effectively integrate multi-source heterogeneous data and accurately quantify the interaction strength between various subsystems of the city to support real-time resilience assessment of the city in dynamic crisis scenarios has become a key technical problem to be solved in the field. SUMMARY

[0006] To solve the above technical problems, the present application proposes a city resilience assessment method based on multi-dimensional data analysis to solve the problems existing in the prior art.

[0007] In the first aspect, to achieve the above object, the present application provides a city resilience assessment method based on multi-dimensional data analysis, comprising the following steps:

[0008] Collecting multi-source heterogeneous data from geographic information systems and Internet of Things sensors, and performing data standardization processing on the multi-source heterogeneous data to convert it into a unified vector representation to obtain a fused data set;

[0009] Building a city subsystem network graph according to the fused data set, wherein the nodes represent infrastructure and social fields, and the edges represent potential interaction relationships;

[0010] Using graph neural networks to perform embedding learning on the network graph to capture the dependence strength between nodes and obtain a quantized interaction matrix;

[0011] If the value of a certain edge in the interaction matrix exceeds a preset threshold, it is judged as a high-risk link, and by aggregating the features of adjacent nodes, a priority list of vulnerable points is obtained;

[0012] For the priority list of vulnerable points, real-time sensor input is obtained as a dynamic update source, and an attention mechanism is used to focus on key interaction changes to determine the updated interaction strength value;

[0013] According to the updated interaction strength value, the propagation path under a dangerous scenario is simulated to judge the influence degree of the failed node on the overall system, and a resilience assessment score is obtained;

[0014] The resilience scores of each city and each region, the scores of each dimension, and the weak links are presented through a visualization tool to generate a city resilience diagnosis report.

[0015] Optionally, the process of performing data standardization processing on the multi-source heterogeneous data comprises:

[0016] The multi-source heterogeneous data is classified and stored using a data collection protocol to obtain an original data set;

[0017] The format differences of the original data set are processed using a data standardization tool to generate a unified format data set;

[0018] The unified format data set is converted into a vector representation using a vector conversion algorithm to obtain a vector data set;

[0019] If the vector data set meets the preset fusion conditions, the vector data set is integrated using a logical processing method to obtain a fused data set.

[0020] Optionally, the process of constructing the city subsystem network graph comprises:

[0021] Extracting attribute sets of infrastructure nodes and social field nodes from the fused data set, determining node attribute sets;

[0022] According to the node attribute set, a graph construction algorithm is used to generate a network graph containing interaction relationship edges and edge weight values;

[0023] Combined with topological structure analysis and dynamic interaction update, the system connection structure is obtained.

[0024] Optionally, the process of obtaining the priority list of vulnerable points comprises:

[0025] Obtaining edge value data from the interaction matrix, using a threshold comparison tool to judge whether the edge value exceeds a preset threshold, and marking it as a high-risk link if it does, to obtain a high-risk link set;

[0026] According to the high-risk link set, an adjacency matrix extraction tool is used to obtain associated adjacent nodes to generate an adjacent node list;

[0027] For the adjacent node list, a feature aggregation tool is used to integrate the feature vectors and dependency strengths of each adjacent node to obtain an aggregated feature set;

[0028] Through the aggregated feature set, a priority sorting tool is used to calculate the priority of each node's vulnerable point to generate a vulnerable point priority list.

[0029] Optionally, the process of determining the updated interaction intensity value comprises:

[0030] Obtaining an input data stream from a real-time sensor, using a data stream processing tool to perform time series segmentation on the data stream to obtain a time series data segment set;

[0031] For the time series data segment set, an attention mechanism tool is used to calculate the interaction change weight of each data segment to obtain a key interaction change set;

[0032] If the weight of a certain data segment in the key interaction change set exceeds a preset threshold, an interaction intensity calculation tool is used to update the interaction intensity value of the data segment to obtain an updated interaction intensity set.

[0033] Optionally, the process of simulating the propagation path under the dangerous scenario comprises:

[0034] Obtaining an input data stream from a real-time sensor, using a time series analysis tool to segment the data stream to obtain a time series segment set;

[0035] For the time sequence fragment set, a graph network tool is used to construct a propagation path model, to calculate the propagation weight of each node, and to obtain a propagation path set;

[0036] If the propagation weight of a node in the propagation path set exceeds a preset threshold, a failure analysis tool is used to simulate failure of the node, to generate a node failure influence set;

[0037] According to the node failure influence set, a resilience calculation tool is used to calculate the resilience contribution to the overall system, to obtain a resilience evaluation score.

[0038] In a second aspect, the present application also provides a city resilience evaluation system based on multi-dimensional data analysis, for implementing a city resilience evaluation method based on multi-dimensional data analysis, the system comprising:

[0039] A data fusion module is configured to collect multi-source heterogeneous data from a geographic information system and Internet of Things sensors, and to perform data standardization processing on the multi-source heterogeneous data to convert into a unified vector representation, to obtain a fused data set;

[0040] A network graph construction module is configured to construct a city subsystem network graph according to the fused data set, wherein nodes represent infrastructure and social fields, and edges represent potential interaction relationships;

[0041] An interaction matrix generation module is configured to use a graph neural network to perform embedding learning on the network graph, to capture the dependence strength between nodes, and to obtain a quantized interaction matrix;

[0042] A vulnerability point identification module is configured to determine a high-risk link if a value of an edge in the interaction matrix exceeds a preset threshold, to obtain a priority list of vulnerability points by aggregating the features of adjacent nodes;

[0043] A dynamic update module is configured to obtain real-time sensor input as a dynamic update source for the priority list of vulnerability points, to focus on key interaction changes using an attention mechanism, and to determine an updated interaction strength value;

[0044] A resilience evaluation module is configured to simulate propagation paths under a danger scenario according to the updated interaction strength value, to determine the influence degree of a failed node on the overall system, and to obtain a resilience evaluation score;

[0045] A visual report generation module is configured to present the resilience total score, the dimension scores, and the weak links of each city and each region using a visualization tool, and to generate a city resilience diagnosis report.

[0046] In a third aspect, the present application also provides a computer terminal device, comprising:

[0047] One or more processors;

[0048] a memory, coupled to the processor, configured to store one or more programs;

[0049] When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the urban resilience evaluation method based on multi-dimensional data analysis in the first aspect.

[0050] In a fourth aspect, the present application further provides a computer readable storage medium, having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the urban resilience evaluation method based on multi-dimensional data analysis in the first aspect.

[0051] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the steps of the urban resilience evaluation method based on multi-dimensional data analysis in the first aspect.

[0052] Compared with the prior art, the present application has the following advantages and technical effects:

[0053] The urban resilience evaluation method based on multi-dimensional data analysis provided by the present application effectively fuses multi-source heterogeneous data, solves the integration problem caused by different data formats and various sources in traditional evaluation methods, accurately quantifies the interaction strength between urban subsystems through a graph neural network, overcomes the limitation of ignoring system correlation in the prior art, combines real-time sensor data and an attention mechanism to realize dynamic updating and hazard scenario simulation, and improves the real-time performance and accuracy of the evaluation. Finally, the resilience score and weak links are intuitively presented through a visualization tool, providing reliable decision support for urban risk management. BRIEF DESCRIPTION OF DRAWINGS

[0054] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the present application, and the illustrative embodiments thereof and their descriptions are used to explain the present application, and do not constitute improper limitations on the present application. In the drawings:

[0055] Figure 1 It is a total method flowchart of the embodiments of the present application;

[0056] Figure 2 It is a step flowchart of S102 and S102 of the embodiments of the present application;

[0057] Figure 3 It is a step flowchart of S103 and S104 of the embodiments of the present application. DETAILED DESCRIPTION

[0058] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0059] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown.

[0060] Embodiment one

[0061] As Figure 1 shown, the present embodiment provides a city resilience assessment method based on multi-dimensional data analysis, comprising:

[0062] Collecting multi-source heterogeneous data from geographic information systems and Internet of Things sensors, and performing data standardization processing on the multi-source heterogeneous data to convert it into a unified vector representation to obtain a fused data set;

[0063] According to the fused data set, a city subsystem network graph is constructed, wherein the nodes represent infrastructure and social fields, and the edges represent potential interaction relationships;

[0064] A graph neural network is used to perform embedding learning on the network graph to capture the dependence strength between nodes and obtain a quantized interaction matrix;

[0065] If the value of a certain edge in the interaction matrix exceeds a preset threshold, it is judged as a high-risk link, and by aggregating the features of adjacent nodes, a priority list of vulnerable points is obtained;

[0066] For the priority list of the vulnerable points, real-time sensor input is obtained as a dynamic update source, an attention mechanism is used to focus on key interaction changes, and an updated interaction strength value is determined;

[0067] According to the updated interaction strength value, the propagation path under a dangerous scenario is simulated, the influence degree of the failed node on the overall system is judged, and a resilience assessment score is obtained;

[0068] Through a visualization tool, the resilience total score, the dimension score and the weak link of each city and each region are presented, and a city resilience diagnosis report is generated.

[0069] As an implementation manner in the present embodiment, the process of performing data standardization processing on the multi-source heterogeneous data comprises:

[0070] The multi-source heterogeneous data is classified and stored by using a data collection protocol to obtain an original data set;

[0071] The data standardization tool is used to process the format difference of the original data set, and a uniform format data set is generated.

[0072] The uniform format data set is converted into a vector representation by using a vector conversion algorithm, and a vector data set is obtained.

[0073] If the vector data set meets the preset fusion condition, the vector data set is integrated by using a logic processing method, and a fusion data set is obtained.

[0074] Specifically, S101, by collecting multi-source heterogeneous data from geographic information and Internet of Things sensors, the data standardization processing is used to convert the information with different formats into a unified vector representation, and a fused data set is obtained.

[0075] Multi-source heterogeneous data is obtained from a geographic information system and Internet of Things sensors, and the multi-source heterogeneous data is classified and stored by using a data acquisition protocol, and an original data set is obtained. For the original data set, a data standardization tool is used to process the format difference, and a uniform format data set is generated. According to the uniform format data set, a vector conversion algorithm is used to convert the data into a vector representation, and a vector data set is obtained. If the vector data set meets the preset fusion condition, the vector data set is integrated by using a logic processing method, and a fusion data set is obtained.

[0076] Specifically, in the geographic information system (GIS) and Internet of Things (IoT) sensor scene, the processing and fusion of multi-source heterogeneous data is the core of building an efficient data analysis system.

[0077] For example, in the intelligent city traffic management, the GIS provides spatial data of the road network, and the IoT sensor collects real-time information such as vehicle flow, speed, and air quality. These data sources are diverse and have different formats, such as SHP file storage of vector map data of GIS, IoT sensor data may be in JSON or CSV format, and the timestamps and coordinate systems are not unified. The data acquisition protocol can be based on MQTT or HTTP to realize classified storage.

[0078] Exemplarily, the MQTT protocol can be used to collect dynamic data of IoT sensors in real time, and the HTTP protocol is suitable for batch transmission of GIS static data. Through protocol classification, the original data set is stored according to the data type, such as storing traffic flow data into a time series database and storing map data into a spatial database, thereby facilitating subsequent processing. For the original data set, the data standardization tool can solve the format difference problem.

[0079] For example, the timestamp format for IoT sensor data (such as ISO8601 or Unix timestamp) and the coordinate system for GIS data (such as WGS84 or CGCS2000) can be converted and cleaned using ETL tools (such as Apache NiFi) to generate a unified format dataset.

[0080] Specifically, the timestamp is unified as UTC format, and the coordinate system is converted to WGS84 standard to ensure data consistency. This standardization process facilitates subsequent analysis and reduces errors caused by inconsistent formats.

[0081] In one embodiment, the vector conversion algorithm can convert the unified format dataset into a vector representation.

[0082] For example, traffic flow data can be converted into a multi-dimensional vector through feature extraction, including time, location, flow value, etc. GIS spatial data can be converted into a vector representation through rasterization or embedding algorithms (such as Word2Vec variants). Assuming that the flow data for a certain road segment is 1000 vehicles per hour, the speed is 40 km / h, and the air quality index is 50, a vector [1000, 40, 50] can be generated, and GIS data can generate a spatial vector based on the geometric characteristics of the road segment. This vector representation facilitates machine learning model processing and improves data analysis efficiency.

[0083] It should be noted that the vector dataset must meet the fusion conditions, such as data integrity, time synchronization, etc. If the conditions are met, the data can be integrated using logical processing methods.

[0084] For example, the flow vector and spatial vector can be fused to generate a fusion dataset through weighted averaging or decision tree methods.

[0085] Specifically, the fusion dataset can comprehensively reflect the traffic state and geographical characteristics of a certain road segment, such as the comprehensive congestion index of a certain road segment at a certain time.

[0086] Preferably, the fusion process can introduce time weights, with higher weights for recent data to reflect real-time performance. This fusion method can improve the accuracy of traffic prediction and support smart city management decisions.

[0087] It can be understood that the above method has the advantage of reducing data processing complexity through standardization and vector conversion; the fusion dataset provides high-quality input for real-time monitoring and prediction.

[0088] For example, the fusion dataset can be used to predict peak congestion or optimize signal light control, significantly improving traffic efficiency and urban management level.

[0089] As an embodiment of the present embodiment, the process of constructing the city subsystem network graph includes:

[0090] extracting attribute sets of infrastructure nodes and social field nodes from the fused data set, determining node attribute sets;

[0091] generating a network graph containing interaction relationship edges and edge weight values according to the node attribute sets, using a graph construction algorithm;

[0092] obtaining system connection structure in combination with topological structure analysis and dynamic interaction update.

[0093] Specifically, S102, according to the fused data set, a city subsystem network graph is constructed, wherein the nodes represent infrastructure and social fields, the edges represent potential interaction relationships, and a preliminary system connection structure is determined, and the step flow chart of S101 and S102 is as shown in Figure 2 .

[0094] Geospatial data and real-time sensor data are obtained from a geographic information system and Internet of Things sensors, the data is stored in layers using a classification protocol, and a classified data set is obtained. For the classified data set, a standardized tool is used to process format differences to generate a uniform format data set, and a vector conversion algorithm is used to convert the uniform format data set into a vector representation form to obtain a vector data set. If the vector data set meets the preset fusion condition, a logical processing method is used to integrate the vector data set, extract attribute sets of infrastructure nodes and social field nodes, and determine node attribute sets. According to the node attribute sets, a graph construction algorithm is used to generate a network graph containing interaction relationship edges and edge weight values, and in combination with topological structure analysis and dynamic interaction update, a system connection structure is obtained.

[0095] For example, in the infrastructure management scenario of a smart city, geographic information system and Internet of Things sensor data are widely used to monitor the status of urban infrastructure. The geographic information system provides spatial location information of roads, bridges, etc., and is usually stored in a vector format, such as a SHP file; Internet of Things sensors collect dynamic data such as bridge vibration and road surface temperature in real time, and the format may be JSON or CSV. The classification protocol can effectively store these data in layers.

[0096] For example, Internet of Things sensor data is transmitted in real time through the MQTT protocol and stored in a time series database such as InfluxDB to support high-frequency queries; geographic information system data is transmitted in batches through the HTTP protocol and stored in a spatial database such as PostGIS. This hierarchical storage facilitates efficient management of different types of data. For the classified data set, a standardized tool can solve the problem of inconsistent formats.

[0097] In an embodiment, the timestamps of the IoT sensors can be Unix timestamps, while the GIS data adopts the WGS84 coordinate system. Using an ETL tool such as Apache NiFi, the timestamps can be unified into the ISO8601 format, and the coordinate system can be converted into a unified standard, generating a unified format dataset. This standardization process ensures data consistency and facilitates subsequent analysis. A vector conversion algorithm further converts the unified format dataset into a vector representation.

[0098] For example, a bridge sensor collects vibration frequency of 5 Hz and temperature of 25℃, which can generate a vector [5, 25]. The bridge location data of the GIS generates a spatial vector through an embedding algorithm. This vector representation facilitates subsequent processing and is suitable for machine learning models. If the vector dataset meets the fusion conditions, such as time synchronization, a logical processing method can be used to integrate the data.

[0099] In an embodiment, the bridge vibration vector and the location vector are fused by a weighted average method, and a node attribute set is extracted, including the geographical location, vibration frequency, and other attributes of the bridge. These attribute sets provide a basis for network graph construction. Based on the node attribute set, a graph construction algorithm can generate a network graph.

[0100] For example, city bridges serve as nodes, and the interaction relationship between nodes is determined based on traffic flow or physical connection, and the edge weight value can be set according to the distance or flow size between bridges, such as a distance of 2 kilometers between two bridges, and a weight value of 2. Combined with topological structure analysis, the network graph can reflect the spatial relationship between bridges; dynamic interaction updates update the edge weight through real-time sensor data, such as changes in vibration frequency reflecting changes in bridge state. This system connection structure supports dynamic monitoring of urban infrastructure.

[0101] It can be understood that the above method constructs a dynamic network model of urban infrastructure through steps such as hierarchical storage, standardization processing, vector conversion, and graph construction. This model facilitates real-time analysis of bridge state, optimization of maintenance plans, and improvement of urban management efficiency.

[0102] Specifically, S103, a graph neural network is used to perform embedding learning on the network graph to capture the dependence strength between nodes, and obtain a quantized interaction matrix.

[0103] The node features and edge weights are obtained from a classification dataset, the graph neural network algorithm is used for embedding processing on the node features to generate a preliminary embedding vector set, and the preliminary embedding vector set is obtained. According to the preliminary embedding vector set, the interaction weight between nodes is calculated by using an adjacency matrix construction tool, and if the interaction weight is greater than a preset threshold, the embedding vector is updated to obtain an optimized embedding vector set. For the optimized embedding vector set, a relationship pattern extraction tool is used to identify the dependence strength between nodes to generate a quantized relationship pattern matrix, and a relationship pattern matrix is obtained. Through the relationship pattern matrix, a matrix conversion tool is used to integrate node classification and attribute set to generate a quantized matrix containing edge weights and dependence strength, and a final interaction matrix is determined.

[0104] Specifically, in the scenario of smart city infrastructure management, node features and edge weights are obtained from a classification dataset, embedding processing is performed using a graph neural network algorithm to generate a preliminary embedding vector set, and an interaction matrix is constructed based on this to ultimately form a relationship pattern matrix of city subsystems. The following analyzes and exemplifies each technical topic, focusing on the field of city bridge management, and is consistent with the infrastructure dynamic network model in the historical dialogue.

[0105] For example, the classification dataset usually contains sensor data and geographic information data of the bridge. The node features can include the physical attributes of the bridge, such as length, material type; the edge weight can be based on the traffic flow or distance between bridges.

[0106] In one embodiment, a city has three bridges, A bridge length 200 meters, material reinforced concrete, B bridge length 150 meters, material steel, C bridge length 180 meters, material concrete. Sensor data shows that A bridge daily traffic flow is 5000 vehicles, B bridge 3000 vehicles, C bridge 4000 vehicles. When extracting node features, the length and material type of the bridge can be used as features to generate an initial feature set, such as A bridge features [200, reinforced concrete], B bridge [150, steel], and C bridge [180, concrete].

[0107] For example, when the graph neural network algorithm is used for embedding processing on the node features, the graph neural network generates embedding vectors by aggregating neighbor node information. In one possible implementation, the neighbor nodes of A bridge are B bridge and C bridge, and the graph neural network generates the preliminary embedding vector of A bridge according to the traffic flow and the difference in bridge material, such as [0.8, 0.3, 0.5], which represents the comprehensive characteristics of the bridge. B bridge and C bridge similarly generate their respective vectors to form a preliminary embedding vector set. This embedding process captures the potential relationship between bridges through multi-layer aggregation.

[0108] For example, based on the initial embedding vector set, the interaction weight between nodes is calculated using an adjacency matrix construction tool. Assuming the preset threshold is 0.5, the interaction weight between A Bridge and B Bridge is calculated as 0.6 based on traffic flow difference, A Bridge and C Bridge is 0.7, and B Bridge and C Bridge is 0.4. Edges with weight greater than 0.5, such as A-B and A-C, trigger embedding vector update. After updating, the optimized embedding vector of A Bridge may become [0.85, 0.35, 0.55], reflecting a stronger interaction relationship.

[0109] For example, for the optimized embedding vector set, the relationship pattern extraction tool identifies the dependency strength between nodes.

[0110] In one embodiment, A Bridge and B Bridge have high flow interaction and high dependency strength, with a quantitative value of 0.8; B Bridge and C Bridge have low flow and low dependency strength, with a quantitative value of 0.4. The generated relationship pattern matrix records these quantitative values, with matrix elements such as [A-B:0.8, A-C:0.7, B-C:0.4], reflecting the dynamic dependency between bridges.

[0111] For example, through the relationship pattern matrix, the matrix conversion tool integrates node classification and attribute set to generate the final interaction matrix. In one possible implementation, node classification includes bridge type (such as trunk bridge, auxiliary bridge), and attribute set includes length, material, flow. The final interaction matrix integrates these information, with matrix elements not only containing edge weight, but also incorporating dependency strength, such as A-B edge weight 0.6 combined with dependency strength 0.8, generating a comprehensive value of 0.7. This matrix provides a comprehensive relationship view for city bridge management.

[0112] It can be understood that the above method gradually refines the relationship representation between bridges through hierarchical processing from feature extraction to matrix construction. This approach supports city managers to analyze the dynamic interaction between bridges and optimize traffic scheduling or maintenance strategies. Each step is closely linked to ensure data consistency and relationship accuracy, providing reliable support for city infrastructure management.

[0113] As one implementation in the embodiment, the process of obtaining the priority list of vulnerable points includes:

[0114] Obtain edge value data from the interaction matrix, and use a threshold comparison tool to determine if the edge value exceeds a preset threshold, and mark it as a high-risk link to obtain a high-risk link set;

[0115] According to the high-risk link set, use an adjacency matrix extraction tool to obtain associated adjacent nodes to generate an adjacent node list;

[0116] For the adjacent node list, use a feature aggregation tool to integrate the feature vector and dependency strength of each adjacent node to obtain an aggregated feature set;

[0117] The priority ranking tool calculates the priority of each node based on the aggregated feature set, and generates a priority list of vulnerable points.

[0118] Specifically, S104, if the edge value in the interaction matrix exceeds the preset threshold, it is judged as a high-risk link. The priority list of vulnerable points is obtained by aggregating the features of adjacent nodes. The step flow chart of S103 and S104 is as shown in Figure 2

[0119] The edge value data is obtained from the interaction matrix. If the edge value exceeds the preset threshold, the threshold comparison tool is used to mark it as a high-risk link, and a high-risk link set is obtained. According to the high-risk link set, the adjacent nodes associated with the high-risk link set are obtained by using the adjacent matrix extraction tool, and an adjacent node list is generated. For the adjacent node list, the feature aggregation tool is used to integrate the feature vectors and dependency strengths of each adjacent node, and an aggregated feature set is obtained. The priority ranking tool calculates the priority of each node based on the aggregated feature set, and generates a priority list of vulnerable points.

[0120] For example, in the field of city bridge management, the interaction matrix records the edge value data between bridges, reflecting the dynamic relationship strength between bridges. The edge value data is usually derived from traffic flow, geographical distance or structural dependence between bridges. When using the threshold comparison tool, assume that the preset threshold is 0.6. If the edge value exceeds this threshold, the link is marked as a high-risk link, indicating that the interaction between these bridges may lead to risk concentration due to high traffic flow or structural association.

[0121] In one embodiment, a city has three bridges: D bridge, E bridge and F bridge. The interaction matrix shows that the edge value between D bridge and E bridge is 0.65, the edge value between D bridge and F bridge is 0.7, and the edge value between E bridge and F bridge is 0.5. After threshold comparison, the edge values between D bridge and E bridge and between D bridge and F bridge exceed 0.6, and are marked as high-risk links, forming a high-risk link set {D-E, D-F}. This process quickly filters out the relationships that need to be focused on by quantifying the edge value.

[0122] For example, based on the high-risk link set, the adjacent matrix extraction tool is used to obtain the associated adjacent nodes. The adjacent matrix records the connection relationship between bridges, and the extraction tool identifies the relevant nodes by traversing the high-risk links. Continuing the above example, the D-E link indicates that D bridge and E bridge are adjacent nodes, and the D-F link indicates that D bridge and F bridge are adjacent nodes. After extraction, the adjacent node list {D bridge, E bridge, F bridge} is generated.

[0123] It should be noted that the adjacent node list is not limited to directly connected nodes, but may also include indirectly affected nodes, depending on the specific implementation. This method ensures that all key nodes are covered in subsequent analysis.

[0124] ​For example, based on the aggregated feature set, the prioritization tool calculates the vulnerability point priorities of each node, generating a priority list. The vulnerability point priority reflects the importance and potential risk of the bridge in the network, usually evaluated in combination with the traffic, material, and dependency strength in the feature set.

[0125] It can be understood that the above process progresses from edge value screening to priority ranking, ensuring accurate identification of risk points in bridge management. Each step utilizes data from the previous step, with logical rigor and consistent data, providing a reliable technical path for dynamic management of urban bridges.

[0126] As an embodiment in this embodiment, the process of determining the updated interaction intensity value includes:

[0127] An input data stream is obtained from real-time sensors, and a data stream processing tool is used to perform time series segmentation on the data stream to obtain a time series data segment set;

[0128] For the time series data segment set, an attention mechanism tool is used to calculate the interaction change weight of each data segment to obtain a key interaction change set;

[0129] If the weight of a data segment in the key interaction change set exceeds a preset threshold, an interaction intensity calculation tool is used to update the interaction intensity value of the data segment to obtain an updated interaction intensity set.

[0130] Specifically, S105, for the priority list of vulnerability points, real-time sensor input is obtained as a dynamic update source, and an attention mechanism is used to focus on key interaction changes to determine updated interaction intensity values.

[0131] An input data stream is obtained from real-time sensors, and a data stream processing tool is used to perform time series segmentation on the data stream to obtain a time series data segment set. For the time series data segment set, an attention mechanism tool is used to calculate the interaction change weight of each data segment to obtain a key interaction change set. If the weight of a data segment in the key interaction change set exceeds a preset threshold, an interaction intensity calculation tool is used to update the interaction intensity value of the data segment to obtain an updated interaction intensity set. According to the updated interaction intensity set, a feature weight adjustment tool is used to re-calculate the vulnerability point priority in combination with the node features to obtain a dynamically updated vulnerability point priority list.

[0132] Specifically, in the field of urban bridge management, real-time sensor data streams provide key support for dynamic monitoring. The process of real-time sensor obtaining input data stream involves sensor devices deployed on the bridge, such as vibration sensors, stress sensors, and flow counters, continuously collecting dynamic state data of the bridge.

[0133] Exemplarily, A bridge, B bridge and C bridge in a city bridge network are respectively installed with sensors, which record vibration frequency, stress change and vehicle flow in real time, forming continuous data streams. The data streams contain time stamps and multi-dimensional features, such as the vibration frequency of A bridge at a certain time is 5 Hz, the stress value is 200 Pa, and the flow is 5000 vehicles / hour. Such data streams provide a real-time and dynamic basis for subsequent analysis. The data stream processing tool is used to perform time series segmentation on the data streams to obtain a time series data segment set. Time series segmentation is to divide continuous data streams into discrete segments according to time windows.

[0134] For example, the data stream processing tool divides the 24-hour data stream of A bridge into 24 time series data segments with 1 hour as the window, and each data segment contains vibration, stress and flow features within the hour.

[0135] Specifically, the data segment of A bridge in a certain hour may be [5 Hz, 200 Pa, 5000 vehicles], and the data segment of B bridge may be [4 Hz, 180 Pa, 3000 vehicles]. This segmentation method facilitates the analysis of the dynamic changes of bridges in different time periods and provides structured data for subsequent interactive analysis. For the time series data segment set, the attention mechanism tool calculates the interactive change weight of each data segment to obtain a key interactive change set. The attention mechanism identifies key changes by evaluating the correlation strength between data segments.

[0136] For example, the attention mechanism tool analyzes the data segments of A bridge and B bridge in a certain hour, and calculates the interactive change weight between the two bridges based on the similarity of vibration and flow as 0.8, while the weight between A bridge and C bridge is 0.4. Assuming that the preset threshold is 0.6, the interactive change between A bridge and B bridge is marked as a key interactive change, forming a key interactive change set {A-B}. This method highlights the significant changes in the dynamic relationship between bridges and facilitates focusing on high-impact areas. If the weight of a data segment in the key interactive change set exceeds the preset threshold, the interactive strength calculation tool updates the interactive strength value of the data segment to obtain an updated interactive strength set.

[0137] For example, the interactive change weight between A bridge and B bridge is 0.8, which exceeds the threshold 0.6, and the interactive strength calculation tool updates the interactive strength value to 0.75 based on the flow and stress data of the two bridges, reflecting a stronger dynamic correlation between them. The updated interactive strength set may be {A-B: 0.75, A-C: 0.45}. This updating mechanism ensures the dynamic nature and accuracy of the interactive strength. According to the updated interactive strength set, the feature weight adjustment tool recalculates the priority of vulnerable points by combining node features, generating a dynamically updated vulnerable point priority list. The feature weight adjustment tool integrates the physical properties (such as material, length) and dynamic properties (such as interactive strength) of the bridge.

[0138] For example, Bridge A is characterized by [300 meters, reinforced concrete, 5000 vehicles, 0.75], Bridge B by [250 meters, steel, 3000 vehicles, 0.75], and Bridge C by [200 meters, concrete, 2000 vehicles, 0.45]. Through weighted calculation, Bridge A has the highest priority due to high traffic volume and strong interaction intensity, ranking 1; Bridge B ranks second, ranking 2; and Bridge C ranks last, ranking 3. The final priority list is {Bridge A: 1, Bridge B: 2, Bridge C: 3}. This dynamic updating method ensures that the priority reflects real-time changes, providing accurate basis for bridge maintenance.

[0139] As an embodiment in the present embodiment, the process of simulating the propagation path under the dangerous scenario includes:

[0140] Obtain an input data stream from real-time sensors, segment the data stream using a time series analysis tool to obtain a time series segment set;

[0141] For the time series segment set, construct a propagation path model using a graph network tool, calculate the propagation weight of each node, and obtain a propagation path set;

[0142] If the propagation weight of a certain node in the propagation path set exceeds a preset threshold, simulate the failure of the node using a failure analysis tool to generate a node failure impact set;

[0143] According to the node failure impact set, calculate the resilience contribution to the overall system using a resilience calculation tool to obtain a resilience evaluation score.

[0144] Specifically, S106, according to the updated interaction intensity value, simulate the propagation path under the dangerous scenario, judge the influence degree of the failure node on the overall system, and obtain the resilience evaluation score.

[0145] Obtain an input data stream from real-time sensors, segment the data stream using a time series analysis tool to obtain a time series segment set. For the time series segment set, construct a propagation path model using a graph network tool, calculate the propagation weight of each node, and obtain a propagation path set. If the propagation weight of a certain node in the propagation path set exceeds a preset threshold, simulate the failure of the node using a failure analysis tool to generate a node failure impact set. According to the node failure impact set, use a clustering analysis tool to group the failure impacts to obtain a failure impact grouping set. For the failure impact grouping set, use a resilience calculation tool to calculate the resilience contribution of each group to the overall system to obtain a resilience contribution set.

[0146] For example, in the field of urban bridge management, the process of obtaining an input data stream from real-time sensors relies on various devices deployed on the bridge. Vibration sensors, stress sensors, and temperature sensors continuously collect bridge state data to form a data stream containing timestamps and multi-dimensional features.

[0147] Specifically, X Bridge, Y Bridge, and Z Bridge in a certain city's bridge network are each equipped with sensors. At a certain time, X Bridge records a vibration frequency of 6 Hz, a stress value of 220 Pa, and a temperature of 25°C. Y Bridge records a vibration frequency of 5 Hz, a stress value of 190 Pa, and a temperature of 24°C. This data stream provides a dynamic basis for subsequent analysis. Using a time series analysis tool, the data stream is segmented into a set of time series segments. Time series segmentation divides the data stream into discrete segments using a fixed time window, such as 30 minutes.

[0148] For example, the 24-hour data of X Bridge is segmented into 48 time series segments, each containing the vibration, stress, and temperature characteristics within that time period, such as [6 Hz, 220 Pa, 25°C]. This facilitates the capture of short-term changes in bridge status. For the set of time series segments, a graph network tool constructs a propagation path model, calculates the propagation weights of each node, and obtains a set of propagation paths. The graph network tool treats bridges as nodes and the dynamic associations between bridges (such as vehicle flow similarity) as edges.

[0149] For example, X Bridge and Y Bridge have a similar vibration pattern, with a propagation weight of 0.7, while X Bridge and Z Bridge have a weight of 0.3. Setting a threshold of 0.5, the set of propagation paths contains {X-Y}, reflecting the key dynamic association. If a node in the set of propagation paths has a weight exceeding the threshold, a failure analysis tool simulates the failure of that node and generates a set of node failure impacts.

[0150] For example, X Bridge has a weight of 0.7, and after simulating its failure, the analysis shows that Y Bridge's traffic increases by 10% and its stress rises by 15 Pa. This forms the set of failure impacts {X: Y Bridge traffic +10%, stress +15 Pa}, revealing the cascading effects of failure. Using a clustering analysis tool, the set of failure impacts is grouped into a set of failure impact groups.

[0151] For example, based on the degree of impact, the clustering tool groups the impact of X Bridge's failure on Y Bridge into a high-impact group and the impact on Z Bridge into a low-impact group. This grouping facilitates the identification of key impact areas. For the set of failure impact groups, a resilience calculation tool calculates the resilience contribution of each group to the overall system and generates a set of resilience contributions.

[0152] For example, the resilience contribution of the high-impact group is 0.6, and that of the low-impact group is 0.2, reflecting the role of different groups in system stability. Using a weighted summation tool, the resilience contributions are integrated to generate a system resilience score.

[0153] For example, the high-impact group has a weight of 0.7, and the low-impact group has a weight of 0.3, resulting in a system resilience score of 0.48. If the threshold is 0.5, because the score is relatively low, an optimization adjustment tool reconfigures the propagation path model by adjusting the association weight between X Bridge and Y Bridge to 0.65, generating an optimized set of propagation paths. This optimization ensures that the system is more adaptable to dynamic changes and improves management efficiency.

[0154] Specifically, S107, through GIS maps, dashboards, radar charts, and other visualization tools, the total score of the resilience of each city and each region, the score of each dimension, and the weak links are intuitively presented, and a precise "city resilience diagnosis report" is generated.

[0155] The resilience evaluation original data of each city and region is obtained from the multi-source data platform, the noise and redundancy are removed by using the data cleaning tool to obtain the standardized data set. According to the standardized data set, the spatial mapping of the data is carried out by using the geographic information system tool to generate the spatial distribution data set of each city and region. If the resilience value of a certain city or region in the spatial distribution data set is lower than the preset threshold value, the clustering analysis tool is used to group it to obtain the weak resilience region set. For the weak resilience region set, the dimension score distribution of each region is generated by using the radar chart tool to obtain the dimension score visualization set. According to the dimension score visualization set, the total score and the dimension score are integrated by using the dashboard tool to generate the comprehensive visualization data set.

[0156] For example, in the field of city bridge management, the original data of the multi-source data platform usually comes from various sensors and external databases, covering indicators such as vibration, stress, temperature, and flow. The implementation of the data cleaning tool can ensure data consistency through denoising and standardization processing.

[0157] Specifically, the X bridge data of a certain city bridge network may contain abnormal values, such as a sudden increase in vibration frequency to 50 Hz, far exceeding the normal range of 6-8 Hz. The data cleaning tool removes such abnormalities through mean filtering and unifies the temperature unit to Celsius to generate a standardized data set. This processing ensures the accuracy of subsequent analysis.

[0158] In one embodiment, the geographic information system tool maps the standardized data to the city spatial coordinates to form a spatial distribution data set.

[0159] For example, X bridge, Y bridge, and Z bridge are located in the northeast, southwest, and central region of city A, respectively, with resilience values of 0.6, 0.4, and 0.7. The geographic information system tool displays the resilience distribution through a heat map, with the low resilience value of Y bridge highlighted in red on the map, indicating it as a potential risk point. This spatial mapping facilitates intuitive identification of weak regions.

[0160] For example, for regions with resilience values lower than the threshold value of 0.5, the clustering analysis tool can group them based on feature similarity. Y bridge is grouped into the weak region set due to its low resilience value, forming a group with the adjacent W bridge, both of which have high stress and low vibration stability. The clustering analysis tool divides the bridges of city A into high-risk and low-risk groups based on stress, vibration, and other characteristics through the K-means algorithm. This grouping helps managers focus on key areas.

[0161] In an embodiment, the radar chart tool generates a dimension score distribution for the set of weak resilience areas.

[0162] For example, the radar chart of Y Bridge shows that its vibration score is 3, stress score is 2, and temperature stability is 4, clearly showing the weak points in each dimension. The radar chart of W Bridge shows that the vibration score is slightly higher at 4, but the stress score is also low. This visualization facilitates the comparison of the weaknesses of different bridges and assists in the development of targeted maintenance strategies.

[0163] For example, the dashboard tool integrates the resilience total score and the dimension score to form a comprehensive visual data set.

[0164] Specifically, the dashboard of Y Bridge shows that the total resilience value is 0.4, and the pointers of the vibration, stress, and temperature dimensions point to the low, medium, and low areas, respectively, intuitively reflecting its overall state. Through the dashboard, managers can quickly understand that Y Bridge needs to prioritize strengthening the stress monitoring equipment. This comprehensive visualization improves decision-making efficiency and facilitates dynamic adjustment of maintenance plans.

[0165] It should be noted that the above method forms a complete resilience evaluation chain from data cleaning to comprehensive visualization through multi-level analysis. The implementation of each link is based on data, ensuring the objectivity and practicality of the evaluation.

[0166] For example, data cleaning ensures input quality, spatial mapping and clustering analysis accurately locate risk areas, and radar charts and dashboards provide intuitive decision-making basis. This multi-faceted support logic structure enables city bridge management to quickly respond to potential crises in a dynamic environment.

[0167] Based on this, the embodiment of the present application provides a city resilience evaluation method based on multi-dimensional data analysis. The present application realizes the effective fusion of multi-source heterogeneous data, solves the integration problem caused by different data formats and various sources in traditional evaluation methods; accurately quantifies the interaction strength between city subsystems through graph neural networks, overcoming the limitations of ignoring system correlation in the prior art; combines real-time sensor data and attention mechanisms to realize dynamic updating and hazard scenario simulation, improving the real-time and accuracy of the evaluation; and finally presents the resilience score and weak links intuitively through visualization tools, providing reliable decision support for city risk management.

[0168] Embodiment Two

[0169] In this embodiment, a computer terminal device is provided, comprising:

[0170] one or more processors;

[0171] a memory coupled to the processor, for storing one or more programs;

[0172] When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-mentioned urban resilience evaluation method based on multi-dimensional data analysis.

[0173] In this embodiment, a computer-readable storage medium is also provided, and the computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned urban resilience evaluation method based on multi-dimensional data analysis are implemented.

[0174] In this embodiment, an electronic device is also provided, and the electronic device includes a memory and a processor. The memory stores a computer program, and the processor is configured to run the computer program to perform the steps of the above-mentioned urban resilience evaluation method based on multi-dimensional data analysis.

[0175] In this embodiment, a computer program product is also provided, and the computer program product includes a computer program. When the computer program is executed by a processor, the steps of the above-mentioned urban resilience evaluation method based on multi-dimensional data analysis are implemented.

[0176] The above-mentioned program can be run in a processor, or can also be stored in a memory (or called a computer-readable medium). The computer-readable medium includes permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0177] These computer programs can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process, and the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flows Figure 1 The steps of the functions specified in one or more flows or one or more blocks can be implemented by different modules. Figure 1 The steps of the functions specified in one or more flows or one or more blocks can be implemented by different modules.

[0178] Such a device or system is provided in this embodiment. The system is called an urban resilience evaluation system based on multi-dimensional data analysis, and includes:

[0179] a data fusion module configured to collect multi-source heterogeneous data from a geographic information system and Internet of Things sensors, and to perform data standardization processing on the multi-source heterogeneous data to convert the multi-source heterogeneous data into a unified vector representation, to obtain a fused data set;

[0180] a network graph construction module configured to construct a city subsystem network graph according to the fused data set, wherein nodes represent infrastructure and social fields, and edges represent potential interaction relationships;

[0181] an interaction matrix generation module configured to perform embedding learning on the network graph using a graph neural network to capture the dependence strength between nodes, to obtain a quantized interaction matrix;

[0182] a vulnerable point identification module configured to determine that a high-risk link exists if a value of an edge in the interaction matrix exceeds a preset threshold, and to obtain a priority list of vulnerable points by aggregating features of adjacent nodes;

[0183] a dynamic update module configured to obtain real-time sensor input as a dynamic update source for the priority list of vulnerable points, to focus on key interaction changes using an attention mechanism, and to determine an updated interaction strength value;

[0184] a resilience evaluation module configured to simulate a propagation path in a danger scenario according to the updated interaction strength value, to determine the influence degree of a failed node on the overall system, and to obtain a resilience evaluation score;

[0185] a visual report generation module configured to present the resilience total score, the dimension scores, and the weak links of each city and each region using a visualization tool, and to generate a city resilience diagnosis report.

[0186] As an implementation manner in the embodiment, the data fusion module comprises:

[0187] a data classification storage unit configured to perform classification storage on the multi-source heterogeneous data using a data collection protocol, to obtain an original data set;

[0188] a data standardization unit configured to process format differences of the original data set using a data standardization tool, to generate a unified format data set;

[0189] a vector conversion unit configured to convert the unified format data set into a vector representation using a vector conversion algorithm, to obtain a vector data set;

[0190] a logical fusion unit configured to integrate the vector data set using a logical processing method if the vector data set satisfies a preset fusion condition, to obtain a fused data set.

[0191] As an implementation manner in the embodiment, the network graph construction module comprises:

[0192] An attribute extraction unit is configured to extract attribute sets of infrastructure nodes and social field nodes from the fused data set, and determine node attribute sets;

[0193] A graph construction unit is configured to generate a network graph containing interaction relationship edges and edge weight values by using a graph construction algorithm according to the node attribute sets;

[0194] A structure updating unit is configured to obtain a system connection structure by combining topology structure analysis and dynamic interaction updating.

[0195] As an implementation manner in the embodiment, the weak point identification module comprises:

[0196] A threshold comparison unit is configured to obtain edge value data from the interaction matrix, and judge whether the edge value exceeds a preset threshold by using a threshold comparison tool, so as to mark a high-risk link and obtain a high-risk link set;

[0197] An adjacent node extraction unit is configured to obtain associated adjacent nodes by using an adjacent matrix extraction tool according to the high-risk link set, and generate an adjacent node list;

[0198] A feature aggregation unit is configured to integrate feature vectors and dependence strengths of each adjacent node by using a feature aggregation tool for the adjacent node list, and obtain an aggregated feature set;

[0199] A priority sorting unit is configured to calculate weak point priorities of each node by using a priority sorting tool through the aggregated feature set, and generate a weak point priority list.

[0200] As an implementation manner in the embodiment, the dynamic updating module comprises:

[0201] A time sequence segmentation unit is configured to obtain an input data stream from a real-time sensor, and perform time sequence segmentation on the data stream by using a data stream processing tool, so as to obtain a time sequence data segment set;

[0202] An attention calculation unit is configured to calculate interaction change weights of each data segment by using an attention mechanism tool for the time sequence data segment set, so as to obtain a key interaction change set;

[0203] A strength updating unit is configured to update an interaction strength value of a data segment in the key interaction change set by using an interaction strength calculation tool if the weight of the data segment exceeds a preset threshold, so as to obtain an updated interaction strength set.

[0204] As an implementation manner in the embodiment, the resilience evaluation module comprises:

[0205] a time sequence analysis unit, configured to acquire an input data stream from a real-time sensor, and segment the data stream by using a time sequence analysis tool to obtain a time sequence segment set;

[0206] a path modeling unit, configured to construct a propagation path model by using a graph network tool for the time sequence segment set, and calculate a propagation weight of each node to obtain a propagation path set;

[0207] a failure simulation unit, configured to simulate failure of a node in the propagation path set by using a failure analysis tool if the propagation weight of the node exceeds a preset threshold, and generate a node failure influence set;

[0208] a resilience calculation unit, configured to calculate a resilience contribution to a system as a whole by using a resilience calculation tool according to the node failure influence set, and obtain a resilience evaluation score.

[0209] The system or device is used to implement the functions of the method in the above-described embodiments, each module in the system or device corresponds to each step in the method, and has been described in the method and will not be described here.

[0210] Through the above-described embodiments, the problem of urban resilience evaluation based on multi-dimensional data analysis in the related art is solved, thereby being able to guarantee to solve the problems in the prior art.

[0211] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for assessing urban resilience based on multidimensional data analysis, characterized in that, Includes the following steps: Multi-source heterogeneous data is collected from geographic information systems and IoT sensors, and the multi-source heterogeneous data is standardized to be transformed into a unified vector representation to obtain a fused dataset. A network graph of the urban subsystem is constructed based on the fused dataset, where nodes represent infrastructure and social domains, and edges represent potential interaction relationships. A graph neural network is used to perform embedding learning on the network graph to capture the dependency strength between nodes and obtain a quantized interaction matrix. If a certain edge value in the interaction matrix exceeds a preset threshold, it is judged as a high-risk link. By aggregating the features of adjacent nodes, a priority list of vulnerable points is obtained. Based on the priority list of the vulnerabilities, real-time sensor input is obtained as a dynamic update source, and an attention mechanism is used to focus on key interaction changes to determine the updated interaction intensity value. Based on the updated interaction strength value, the propagation path under the crisis scenario is simulated, the impact of the failure node on the overall system is determined, and a resilience assessment score is obtained. Visualization tools are used to present the total resilience score, scores for each dimension, and weaknesses of each city and region, generating a city resilience diagnostic report.

2. The method according to claim 1, characterized in that, The process of standardizing multi-source heterogeneous data includes: The multi-source heterogeneous data is classified and stored using a data acquisition protocol to obtain the original dataset; The format differences of the original dataset are processed using data standardization tools to generate a dataset with a unified format; The unified format dataset is converted into a vector representation using a vector transformation algorithm to obtain a vector dataset. If the vector dataset meets the preset fusion conditions, then a logical processing method is used to integrate the vector dataset to obtain a fused dataset.

3. The method according to claim 1, characterized in that, The process of constructing the urban subsystem network diagram includes: Extract the attribute sets of infrastructure nodes and social domain nodes from the fused dataset to determine the node attribute sets; Based on the node attribute set, a graph construction algorithm is used to generate a network graph containing interaction relationship edges and edge weight values. By combining topological analysis and dynamic interactive updates, the system's connection structure is obtained.

4. The method according to claim 1, characterized in that, The process of obtaining the priority list of vulnerabilities includes: Edge value data is obtained from the interaction matrix. If the edge value exceeds a preset threshold, it is marked as a high-risk link, thus obtaining a set of high-risk links. Based on the set of high-risk links, an adjacency matrix extraction tool is used to obtain the associated adjacency nodes and generate an adjacency node list. For the list of adjacent nodes, a feature aggregation tool is used to integrate the feature vectors and dependency strengths of each adjacent node to obtain an aggregated feature set; Using the aggregated feature set, a priority sorting tool is used to calculate the vulnerability priority of each node, generating a vulnerability priority list.

5. The method according to claim 1, characterized in that, The process of determining the updated interaction strength value includes: The input data stream is acquired from a real-time sensor, and the data stream is time-series segmented using a data stream processing tool to obtain a time-series data segment set. For the aforementioned time-series data segment set, an attention mechanism tool is used to calculate the interaction change weights of each data segment, thereby obtaining a key interaction change set; If the weight of a certain data segment in the set of key interaction changes exceeds a preset threshold, the interaction intensity value of the data segment is updated using an interaction intensity calculation tool to obtain an updated set of interaction intensity.

6. The method according to claim 1, characterized in that, The process of the propagation path in the simulated crisis scenario includes: The input data stream is acquired from real-time sensors, and the data stream is segmented using a time series analysis tool to obtain a set of time series segments; For the aforementioned time series fragment set, a propagation path model is constructed using graph network tools, and the propagation weight of each node is calculated to obtain the propagation path set; If the propagation weight of a node in the propagation path set exceeds a preset threshold, a failure analysis tool is used to simulate the failure of that node and generate a node failure impact set. Based on the set of node failure impacts, a resilience calculation tool is used to calculate the overall system resilience contribution, resulting in a resilience assessment score.

7. A city resilience assessment system based on multidimensional data analysis, characterized in that, The system includes: The data fusion module is used to collect multi-source heterogeneous data from geographic information systems and IoT sensors, and to perform data standardization processing on the multi-source heterogeneous data to transform it into a unified vector representation, thereby obtaining a fused dataset. A network graph construction module is used to construct a network graph of the urban subsystem based on the fused dataset, where nodes represent infrastructure and social domains, and edges represent potential interaction relationships. The interaction matrix generation module is used to perform embedding learning on the network graph using a graph neural network, capture the dependency strength between nodes, and obtain a quantized interaction matrix. The vulnerability identification module is used to determine a high-risk link if a certain edge value in the interaction matrix exceeds a preset threshold, and obtain a priority list of vulnerability points by aggregating the features of adjacent nodes. The dynamic update module is used to obtain real-time sensor input as a dynamic update source for the priority list of the vulnerability points, and use an attention mechanism to focus on key interaction changes to determine the updated interaction intensity value. The resilience assessment module is used to simulate the propagation path under crisis scenarios based on the updated interaction strength value, determine the impact of failure nodes on the overall system, and obtain a resilience assessment score. The visualization report generation module is used to present the total resilience score, scores of each dimension, and weak links of each city and region through visualization tools, and generate a city resilience diagnostic report.

8. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6.

Images