Time sequence toughness mode recognition method and system for urban emergency network

Abstract

The invention provides a time sequence toughness mode recognition method and system for an urban emergency network, and belongs to the technical field of urban emergency rescue, and the method comprises the steps: obtaining the network flow data of the urban emergency network; constructing a time sequence network based on the network flow data, analyzing time sequence network indexes, and calculating network toughness; analyzing the network toughness based on a singular spectrum, and extracting a sequential network trend and a feature vector; performing independent clustering analysis on the feature vectors by adopting various clustering methods to obtain a plurality of clustering results; and screening a consistency result of multiple clustering methods through a frequency threshold, and identifying a significant toughness mode. The method can systematically identify a significant and stable toughness mode in a city emergency scene.

Description

Technical Field

[0001] The present invention belongs to the technical field of urban emergency rescue, and in particular relates to a temporal resilience pattern recognition method and system for urban emergency networks. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the acceleration of urbanization, urban emergency rescue systems face increasingly complex challenges. As critical infrastructure for public safety, the resilience of urban emergency rescue networks is directly related to the efficiency and effectiveness of emergency response during disasters. In recent years, with the development of big data and complex network analysis technologies, the dynamic characteristics of urban emergency rescue networks have become a research hotspot. However, existing technologies primarily focus on static network analysis, which struggles to effectively capture the dynamic evolution of emergency rescue networks across time and space.

[0004] In urban emergency rescue scenarios, the connections between network nodes (such as hospitals, fire stations, and material storage centers) dynamically adjust over time. For example, after a disaster, some nodes may fail due to resource depletion or traffic disruption, while other nodes may temporarily connect to enhance the overall resilience of the network. Therefore, how to dynamically quantify the temporal resilience of urban emergency rescue networks and identify significant and stable resilience patterns has become a pressing technical challenge.

[0005] Existing technologies often rely on single indicators (such as network connectivity) or static models, which cannot fully reflect the dynamic changes of the network in the spatial and temporal dimensions. In addition, traditional clustering methods lack robustness in dynamic pattern recognition and have difficulty screening out significant and stable resilience patterns. Existing technologies specifically have the following problems:

[0006] Static network analysis methods, including graph-theory-based network analysis (such as shortest path algorithms and community detection algorithms) and statistical-based network feature extraction (such as node degree distribution and clustering coefficient), are often based on network modeling within a fixed time window. By constructing a point-in-time emergency rescue network graph and analyzing static metrics such as connectivity and centrality, they ignore the dynamic evolution of the network. Static network models are unable to adapt to the dynamic changes in urban emergency rescue networks across time and space. For example, after a disaster, the network structure may undergo drastic changes, and static models cannot capture these dynamic characteristics.

[0007] Single-metric resilience assessment methods, including connectivity-based and node importance-based resilience assessment models, often rely on a single metric (such as network connectivity or node centrality) to evaluate network resilience. For example, they measure network resilience by calculating the maximum size of a network's connected subgraph. Relying on a single metric (such as network connectivity) to assess network resilience fails to fully reflect the complex characteristics of a network in both temporal and spatial dimensions. For example, a single connectivity metric cannot quantify network heterogeneity or dynamic evolution.

[0008] Traditional clustering methods, including distance-based and density-based clustering algorithms, use the K-means algorithm to cluster network nodes and identify key nodes or community structures. These methods often use a single clustering algorithm (such as K-means or hierarchical clustering) to analyze network structure without considering dynamic trends or algorithm co-validation. Using a single algorithm is susceptible to noise, leading to pattern misjudgment. When dealing with complex spatiotemporal data, the stability and reliability of the results are difficult to guarantee. For example, the K-means algorithm is sensitive to initial cluster centers, which can lead to unstable clustering results.

[0009] Trend extraction techniques, including sliding window-based trend extraction and frequency domain analysis-based signal decomposition, use moving averages to smooth time series data and extract long-term trends. However, simple moving averages or Fourier transforms are often used to extract time series trends, making it difficult to separate noise and extract key trend components. Simple smoothing methods are also difficult to effectively separate noise and extract key trend components. For example, moving averages can mask sudden changes in time series data, leading to inaccurate trend analysis.

[0010] Existing technologies lack comprehensive quantitative methods for the temporal resilience of urban emergency rescue networks, making it difficult to provide accurate decision-making support for disaster response. Existing methods are unable to identify significant resilience patterns across time and space, making it difficult to guide the optimal allocation of emergency rescue resources. Summary of the Invention

[0011] To overcome the deficiencies of the above-mentioned prior art, the present invention proposes a temporal resilience pattern recognition method and system for urban emergency networks. By constructing a temporal resilience clustering method that integrates multi-dimensional indicator calculation, trend feature extraction, and multi-algorithm collaboration, it achieves accurate identification and dynamic evaluation of the resilience pattern of urban emergency rescue networks. This method can not only reflect the evolutionary characteristics of network structure and function in real time, but also significantly improve the reliability of pattern recognition results through a multi-algorithm cross-validation mechanism. This has important practical significance for optimizing emergency response strategies, improving the efficiency of rescue resource scheduling, and warning of potential network risks.

[0012] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0013] In a first aspect, a temporal resilience pattern recognition method for urban emergency networks is disclosed, comprising:

[0014] Obtain network flow data for the city’s emergency network;

[0015] Building a time series network based on the network flow data, analyzing time series network indicators, and calculating network resilience;

[0016] Extracting a time series network trend based on the network resilience through singular spectrum analysis, extracting key features based on the time series network trend, wherein the key features include trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and constructing a feature vector based on the key features;

[0017] Multiple clustering methods are used to perform independent cluster analysis on the feature vectors to obtain multiple clustering results;

[0018] Frequency thresholds were used to screen the consistency results of multiple clustering methods and identify significant resilience patterns.

[0019] Second, a temporal resilience pattern recognition system for urban emergency networks is disclosed, including:

[0020] A data acquisition module is configured to: acquire network flow data of a city emergency network;

[0021] A temporal complex network construction module, configured to: construct a temporal network based on the network flow data;

[0022] A resilience index calculation module is configured to: analyze time series network indicators and calculate network resilience;

[0023] a trend extraction module configured to: extract a time series network trend based on the network resilience through singular spectrum analysis, extract key features based on the time series network trend, the key features including trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and construct a feature vector based on the key features;

[0024] A multi-algorithm clustering module is configured to: use multiple clustering methods to perform independent clustering analysis on feature vectors to obtain multiple clustering results;

[0025] The pattern recognition module is configured to: filter the consistency results of multiple clustering methods through frequency thresholds to identify significant resilience patterns.

[0026] In a third aspect, an electronic device is disclosed, including a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are run by the processor, the steps of the above-mentioned temporal resilience pattern recognition method for urban emergency networks are completed.

[0027] In a fourth aspect, a computer-readable storage medium is disclosed for storing computer instructions. When the computer instructions are executed by a processor, the steps of the above-mentioned temporal resilience pattern recognition method for urban emergency networks are completed.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The present invention proposes a complete framework for quantifying temporal network resilience, which constructs a dynamic network model by capturing changes in network node states and evolution of edge weights in real time. The framework innovatively integrates multi-dimensional indicators such as network connectivity, resource distribution heterogeneity, and node centrality to achieve a comprehensive measurement of network resilience. In the indicator calculation process, the Gini coefficient is used to characterize the uneven distribution of resources, and the dynamic evolution rate is introduced to reflect the characteristics of network changes, forming a scientific and complete evaluation system. This framework breaks through the limitations of traditional static network analysis and can accurately reflect the dynamic evolution characteristics of network structure and function. The fusion of multi-dimensional indicators provides more comprehensive and in-depth insights into network resilience, and provides reliable data support for emergency decision-making. As the core technical foundation of the present invention, the framework provides key support for subsequent pattern recognition and decision optimization, and is the fundamental guarantee for achieving accurate resilience assessment.

[0030] This invention innovatively applies singular spectrum analysis (SSA) to trend extraction of time series resilience data. Through the three key steps of constructing an embedding matrix, singular value decomposition, and trend reconstruction, the precise separation of long-term trends, periodic fluctuations, and mutation characteristics is achieved. This method introduces an adaptive window length selection mechanism to ensure the accuracy of trend extraction. This framework breaks through the limitations of traditional static network analysis and can accurately reflect the dynamic evolution characteristics of network structure and function. The fusion of multi-dimensional indicators provides more comprehensive and in-depth insights into network resilience, providing reliable data support for emergency decision-making.

[0031] This paper designs a multi-algorithm collaborative resilient pattern recognition mechanism that applies multiple clustering algorithms in parallel, combined with consistency verification and threshold screening to ensure the reliability of recognition results. This mechanism innovatively introduces a screening criterion based on pattern frequency, effectively improving the stability of the results. Through cross-validation of multiple algorithms, the potential bias introduced by a single algorithm is significantly reduced, improving the accuracy and reliability of pattern recognition.

[0032] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0034] Figure 1 This is a flow chart of a method for identifying temporal resilience patterns for urban emergency networks described in Example 1 of the present invention.

[0035] Figure 2 This is a diagram of the specific steps of a method for identifying temporal resilience patterns for urban emergency networks described in Example 1 of the present invention.

[0036] Figure 3 A timing resilience pattern diagram is extracted for a timing resilience pattern recognition method for a city emergency network described in Example 1 of the present invention.

[0037] Figure 4 This is a diagram illustrating a pattern sample extracted from a method for identifying temporal resilience patterns for urban emergency networks described in the first embodiment of the present invention.

[0038] Figure 5 This is a pattern index diagram extracted from a temporal resilience pattern recognition method for a city emergency network described in the first embodiment of the present invention. DETAILED DESCRIPTION

[0039] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0040] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0041] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0042] Explanation of terms:

[0043] 1. Temporal Complex Network

[0044] Describes a network system that evolves over time, where the set of nodes remains constant while the set of edges changes over time. This network model can capture the evolution of the dynamic interactions between system components.

[0045] 2. Singular Spectrum Analysis (SSA)

[0046] A nonparametric time series analysis method that decomposes a time series into trend, cycle, and noise components through a decomposition and reconstruction step. This method does not require assumptions about the stationarity or distribution of the data.

[0047] 3. Resilience Indicators

[0048] Metrics used to quantify a system's ability to resist disturbances and recover functionality include:

[0049] Network Connectivity: measures the degree of interconnection between nodes in a network;

[0050] Network Density: represents the ratio of the actual number of connections to the possible number of connections;

[0051] Network Centrality: Evaluates the importance of a node in the network.

[0052] 4. Multi-algorithm Clustering

[0053] A method that uses a combination of multiple clustering algorithms to improve the reliability of clustering results through a consistency evaluation mechanism.

[0054] Example 1

[0055] In one or more embodiments, a sequential resilience clustering (SRC) algorithm is proposed to systematically identify significant and stable resilience patterns from temporally complex networks. The SRC method integrates temporally complex network construction, resilience metric calculation, singular spectrum analysis (SSA) trend extraction, application of multiple clustering algorithms, and a mechanism to ensure pattern recognition accuracy and robustness.

[0056] This embodiment discloses a temporal resilience pattern recognition method for urban emergency networks, such as Figure 1-Figure 2 As shown, the following steps are included:

[0057] Step S1: Obtain network flow data of the city emergency network;

[0058] Network flow data specifically include: traffic flow data (GPS trajectory, speed, direction, and frequency of floating vehicles), crowd movement trajectory (spatiotemporal distribution, aggregation density, and migration patterns based on mobile devices), emergency resource allocation information (material flow direction, quantity, and timestamp), key node traffic (inbound and outbound traffic of transportation hubs, medical institutions, and shelters), communication network load (information transmission frequency, bandwidth occupancy, and connectivity changes), energy supply flow (flow and pressure changes in electricity, water, and gas networks), and abnormal traffic fluctuations caused by emergencies (cascading failure modes and dynamic changes in topological structures), and other multi-dimensional and multi-scale spatiotemporal data sets.

[0059] Step S2: constructing a time series network based on the network flow data, analyzing time series network indicators, and calculating network resilience;

[0060] Step S2-1: Based on multiple time nodes t, construct the node set V and the dynamic edge set E(t), calculate the edge weight w, and build a temporal network;

[0061] In this example, the emergency sequence network is constructed as follows: each airport is represented as a node, and the connections between nodes are represented as weighted edge sets. For any two nodes, if there is a path between them, an edge is established between them. The weight of the edge is determined by the number of connections between the two nodes.

[0062] The emergency timing network can be expressed as:

[0063] G=(V,E) (1)

[0064] Where G represents the aviation network graph, V represents the node set, and E represents the edge set. Each edge e ij ∈E is defined by the following triples:

[0065] e ij =(i,j,w ij ) (2)

[0066] Among them, w ij is the weight of the edge, defined as the weight number between nodes; i, j are the departure node and arrival node of the directed edge.

[0067] w ij =n ij (3)

[0068] Among them, n ij is the weight of the node in the emergent network. In this representation, the weight of the edge is included in the structure of the graph as an attribute of the edge.

[0069] Step S2-2: Analyze network connectivity NV, size NS, density ND, centrality NC and Gini coefficient Gini, and calculate the resilience index;

[0070] Temporal resilience refers to the ability of a network to maintain its functionality over time despite dynamic disruptions or changes. Temporal resilience incorporates the temporal evolution of a network's structural properties and assesses how these changes affect the network's robustness.

[0071] Temporal resilience is defined as:

[0072]

[0073] Among them, R t For timing resilience, NV t For network connectivity, NS t is the network size, ND t is the network density, NC t is the network centrality, Gini(k t ) is the Gini coefficient.

[0074] Network connectivity indicates the degree of interconnection between different ports at time t. t Indicates the number of possible paths from the starting point to the destination in the network. t A higher value indicates a well-connected network that can provide alternative routes during disruptions, thus increasing resilience. t Quantize using the maximum number of possible connections:

[0075] NV t =n t ×(n t -1) (5)

[0076] Among them, n t Indicates the number of nodes in the network.

[0077] The network size refers to the total number of nodes and edges in the network at a certain time. t Reflects its scale and potential redundancy. The calculation formula is:

[0078] NS t =n t +m t (6)

[0079] Among them, n t Indicates the number of nodes each time, m t The number of edges representing the moment.

[0080] Network density measures the ratio of actual connections to all possible connections at the moment, indicating the overall connectivity level. t It is defined as:

[0081]

[0082] Among them, |E t | represents the number of edges at time t; n t represents the number of nodes at time t;

[0083] Network centrality evaluates the average importance or influence of nodes in the network over time, usually measured by degree centrality. t The calculation formula is:

[0084]

[0085] Among them, DC i,t represents the degree centrality of node i at time t, n t Represents the number of nodes at time t.

[0086] The Gini coefficient reflects the inequality of the distribution of node degrees over time, indicating the heterogeneity of the network. t The lower the heterogeneity, the more balanced the network distribution and the higher the resilience. The Gini coefficient is calculated using the Lorenz curve of the network degree distribution:

[0087]

[0088] Among them, P i represents the cumulative proportion of node i, indicating the relative position of the node when the nodes are sorted in ascending order of degree, and L i represents the cumulative proportion of node degrees to node i, and d j,t represents the degree of node j at time t.

[0089] This embodiment uses the Gini coefficient to characterize resource inequality and introduces a dynamic evolution rate to reflect network change characteristics. The fusion of these multi-dimensional indicators provides more comprehensive and in-depth reference parameters for network resilience insights. These multi-dimensional indicators can more sensitively capture subtle changes in system status and provide early warning windows.

[0090] According to the above analysis process, it encapsulates how the temporal evolution of network structural properties and heterogeneity affects its ability to withstand disruptions, providing a comprehensive measure of temporal resilience.

[0091] Step S3: extracting a time series network trend based on the network resilience through singular spectrum analysis, extracting key features based on the time series network trend, wherein the key features include trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and constructing a feature vector based on the key features;

[0092] Step S3-1: Extract resilience trends from time series resilience based on singular spectrum analysis (SSA), and obtain main trend components through singular value decomposition and reconstruction.

[0093] Consider a temporal complex network G = (V, E, T), where V is the set of nodes, E is the set of edges per time step, and T is the total number of time steps. For each time step, the network structure may change, reflecting the dynamic interactions within the system.

[0094] Taking into account the temporal nature, formula (1) can be rewritten as G(t) = (V, E(t)), where t is the time node, t∈{1, 2, ..., T}; by extracting multiple sub-networks and calculating their resilience indicators, the temporal resilience sequence {R k (1),R k (2),...,R k (T)}, k represents the subnetwork number.

[0095] In order to capture the main trend components in the time series resilience, singular spectrum analysis (SSA) is applied to each time series for trend extraction. SSA includes three steps: embedding, singular value decomposition (SVD), and trend reconstruction.

[0096] Specifically, we first choose a window length and construct the embedding matrix:

[0097]

[0098] Among them, X k is the embedding matrix, X k ∈R L×K' ; R k (T) is the resilience value of node k in the time series network at time T; R represents the resilience, k represents the kth time series; L is the number of layers in the time series network, where L ≈ T / 2; and K′ represents the number of nodes. This embodiment ensures the accuracy of trend extraction through an adaptive window length selection mechanism.

[0099] Next, perform Singular Value Decomposition (SVD) on the embedding matrix:

[0100]

[0101] Among them, λ i is a singular value, U i and V i are the corresponding left and right singular vectors respectively.

[0102] Then, the component corresponding to the highest singular value is selected to reconstruct the trend component:

[0103]

[0104] in, is the trend component; m is a parameter, which is usually determined based on the cumulative variance contribution rate or scree plot analysis to ensure that the trend component fully represents the main changes in the time series data.

[0105] By constructing an embedding matrix, singular value decomposition, and trend reconstruction, three key steps are taken to achieve precise separation of long-term trends, cyclical fluctuations, and mutation characteristics. This breaks through the limitations of traditional static network analysis and can accurately reflect the dynamic evolution characteristics of network structure and function.

[0106] Step S3-2: extract key features from trend components to construct feature vectors;

[0107] Extract key features reflecting resilience characteristics from the trend component and construct feature vectors:

[0108]

[0109] Where: x k is a eigenvector, and X k ∈R 4 β k is the trend slope; ROC k is the overall rate of change of the trend component series; C k is the number of trend change points; β 2,k is the mean of the second derivative of the trend series.

[0110] The trend slope is calculated using a linear regression fit to the trend component:

[0111]

[0112] Where t is the time point in the time series; is the average time point of the time series, and T is the total length of the time series; is the trend component of node s at time t; is the mean of the trend component of node s, and

[0113] Compute the overall rate of change of the trend component series:

[0114]

[0115] in, is the trend value at the initial moment, is the trend value at the final moment;

[0116] The number of trend change points is identified using a peak detection algorithm and is defined as the number of local minima in the trend series.

[0117] The second-order trend calculates the mean of the second-order derivative of the trend series, reflecting the acceleration of trend changes:

[0118]

[0119] in, is the trend component at time t;

[0120] In order to eliminate the influence of different feature dimensions and scales, all feature vectors are normalized to obtain standardized feature vectors. Normalization can be expressed as:

[0121]

[0122] Among them, x k′ is the standardized feature vector, μ and σ are the mean and standard deviation vectors of each feature, respectively.

[0123] Step S4: using multiple clustering methods in parallel to perform independent cluster analysis on the feature vectors and perform multi-algorithm clustering;

[0124] Select multiple clustering algorithms, the set of clustering algorithms is A={A1,A2,…,A M}, A j ∈A, apply various clustering algorithms of different properties to perform independent clustering analysis on the standardized feature data, and each clustering algorithm is independently applied to generate cluster labels C j ={c j1 ,c j2 ,…,c jK}, all cluster labels are summarized into a matrix, where C jk Represents the cluster label assigned by the clustering algorithm to the kth time series data. Applying a variety of clustering algorithms with different properties can fully explore the diversity and complexity in the data.

[0125] Step S5: Filter the consistency results of multiple algorithms by the frequency threshold s to identify significant resilience patterns.

[0126] To filter out time series that are consistently classified in multiple clustering results, the same cluster label must be assigned to the same time series data according to the clustering method to set the threshold s, specifically:

[0127] For each standardized feature vector, that is, the time series S k , determine its pattern cluster labels and their frequencies:

[0128]

[0129] in, is the pattern cluster label, is the frequency of label occurrence, I is the ratio calculation, c is the cluster label, j represents different clustering algorithms, j∈{1,2,...,M}, and M is the total number of clustering algorithms used.

[0130] Determine the indicator function based on the frequency of occurrence of pattern cluster labels is the indicator function.

[0131] If r k ≥τ, then S k Marked as significant resilience mode, where τ is the significant resilience threshold and τ∈(0,1);

[0132] P={S k ∣r k ≥τ} (20)

[0133] where P is the set of significant resilience patterns.

[0134] Significant resilience patterns are statistically significant and spatiotemporally stable evolutionary patterns of urban emergency network resilience, identified through multi-algorithm clustering consistency verification and frequency threshold screening. These patterns characterize the dynamic response of urban emergency networks to disturbances, including the evolutionary trajectory of network structure, temporal changes in functional recovery capabilities, and the dynamic patterns of synergy between key nodes. These patterns can quantify the evolution of network resilience under different emergency scenarios, reflecting the complete dynamic process from the onset of disturbance to functional recovery.

[0135] It is understandable that, based on experimental verification, typical significant resilience patterns include: (1) progressive resilience recovery pattern: network functions gradually recover in a steady and increasing trend; (2) transitional resilience reconstruction pattern: the network undergoes structural adjustments in a short period of time to achieve a functional leap; (3) oscillating resilience regulation pattern: network resilience seeks a new equilibrium state in fluctuations; (4) cascading resilience attenuation pattern: local disturbances trigger chain reactions that lead to a decline in system performance.

[0136] This embodiment uses the Sequential Resilience Clustering (SRC) algorithm to identify significant and stable resilience patterns by constructing a temporal complex network, calculating temporal resilience indicators for multiple subnetworks, applying singular spectrum analysis (SSA) to extract trends, extracting and standardizing key features, using multiple clustering algorithms for independent clustering analysis, and selecting consistently classified time series through a mechanism. This approach leverages the advantages of multiple clustering algorithms and enhances the accuracy and robustness of pattern recognition through this mechanism. It is suitable for system analysis and pattern discovery tasks involving resilience data in complex temporal networks.

[0137] Through systematic experimental verification and theoretical analysis, it is demonstrated that the temporal resilience clustering method proposed in this invention exhibits significant technical advantages and innovative effects in urban emergency scenarios.

[0138] This paper innovatively introduces the singular spectrum analysis (SSA) method for trend extraction. Through embedding matrix construction and singular value decomposition, it can adaptively separate the trend, cycle and noise components in time series data. Experiments show that, if Figure 3 As shown, compared to traditional methods such as moving average, SSA trend extraction can improve the signal-to-noise ratio by approximately 40% while maintaining the key signal characteristics, providing a more reliable data foundation for subsequent pattern recognition. This solves the problem that existing technologies, which primarily use fixed time windows or simple sliding windows to process time series data, are unable to effectively capture long-term dependencies during system evolution.

[0139] The present invention constructs a multidimensional resilience indicator system that includes network connectivity, network scale, network density, and network centrality. Experimental verification shows that this multidimensional indicator system can explain more than 85% of system variability, significantly better than the 50-60% explanation rate of existing single indicator methods. Especially when emergencies occur, multidimensional indicators can more keenly capture subtle changes in system status, and the early warning time window can be advanced by 30-50%. This solves the problem that traditional methods often rely on a single or a small number of indicators to evaluate system resilience, making it difficult to fully reflect the dynamic characteristics of the system.

[0140] This paper innovatively proposes a multi-algorithm clustering mechanism, which overcomes the limitations of a single algorithm in different scenarios by combining the advantages of multiple clustering algorithms. Experimental results show that Figure 4-5 As shown, in complex urban emergency scenarios, this mechanism's pattern recognition accuracy can reach 92%, an improvement of 15-20 percentage points over a single clustering algorithm. More importantly, by introducing a consistency screening mechanism, the false alarm rate is significantly reduced from 15-20% with traditional methods to below 5%.

[0141] This paper uses a matrix-based computational approach to achieve efficient data processing. When processing network data containing 10,000 time steps and 100 nodes, the complete analysis process takes only 3-5 minutes, approximately 60% faster than existing methods. Furthermore, thanks to its modular design, the system can be easily expanded to larger datasets, with computational time growing approximately linearly with data size.

[0142] The method presented in this paper has been validated in multiple real-world urban emergency management scenarios. In an emergency response network involving 30 critical infrastructure nodes, it successfully identified four typical resilience evolution patterns and accurately predicted 85% of system performance mutation points, with an average lead time of two hours, providing a sufficient response window for emergency decision-making. These results significantly outperform the 60% prediction accuracy and 45-minute warning time achieved by traditional methods.

[0143] This method demonstrates significant technical advantages in time series feature extraction, multidimensional resilience assessment, pattern recognition accuracy, and system scalability, providing a more scientific and efficient methodology for urban emergency resilience analysis. These benefits are achieved through key innovations in the method, including the integration of SSA trend extraction, a multidimensional resilience indicator system, and a multi-algorithm clustering mechanism, resulting in a complete technical solution.

[0144] Example 2

[0145] In one or more embodiments, a temporal resilience pattern recognition system for urban emergency networks is disclosed, specifically including:

[0146] A data acquisition module is configured to: acquire network flow data of a city emergency network;

[0147] A temporal complex network construction module, configured to: construct a temporal network based on the network flow data;

[0148] A resilience index calculation module is configured to: analyze time series network indicators and calculate network resilience;

[0149] a trend extraction module configured to: extract a time series network trend based on the network resilience through singular spectrum analysis, extract key features based on the time series network trend, the key features including trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and construct a feature vector based on the key features;

[0150] A multi-algorithm clustering module is configured to: use multiple clustering methods to perform independent clustering analysis on feature vectors to obtain multiple clustering results;

[0151] The pattern recognition module is configured to: filter the consistency results of multiple clustering methods through frequency thresholds to identify significant resilience patterns.

[0152] Example 3

[0153] This embodiment provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps of the above-mentioned temporal resilience pattern recognition method for urban emergency networks are completed.

[0154] Example 4

[0155] This embodiment provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps of the above-mentioned temporal resilience pattern recognition method for urban emergency networks are completed.

[0156] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0159] The descriptions of the various embodiments in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0160] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A temporal resilience pattern recognition method for urban emergency networks, characterized by: include: Obtain network flow data for the city’s emergency network; Building a time series network based on the network flow data, analyzing time series network indicators, and calculating network resilience; Extracting a time series network trend based on the network resilience through singular spectrum analysis, extracting key features based on the time series network trend, wherein the key features include trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and constructing a feature vector based on the key features; Multiple clustering methods are used to perform independent cluster analysis on the feature vectors to obtain multiple clustering results; Frequency thresholds were used to screen the consistency results of multiple clustering methods and identify significant resilience patterns.

2. A method for identifying temporal resilience patterns for urban emergency networks according to claim 1, characterized in that: The network resilience expression is: Among them, R t For timing resilience, NV t For network connectivity, NS t is the network size, ND t is the network density, NC t is the network centrality, Gini(k t ) is the Gini coefficient.

3. The method for identifying temporal resilience patterns for urban emergency networks according to claim 1, characterized in that: The temporal network trend is obtained by constructing an embedding matrix, performing singular value decomposition, and reconstructing the trend, specifically: Choose a window length and construct the embedding matrix: Among them, X k is the embedding matrix, R represents the size of resilience, k represents the kth time series, R k represents the resilience value of the kth time series; L is the number of layers in the time series network, and L≈T / 2; K′ is the node; Perform singular value decomposition on the embedded matrix to obtain the corresponding components, including left singular vectors and right singular vectors; Based on the singular value decomposition, the corresponding components are obtained to reconstruct the trend components as the time series network trend: in, is the trend component; m is the parameter.

4. The method for identifying temporal resilience patterns for urban emergency networks according to claim 1, wherein: The key features include: The trend slope is calculated using linear regression fitting of the trend component: Where t is the time point in the time series; is the average time point of the time series, and T is the total length of the time series; is the trend component of node s at time t; is the mean of the trend component of node s, and Compute the overall rate of change of the trend component series: in, is the trend value at the initial moment, is the trend value at the final moment; Calculate the mean of the second derivative of the trend series to reflect the acceleration of trend changes: in, is the trend component at time t.

5. A method for identifying temporal resilience patterns for urban emergency networks according to claim 4, characterized in that: A feature vector is constructed based on the key features, and the feature vector is expressed as: Among them, x k is a eigenvector, and X k ∈R 4 β k is the trend slope; ROC k is the overall rate of change of the trend component series; C k is the number of trend change points; β 2,k is the mean of the second derivative of the trend series.

6. A method for identifying temporal resilience patterns for urban emergency networks according to claim 1, characterized in that: Multiple clustering methods are used to perform independent cluster analysis on the feature vectors to obtain multiple clustering results, specifically: in, is the pattern cluster label, is the frequency of label occurrence, I is the ratio calculation, c is the cluster label, j represents different clustering algorithms, and M is the total number of clustering algorithms.

7. The method for identifying temporal resilience patterns for urban emergency networks according to claim 1, wherein: The method of screening the consistency results of multiple clustering algorithms by frequency threshold and identifying significant resilience patterns is as follows: Filter time series that are consistently classified in multiple clustering results. According to the clustering method, the same time series data must be assigned the same cluster label to set the threshold; For each time series S k , determine its pattern cluster label and its frequency of occurrence, and then determine the indicator function according to the frequency of occurrence of the pattern cluster label: in, represents the frequency of occurrence of pattern clustering labels, M represents the total number of clustering algorithms used, and α is the consistency threshold.

8. A temporal resilience pattern recognition system for urban emergency networks, characterized by: include: A data acquisition module is configured to: acquire network flow data of a city emergency network; A temporal complex network construction module is configured to: construct a temporal network based on the network flow data; A resilience index calculation module is configured to: analyze time series network indicators and calculate network resilience; a trend extraction module configured to: extract a time series network trend based on the network resilience through singular spectrum analysis, extract key features based on the time series network trend, the key features including trend slope, overall rate of change of trend component sequence, number of trend change points, and mean of second-order derivative of trend sequence, and construct a feature vector based on the key features; A multi-algorithm clustering module is configured to: use multiple clustering methods to perform independent clustering analysis on feature vectors to obtain multiple clustering results; The pattern recognition module is configured to: filter the consistency results of multiple clustering methods through frequency thresholds to identify significant resilience patterns.

9. An electronic device, characterized in that: It includes a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are run by the processor, the temporal resilience pattern recognition method for urban emergency networks described in any one of claims 1 to 7 is completed.

10. A computer-readable storage medium, characterized in that Used to store computer instructions, which, when executed by a processor, complete the temporal resilience pattern recognition method for urban emergency networks as described in any one of claims 1 to 7.

Images