Sea-land coupling extreme parallel meteorological recognition disaster chain analysis method

Abstract

The invention provides a sea-land coupling extreme parallel meteorological recognition disaster chain analysis method, and relates to the field of meteorology, and the method comprises the steps: obtaining and preprocessing meteorological and monitoring data of a sea, land and a power transmission channel; based on meteorological and monitoring data, calculating a sea-land energy coupling index through a boundary integral constrained cross-domain energy flow tensor fusion algorithm BI-CETF; based on meteorological and monitoring data, constructing a disaster event triggering network and calculating a link anti-fact improvement degree through a dynamic disaster chain causal reasoning algorithm of a mixed Horkes process and a hypergraph structure; based on meteorological and monitoring data, calculating a parallel synchronization index by adopting a topological directional multiband extreme value synchronization recognition algorithm TD-PESR; fusing, performing CLift and constructing a disaster chain risk matrix, and performing extreme parallel meteorological event identification and risk assessment to obtain a risk distribution diagram or a risk evolution sequence. According to the technical scheme, the limitation of a traditional single-factor early warning or empirical threshold model is broken through.

Description

Technical Field

[0001] This application relates to the field of meteorology, and in particular to a method for identifying disaster chains in extreme parallel meteorological events coupled with land and sea conditions. Background Technology

[0002] In recent years, global climate change has led to frequent extreme weather events. The land-sea interface area, especially coastal power transmission channels, ultra-high voltage lines and energy distribution corridors, is facing an increasingly complex multi-hazard environment. Typhoons, rainstorms, wind shear, thunderstorms, icing, landslides and other disasters often exhibit parallel superposition and cascading triggering characteristics, forming the so-called "disaster chain".

[0003] These types of disaster chains often span two media systems: high-energy convective disturbances on the ocean side (such as typhoons and storm surges) affect localized severe convection, rainstorms, icing, and even secondary geological disasters on the land side through thermal and momentum transfer, causing concentrated damage to power grid transmission lines, tower foundations, and monitoring equipment.

[0004] For online monitoring and risk early warning systems for power transmission lines, traditional meteorological models or single-parameter monitoring cannot reveal the spatiotemporal chain propagation patterns of sea-land energy coupling, making it difficult to accurately identify parallel extreme events and conduct causal analysis of disaster chains. Therefore, an innovative algorithm system is needed that can simultaneously process multi-source heterogeneous meteorological data, reveal cross-medium energy flow transmission mechanisms, and identify the coupling relationships of parallel extreme events.

[0005] China has made some progress in meteorological monitoring and disaster early warning for power transmission lines. For example, the State Grid Corporation of China and the Meteorological Bureau have built meteorological monitoring systems for power transmission lines, icing early warning platforms, and wildfire video recognition systems. However, these systems mostly use single-factor, single-model analysis, for example: Early warning is based on statistical thresholds such as local wind speed and temperature; trend prediction is based on common deep learning methods such as CNN and LSTM; the leading role of ocean-side dynamic processes in land disasters is ignored; and multi-level causal relationship reasoning between disaster events is lacking.

[0006] Insufficient coupling: Existing algorithms cannot achieve unified representation of data from the three domains of ocean, land, and power transmission channels; poor real-time performance: Highly complex models are difficult to deploy at the edge and cannot support online identification; lack of causality: Most algorithms remain at the level of statistical correlation and cannot identify disaster chains with multiple causes and common causes; insufficient robustness: The accuracy of identifying parallel extreme events (typhoon + rainstorm + icing) is low, and there is a lack of directional constraints and physical consistency. Summary of the Invention

[0007] The purpose of this invention is to provide a method for analyzing disaster chains in marine-terrestrial coupled extreme parallel meteorological events, in order to solve the problems of fragmented modeling of marine and terrestrial meteorological systems, insufficient single-element disaster identification capabilities, unclear disaster chain triggering relationships, and low accuracy of parallel extreme event identification in the existing technology.

[0008] The above-mentioned objective of this application is achieved through the following technical solution: S1: Acquire meteorological and monitoring data from the ocean, land, and power transmission channels and perform preprocessing; S2: Based on meteorological and monitoring data, the land-sea energy coupling index is calculated using the BI-CETF cross-domain energy flow tensor fusion algorithm with boundary integral constraints. ; S3: A dynamic disaster chain causal reasoning algorithm based on meteorological and monitoring data, using a hybrid Hawkes process and hypergraph structure. Construct a disaster event triggering network and calculate the link counterfactual lift. ; S4: Based on meteorological and monitoring data, the Topology-Oriented Multi-Band Extremum Synchronization Recognition Algorithm (TD-PESR) is used to calculate the parallel synchronization index. ; S5: Fusion , CLIFT and Construct a disaster chain risk matrix, identify and assess the risks of extreme parallel meteorological events, and obtain risk distribution maps or risk evolution sequences.

[0009] Optionally, step S1 includes: Meteorological and monitoring data include: sea surface temperature, sea surface wind speed, air pressure, land temperature and humidity and rainfall, electric field strength, conductor vibration amplitude and partial discharge signals; Preprocessing includes: unified timestamps and spatial standardization.

[0010] Optionally, step S2 includes: S21: The coupling strength of land-sea energy is measured through meteorological and monitoring data, as detailed below: For the ocean side, the energy flux density consists of two parts: thermal flux and kinetic flux, expressed as:

[0011] in, Indicates the energy flux density on the ocean side; Indicates the density of seawater; This indicates the specific heat capacity of seawater; Indicates sea surface temperature; Represents the sea surface current velocity vector; These represent longitude, latitude coordinates, and time variables, respectively. For the land side, energy flux density includes thermal flux and kinetic flux, defined as follows:

[0012] in, This represents the energy flow density on the land side; Indicates air density; This indicates the specific heat capacity of air at constant pressure. Indicates land surface temperature; Represents the land wind speed vector; S22: Define the energy flux conservation relationship on the land-sea boundary curve to construct the energy transfer process between the ocean and land. Let the boundary normal vector be... The energy flux across the boundary is then expressed by the following integral:

[0013] in, Indicates at time Along the boundary Net energy flux; Indicates the density of the medium; Represents the land-sea boundary curve; Indicates specific heat capacity; Represents the temperature field; Represents the velocity field; A boundary flux constraint term is introduced, as follows:

[0014] in, This indicates the energy flux output from the ocean side; This represents the energy flux input from the land side; Indicates the energy difference threshold; Let the ocean characteristic tensor be represented as The land feature tensor is represented as The two obtain the latent factor matrix through tensor decomposition; The optimization objective of the cross-domain energy flow tensor fusion algorithm can be expressed as:

[0015] in, The latent factor matrix representing the ocean and land feature tensors; and Represents the measured ocean tensor and the reconstructed tensor; and Represents the measured tensor and the reconstructed tensor of the land; and Represents the regularization coefficient; Represents a distance metric based on optimal transmission; and Indicates the distribution of marine and terrestrial features; The optimization process of the cross-domain energy flow tensor fusion algorithm achieves rapid convergence through the Sinkhorn iterative algorithm; Introducing the Helmholtz–Hodge decomposition, the wind speed vector field is... It can be decomposed into two parts: irrotational potential flow and vortex flow.

[0016] in, Represents the wind speed vector field; Represents the irrotational potential flow component; Represents the gradient operator; Represents the vector potential function; In the output phase of the fusion results, the land-sea energy coupling index is defined. + is used to quantify the degree of energy coherence and transmission intensity between land and sea, and its calculation formula is:

[0017] in, Represents the natural exponential function; Indicates the activation function; Indicates temperature; The infinitesimal arc length representing the boundary curve; The larger the value, the stronger the energy transfer between land and sea and the higher the degree of coupling.

[0018] Optionally, step S3 includes: During the observation period Internal identification Set of disaster-related event nodes Each node This indicates a specific type of meteorological or power transmission monitoring event. ; Meteorological or power transmission monitoring events include: typhoon landfall. 1. Rainstorm intensifies 2. Ice accumulation 3. Conductor vibration amplification 4 and partial discharge 5; Each event node has an attribute feature vector, including: wind speed, temperature and humidity, electric field strength and vibration energy spectrum features; Constructing a hypergraph structure for disaster chains =( , ),in Let be the set of superedges, each superedge This indicates a causal relationship triggered by multiple parent events. In mathematical modeling, a labeled multidimensional Hawkes process is used to describe the event triggering intensity, and target event nodes are defined. The conditional strength function is:

[0019] in Represents the target event At any moment Trigger strength; Indicates base strength; Represents the time decay kernel function; Represents the time constant; Indicates the function that triggers the effect; Represents the feature vector of the parent event; Represents the differentiation of the event counting process; Using the land-sea energy coupling index as a priori constraint for the hyperedge weights, the hyperedge weight function is defined as follows:

[0020] in Indicates the superedge At any moment The weights; , and Indicates the weighting coefficient; Indicates time The energy coupling index; This represents the correlation coefficient between features of parent events. Indicates the first Feature vectors of each parent event; Represents the natural exponential function; Indicates the spatial attenuation coefficient; Indicates the spatial distance between the parent event and the child event; A time attention mechanism is introduced to adaptively adjust the influence of the parent event at different time intervals. The time attention weight is defined as follows:

[0021] in Indicates time Lower parent event Pair Events Attention weights; Represents the query vector; Represents the key vector; Indicates the dimension of a vector; Indicates sub-events The set of parent events; Represents the key vector; Design a soft intervention mechanism to estimate the counterfactual lift of a set of parent events to the target event. Its definition is as follows:

[0022] in and Indicates the start and end times of the observation; Indicates the original trigger strength; Indicates the triggering intensity after soft intervention; This represents the smoothing constant.

[0023] Optionally, step S4 includes: Meteorological and monitoring data are divided into categories after bandpass filtering. Characteristic frequency bands To simultaneously capture high-frequency local disturbances and low-frequency regional variations; For each frequency band The instantaneous phase of the signal is obtained using the Hilbert transform. ; In phase synchronization analysis, an extreme value weighting mechanism is introduced. The weighted multi-source phase synchronization coefficient is defined as follows:

[0024] in Indicates frequency band The weighted synchronization coefficient is below; Indicates the total number of signals; Represents the phase of a complex number; This represents the normalized extremum weight function for each signal; Introducing directional propagation constraints by defining the main propagation direction vector in geographic coordinates. Construct an anisotropic phase diffusion operator Its mathematical expression is:

[0025] in Represents the directional loss function; Indicates a spatial region; This represents the gradient operator along the main propagation direction; Indicates frequency band The phase field below; Represents the anisotropy coefficient; This represents the gradient operator perpendicular to the main propagation direction; Topological persistent cohomology analysis is introduced, the phase similarity matrix is ​​thresholded, and the length of the persistent stripe of its cohomology group as a function of the threshold is calculated. ; Considering extreme value weighting, directional constraints, and topological stability, a comprehensive parallel synchronization index is defined. :

[0026] in Indicates the total number of frequency bands; Indicates the direction-weighted synchronization coefficient; Indicates the frequency band weighting coefficient; when When the threshold is exceeded, it can be determined that multiple disasters are occurring in parallel extreme events.

[0027] Optionally, step S5 includes: Energy Coupling Index + Causal triggering strength Parallel synchronization metrics The risk values ​​of the units in the disaster chain risk matrix are formed by weighted superposition and defined as follows:

[0028] in Represents longitude, latitude coordinates, and time variables; , and This represents the risk weighting coefficient.

[0029] An electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform a land-sea coupled extreme parallel meteorological disaster chain analysis method.

[0030] A computer-readable storage medium storing instructions that, when executed, perform a land-sea coupled extreme parallel meteorological disaster chain identification analysis method.

[0031] The beneficial effects of the technical solution provided in this application are: The "Sea-Land Coupled Extreme Parallel Meteorological Disaster Chain Identification Method" proposed in this invention, through the construction of a cross-domain energy flow tensor fusion model (BI-CETF), a dynamic disaster causal hypergraph model (H³-DDCG), and a topology-oriented extreme value synchronization identification model (TD-PESR), can reveal the mechanism of sea-land meteorological interaction from the perspective of energy conservation and physical boundaries; capture the parallel triggering relationships of various extreme events (typhoons, rainstorms, icing, wildfires, etc.); output the spatiotemporal evolution path and key control factors of disaster chains; and form a quantitative early warning indicator system applicable to power grids, disaster prevention, meteorological emergency response, and other fields. This method breaks through the limitations of traditional single-factor early warning or empirical threshold models, providing a new technical approach for disaster prevention and mitigation in power transmission channels, sea-land coupled meteorological monitoring, and smart grid safety assessment, and is of great significance for improving the meteorological resilience of my country's coastal energy corridors and reducing power outages and equipment damage. Attached Figure Description

[0032] The present application will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a step diagram of an embodiment of this application; Figure 2 This is the login window ECI in the embodiments of this application. + Plot showing uplift and boundary flux changes; Figure 3 This is a diagram showing the inverse relationship between OT distance and energy intrusion in the embodiments of this application; Figure 4 This is a diagram illustrating the impact of soft intervention on the CLIFT and trigger rate of the icing / vibration link in the embodiments of this application. Figure 5 This is the TD-PESR parallel synchronization index diagram in the embodiments of this application; Figure 6 This is the spatiotemporal risk matrix and nearshore high-risk zone map in the embodiments of this application; Figure 7 This is the normalized overlay consistency map in the embodiments of this application; Figure 8 This is a schematic diagram of the electronic device structure in the embodiments of this application. Detailed Implementation

[0033] To provide a clearer understanding of the technical features, objectives, and effects of this application, the specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0034] The embodiments of this application provide a method for identifying disaster chains in extreme parallel meteorological events coupled with land and sea conditions.

[0035] Please refer to Figure 1 , Figure 1This is a flowchart illustrating the steps of a land-sea coupled extreme parallel meteorological disaster chain identification method in an embodiment of this application, including: S1: Acquire meteorological and monitoring data from the ocean, land, and power transmission channels and perform preprocessing; S2: Based on meteorological and monitoring data, the land-sea energy coupling index is calculated using the BI-CETF cross-domain energy flow tensor fusion algorithm with boundary integral constraints. ; S3: A dynamic disaster chain causal reasoning algorithm based on meteorological and monitoring data, using a hybrid Hawkes process and hypergraph structure. Construct a disaster event triggering network and calculate the link counterfactual lift. ; S4: Based on meteorological and monitoring data, the Topology-Oriented Multi-Band Extremum Synchronization Recognition Algorithm (TD-PESR) is used to calculate the parallel synchronization index. ; S5: Fusion , CLIFT and Construct a disaster chain risk matrix, identify and assess the risks of extreme parallel meteorological events, and obtain risk distribution maps or risk evolution sequences.

[0036] This application employs the above-mentioned technical solutions to construct a boundary integral-constrained energy flow tensor model to quantify energy exchange between land and sea; designs a Hawkes-hypergraph structure-based dynamic reasoning model for disaster chains to identify multi-parent common-cause triggering mechanisms; and introduces a topology-oriented extreme value synchronization analysis method to characterize the co-evolution of parallel extreme weather events; thereby achieving accurate identification, causal tracking, and chain-based early warning of land-sea coupled parallel meteorological events.

[0037] Step S1 includes: Meteorological and monitoring data include: sea surface temperature, sea surface wind speed, air pressure, land temperature and humidity and rainfall, electric field strength, conductor vibration amplitude and partial discharge signals; Preprocessing includes: unified timestamps and spatial standardization.

[0038] As one example, at the data acquisition level, it is first necessary to standardize the time and space of multi-source observation data for both ocean and land. Assuming that within a given time period, there exist ocean datasets and land observation datasets, both of which are four-dimensional tensor structures, all data are acquired and registered in real time via an edge computing gateway to ensure the comparability of ocean and land data at the same timestamp.

[0039] As one example, to measure the coupling strength of land and sea energy from a physical perspective, the energy flux density per unit area is first defined.

[0040] Step S2 includes: S21: The coupling strength of land-sea energy is measured through meteorological and monitoring data, as detailed below: For the ocean side, the energy flux density consists of two parts: thermal flux and kinetic flux, expressed as:

[0041] in, Indicates the energy flux density on the ocean side; Indicates the density of seawater; This indicates the specific heat capacity of seawater; Indicates sea surface temperature; Represents the sea surface current velocity vector; These represent longitude, latitude coordinates, and time variables, respectively. For the land side, energy flux density includes thermal flux and kinetic flux, defined as follows:

[0042] in, This represents the energy flow density on the land side; Indicates air density; This indicates the specific heat capacity of air at constant pressure. Indicates land surface temperature; Represents the land wind speed vector; As one example, in order to reveal the energy transfer process between the ocean and the land, it is necessary to define the energy flux conservation relationship on the ocean-land boundary curve.

[0043] S22: Define the energy flux conservation relationship on the land-sea boundary curve to construct the energy transfer process between the ocean and land. Let the boundary normal vector be... The energy flux across the boundary is then expressed by the following integral:

[0044] in, Indicates at time Along the boundary Net energy flux; Indicates the density of the medium; Represents the land-sea boundary curve; Indicates specific heat capacity; Represents the temperature field; Represents the velocity field; As one embodiment, in order to ensure energy conservation, the present invention introduces a boundary flux constraint term in the tensor fusion process, so that the energy flux output from the ocean side is statistically equal to the energy flux input from the land side.

[0045] A boundary flux constraint term is introduced, as follows:

[0046] in, This indicates the energy flux output from the ocean side; This represents the energy flux input from the land side; Indicates the energy difference threshold; As one example, the energy difference threshold (in W / m²) can be set based on experience or historical climate data. This constraint ensures that the model focuses not only on the accuracy of data fitting during optimization but also on maintaining the consistency of physical energy conservation.

[0047] Let the ocean characteristic tensor be represented as The land feature tensor is represented as The two obtain the latent factor matrix through tensor decomposition; The optimization objective of the cross-domain energy flow tensor fusion algorithm can be expressed as:

[0048] in, The latent factor matrix representing the ocean and land feature tensors; and Represents the measured ocean tensor and the reconstructed tensor; and Represents the measured tensor and the reconstructed tensor of the land; and Represents the regularization coefficient; Represents a distance metric based on optimal transmission; and Indicates the distribution of marine and terrestrial features; The optimization process of the cross-domain energy flow tensor fusion algorithm achieves rapid convergence through the Sinkhorn iterative algorithm; As one embodiment, in order to further suppress spurious coupling caused by vortex terms in the wind field, this invention introduces Helmholtz–Hodge decomposition to transform the wind speed vector field. It is decomposed into two parts: irrotational potential flow and vortex flow. Only the irrotational part is included in the boundary energy flux calculation to eliminate spurious energy transfer caused by vortex disturbances.

[0049] Introducing the Helmholtz–Hodge decomposition, the wind speed vector field is... It can be decomposed into two parts: irrotational potential flow and vortex flow.

[0050] in, Represents the wind speed vector field; Represents the irrotational potential flow component; Represents the gradient operator; Represents the vector potential function; In the output phase of the fusion results, the land-sea energy coupling index is defined. + is used to quantify the degree of energy coherence and transmission intensity between land and sea, and its calculation formula is:

[0051] in, Represents the natural exponential function; Indicates the activation function; Indicates temperature; The infinitesimal arc length representing the boundary curve; The larger the value, the stronger the energy transfer between land and sea and the higher the degree of coupling.

[0052] As one embodiment, the BI-CETF algorithm of this invention adopts an "edge-cloud collaborative" model. Edge devices are responsible for low-order feature extraction and local tensor updates (such as calculating local wind field gradients, temperature deviations, and air pressure disturbances), while the cloud platform is responsible for high-dimensional tensor fusion, optimal transmission optimization, and boundary flux integration calculation. This reduces the computational burden on the edge devices while ensuring the model's real-time performance and high accuracy. The energy coupling index output by the algorithm... + can be directly input into subsequent disaster chain causal reasoning algorithms to characterize the potential impact intensity of ocean disturbances on terrestrial disasters.

[0053] As one example, the BI-CETF algorithm achieves cross-media energy flow fusion modeling of multi-source heterogeneous meteorological data. It not only maintains the physical energy conservation relationship but also ensures the consistency of land-sea feature spaces through optimal transmission metrics. This algorithm provides a solid physical and mathematical foundation for subsequent disaster chain reasoning and synchronous identification of extreme events, and is the first core supporting algorithm of the entire invention system.

[0054] Step S3 includes: As one example, after completing the fusion of ocean-land energy flows and the calculation of the energy coupling index, the system is able to obtain physical quantitative indicators characterizing the intensity of energy transfer between the ocean and land. However, a description of the energy field alone is insufficient to reveal the propagation mechanism and triggering chain of disasters. Extreme weather disasters are often not linear responses to a single factor, but rather the result of the spatiotemporal interaction and common-cause triggering of multiple environmental variables. To address this, this invention proposes a Hybrid-Hawkes Hypergraph-based Dynamic Disaster CausalityGraph (H³-DDCG) algorithm, which aims to identify the causal relationships, triggering intensities, and propagation paths between multi-source meteorological events and power transmission monitoring events, thereby constructing a complete spatiotemporal inference network for the "disaster chain" based on the results of energy coupling analysis.

[0055] As one embodiment, the overall idea of ​​this algorithm is to abstract various meteorological and monitoring events into time-series nodes, characterize the causal relationships between events using a hypergraph structure with a multi-parent triggering mechanism, then combine a Hawkes process with a time decay kernel to quantitatively estimate the triggering intensity, and introduce a soft intervention (Hybrid-Do) mechanism to evaluate the counterfactual contribution of each triggering path. In this way, the system can comprehensively depict the chain propagation pattern of extreme events from temporal, spatial, and causal perspectives. Unlike traditional Bayesian networks or static correlation models, this invention... The model possesses time dynamics, multiple parent triggering, and causal intervention capability, and can reflect the parallel triggering and cascading amplification characteristics of disaster events in real-world environments.

[0056] During the observation period Internal identification Set of disaster-related event nodes Each node This indicates a specific type of meteorological or power transmission monitoring event. ; Meteorological or power transmission monitoring events include: typhoon landfall. 1. Rainstorm intensifies 2. Ice accumulation 3. Conductor vibration amplification 4 and partial discharge 5; Each event node has an attribute feature vector, including: wind speed, temperature and humidity, electric field strength and vibration energy spectrum features; Constructing a hypergraph structure for disaster chains =( , ),in Let be the set of superedges, each superedge This indicates a causal relationship triggered by multiple parent events. As one example, a hypergraph structure for disaster chains is constructed to represent the situation where multiple parent events may jointly trigger a child event.

[0057] In mathematical modeling, a labeled multidimensional Hawkes process is used to describe the event triggering intensity, and target event nodes are defined. The conditional strength function is:

[0058] in Represents the target event At any moment Trigger strength; Indicates base strength; Represents the time decay kernel function; Represents the time constant; Indicates the function that triggers the effect; Represents the feature vector of the parent event; Represents the differentiation of the event counting process; As one embodiment, to ensure the physical rationality of the model, this invention uses the energy coupling index as a priori constraint for the hyperedge weights during feature fusion, making events with a strong background of land-sea energy transfer more easily identified as high-probability triggering relationships. This weighting function unifies physical energy connections, statistical correlations, and spatial propagation constraints within the same causal reasoning framework.

[0059] Using the land-sea energy coupling index as a priori constraint for the hyperedge weights, the hyperedge weight function is defined as follows:

[0060] in Indicates the superedge At any moment The weights; , and Indicates the weighting coefficient; Indicates time The energy coupling index; This represents the correlation coefficient between features of parent events. Indicates the first Feature vectors of each parent event; Represents the natural exponential function; Indicates the spatial attenuation coefficient; Indicates the spatial distance between the parent event and the child event; As one example, at the temporal modeling level, to capture the dynamics and memory of disaster events, a Temporal Graph Attention (T-GAT) mechanism is introduced to adaptively adjust the influence of parent events at different time intervals. This attention mechanism achieves adaptive learning of the causal strength of time-varying disaster chains by dynamically adjusting the influence weights of different parent events.

[0061] A time attention mechanism is introduced to adaptively adjust the influence of the parent event at different time intervals. The time attention weight is defined as follows:

[0062] in Indicates time Lower parent event Pair Events Attention weights; Represents the query vector; Represents the key vector; Indicates the dimension of a vector; Indicates sub-events The set of parent events; Represents the key vector; As one embodiment, to improve the interpretability of the model at the causal reasoning level, this invention further designs a "soft intervention" (Hybrid-Do) mechanism to estimate the counterfactual contribution of a set of parent events to the target event. This mechanism, without disrupting the overall causal network structure, calculates the change in the target event trigger rate by perturbing the amplitude or threshold of specific parent node features, thereby obtaining the "CLOFT" (Counterfactual Lift).

[0063] Design a soft intervention mechanism to estimate the counterfactual lift of a set of parent events to the target event. Its definition is as follows:

[0064] in and Indicates the start and end times of the observation; Indicates the original trigger strength; Indicates the triggering intensity after soft intervention; This represents the smoothing constant.

[0065] As one example, in actual implementation, The algorithm iteratively updates the hyperedge weights and temporal attention coefficients to achieve adaptive optimization of the disaster chain structure. The resulting dynamic causal network not only outputs a disaster chain topology map but also identifies the "Minimal Hitting Set" (MHS), which is the minimum set of intervention nodes required to interrupt the spread of a disaster. This characteristic allows the model to not only be used for disaster identification but also to guide the formulation of prevention and control strategies, such as activating icing and de-icing devices in advance, adjusting line loads, or pre-setting cross-section protection strategies, thereby achieving proactive disaster prevention.

[0066] From the perspective of its connection with the previous algorithm (BI-CETF), The model will output the energy coupling index from the former. The energy prior, used as a causal graph construction term, provides a physical basis for disaster chain reasoning, avoiding the spurious correlation problem that may arise from purely statistical causal analysis. Through energy-causal coupling, the model achieves a logical extension from the "physical process layer" to the "event triggering layer," enabling energy transfer and disaster evolution to be uniformly characterized within the same reasoning framework.

[0067] In summary, The algorithm occupies a pivotal position in this invention, bridging the gap between previous and subsequent stages. It transforms the land-sea energy flow relationship revealed by the BI-CETF algorithm into a multi-event triggering probability distribution, providing causal constraints and triggering logic for the next step of synchronous identification of parallel extreme events. Through this algorithm, the system can perform structured modeling, quantitative analysis, and strategic intervention on complex, multi-factor disaster chains, achieving a leap from passive monitoring to proactive early warning.

[0068] Step S4 includes: As an example, after completing the aforementioned sea-land energy coupling modeling (BI-CETF) and dynamic disaster chain causal reasoning, the system has obtained quantitative indicators that reflect the relationship between energy transfer intensity and disaster triggering. However, extreme meteorological disasters in the real environment often exhibit the characteristics of multiple disasters occurring in parallel and in synergy. For example, typhoon landfall may be accompanied by torrential rain, electric field jumps, and conductor galloping; mountain torrential rain and strong convection may trigger icing and partial discharge in parallel. To reveal the synergistic characteristics of these parallel extreme events in terms of temporal phase and spatial distribution, this invention proposes a Topology-Directed Parallel Extreme Synchrony Recognition and Risk Assessment Algorithm (TD-PESR). The core idea of ​​this algorithm is to quantitatively characterize the degree of synchronization and stability between multiple meteorological events through multi-band phase synchronization analysis, directional propagation constraints, extreme segment weighting, and topological coherence feature extraction, and transform it into a comprehensive risk indicator, thereby completing the spatiotemporal synergistic identification of disaster chains within an energy-causal framework. The overall goal of the algorithm is to measure the synergistic characteristics of multiple meteorological events in time series. and Whether there is strong synchronization behavior between them, and determine whether the synchronization has directionality, extreme value correlation and topological stability.

[0069] Meteorological and monitoring data are divided into categories after bandpass filtering. Characteristic frequency bands To simultaneously capture high-frequency local disturbances and low-frequency regional variations; For each frequency band The instantaneous phase of the signal is obtained using the Hilbert transform. ; As one embodiment, to improve the algorithm's sensitivity to extreme events, this invention introduces an extreme value weighting mechanism in the phase synchronization analysis. Let the normalized extreme value weighting function for each signal be... When the observed value When the value exceeds its 95th percentile threshold, the weight is significantly increased to emphasize the contribution of extreme segments to the synchronization index.

[0070] In phase synchronization analysis, an extreme value weighting mechanism is introduced. The weighted multi-source phase synchronization coefficient is defined as follows:

[0071] in Indicates frequency band The weighted synchronization coefficient is below; Indicates the total number of signals; Represents the phase of a complex number; This represents the normalized extremum weight function for each signal; As one example, since meteorological processes in the land-sea coupling zone exhibit significant propagation directionality, such as energy intrusion from the ocean to the land or the landward expansion of storms, simple phase consistency cannot reflect the spatial evolution of events. Therefore, this invention introduces a directional propagation constraint term. By defining the main propagation direction vector in geographic coordinates, an anisotropic phase diffusion operator is constructed to suppress phase alignment that deviates excessively from the main direction.

[0072] Introducing directional propagation constraints by defining the main propagation direction vector in geographic coordinates. Construct an anisotropic phase diffusion operator Its mathematical expression is:

[0073] in Represents the directional loss function; Indicates a spatial region; This represents the gradient operator along the main propagation direction; Indicates frequency band The phase field below; Represents the anisotropy coefficient; This represents the gradient operator perpendicular to the main propagation direction; As one embodiment, the model emphasizes phase consistency along the main direction, forming a synchronous cone structure with propagation directionality. After obtaining the directional synchronous distribution, to eliminate random noise and transient resonances at the topological level, this invention introduces topological persistent homology (PH) analysis, thresholding the phase similarity matrix and calculating the length of the persistent stripe of its homology group as the threshold changes. This value reflects the duration of different phase clusters at multiple scale thresholds, thus measuring the topological stability of the synchronization structure.

[0074] Topological persistent cohomology analysis is introduced, the phase similarity matrix is ​​thresholded, and the length of the persistent stripe of its cohomology group as a function of the threshold is calculated. ; Considering extreme value weighting, directional constraints, and topological stability, a comprehensive parallel synchronization index is defined. :

[0075] in Indicates the total number of frequency bands; Indicates the direction-weighted synchronization coefficient; Indicates the frequency band weighting coefficient; when When the threshold is exceeded, it can be determined that multiple disasters are occurring in parallel extreme events.

[0076] Step S5 includes: As one embodiment, in order to achieve a unified risk output of multiple algorithm results, this invention establishes a comprehensive risk function in the fusion inference layer.

[0077] Energy Coupling Index + Causal triggering strength Parallel synchronization metrics The risk values ​​of the units in the disaster chain risk matrix are formed by weighted superposition and defined as follows:

[0078] in Represents longitude, latitude coordinates, and time variables; , and This represents the risk weighting coefficient.

[0079] As one example, the risk matrix The output can form both a spatial risk distribution map and a temporal risk evolution sequence, providing quantitative basis for real-time disaster early warning and prevention decisions. In engineering implementation, the TD-PESR algorithm is deployed in a collaborative system of a cloud-based main control platform and edge intelligent terminals. The edge terminal performs preprocessing such as signal sampling, band filtering, and Hilbert transform, while the platform side is responsible for directional constraint optimization, topology coherence calculation, and comprehensive risk assessment. The algorithm output... The indicators and risk matrix can be directly connected to the transmission line operation status monitoring platform to achieve three-dimensional visualization and automatic early warning push. When the system detects a high-risk parallel extreme event chain, it can trigger preset prevention and control strategies, such as line current limiting, de-icing heating, and drone inspection and dispatch, to achieve dynamic closed-loop response to the disaster chain.

[0080] From a system integration perspective, the TD-PESR algorithm occupies the final stage of the "identification-assessment-decision" process in the entire invention methodology. It uses the results of the first two steps as input, transforming energy coupling and causal relationships into specific parallel extreme event identification and risk level output, forming a complete physical-causal-topological triple fusion framework. This algorithm represents a leap from single-hazard analysis to multi-hazard parallel identification, enabling the system to perform comprehensive spatiotemporal prediction and graded early warning of complex disaster chains in a land-sea coupled context, thereby significantly improving the operational safety and resilience of the power grid and critical infrastructure under extreme weather conditions.

[0081] In one embodiment, the coastline is simplified to a straight line. =0, left side <0 represents the sea area, on the right. >0 represents land area. The time discretization is 49 hours (0–48h), with the warm tongue of the sea and the sea breeze moving at a constant phase velocity. =2km / h advancing landward, resulting in ≈20 A significant transboundary energy intrusion from sea to land was observed over 32 hours; we used a 1D spatial profile (along the normal) to characterize the energy distribution and delineate the boundary... =0 is used to approximate the flux integral position. BI-CETF section: using the thermal-kinetic approximation. = Construct ocean / land energy tensor slices, approximate the optimal transmission distance using one-dimensional EMD (Wasserstein-1), and superimpose boundary fluxes to obtain... +. H³-DDCG part: Synthesizes four types of events (sudden wind surge W, heavy rainfall R, icing I, vibration V), whose triggering intensity follows a simplified multi-parent Hawkes structure. "Soft intervention" involves downsizing the weight of a parent chain, followed by integration and difference to obtain the CLIFT. TD-PESR part: Bandpass + Hilbert phase is applied to the wind speed, electric field, and vibration signals. Extreme value segments are weighted, and directional weights (sea → land) are used to scan the "pseudo-persistent" duration using a threshold, forming... A workable approximation is obtained. Finally, a weighted average is used to form a risk matrix and an overview.

[0082] In another embodiment, Figure 2 This displays the time-varying flux at the sea / landside boundary and the ECI composed of OT distance and flux. + ,exist ≈20 Within the 32-hour login window, ECI +The significant increase indicates enhanced cross-boundary energy coupling; Figure 3 The OT distance of the one-dimensional sea / land distribution is given, which changes in the opposite direction to the energy intrusion, further supporting the evidence that the energy distribution in the sea area is gradually aligning towards the land side; Figure 4 The left side shows that the CLIFT values ​​for both the icing (I) and vibration (V) links were positive and significant before and after soft intervention. The trigger rate curve on the right side also shows that intervention can suppress the trigger rate. , The area; Figure 5 The TD-PESR parallel synchronization index is given, which significantly improves direction-weighted synchronization within the landing window. Combined with the "pseudo-persistence" factor, a stable synchronization is obtained. peak; Figure 6 Risk matrix ( , A high-risk zone emerged in the nearshore 0–40 km area, where the landing window overlapped with the landing time. Figure 7 The overview map shows , and After normalization and superposition, the three are highly consistent during the critical period, indicating that the closed-loop logic of energy-causality-synchronization is self-consistent and verifiable in this embodiment.

[0083] This application also discloses an electronic device. (See reference...) Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 500 may include: at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0084] The communication bus 502 is used to enable communication between these components.

[0085] The user interface 503 may include a display screen, and optionally, the user interface 503 may also include a standard wired interface or a wireless interface.

[0086] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0087] This application also discloses a computer-readable storage medium storing multiple instructions adapted for loading by a processor to execute the aforementioned method for analyzing disaster chains in a combined land-sea extreme meteorological environment.

[0088] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure.

[0089] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for identifying disaster chains in a combined land-sea extreme parallel meteorological event, characterized in that, The method includes the following steps: S1: Acquire meteorological and monitoring data from the ocean, land, and power transmission channels and perform preprocessing; S2: Based on meteorological and monitoring data, the land-sea energy coupling index is calculated using the BI-CETF cross-domain energy flow tensor fusion algorithm with boundary integral constraints. ; S3: A dynamic disaster chain causal reasoning algorithm based on meteorological and monitoring data, using a hybrid Hawkes process and hypergraph structure. Construct a disaster event triggering network and calculate the link counterfactual lift. ; S4: Based on meteorological and monitoring data, the Topology-Oriented Multi-Band Extremum Synchronization Recognition Algorithm (TD-PESR) is used to calculate the parallel synchronization index. ; S5: Fusion , CLIFT and Construct a disaster chain risk matrix, identify and assess the risks of extreme parallel meteorological events, and obtain risk distribution maps or risk evolution sequences.

2. The method for identifying disaster chains in a combined land-sea extreme parallel meteorological event as described in claim 1, characterized in that, Step S1 includes: Meteorological and monitoring data include: sea surface temperature, sea surface wind speed, air pressure, land temperature and humidity and rainfall, electric field strength, conductor vibration amplitude and partial discharge signals; Preprocessing includes: unified timestamps and spatial standardization.

3. The method for identifying disaster chains in a combined land-sea extreme parallel meteorological event as described in claim 1, characterized in that, Step S2 includes: S21: The coupling strength of land-sea energy is measured through meteorological and monitoring data, as detailed below: For the ocean side, the energy flux density consists of two parts: thermal flux and kinetic flux, expressed as: in, Indicates the energy flux density on the ocean side; Indicates the density of seawater; This indicates the specific heat capacity of seawater; Indicates sea surface temperature; Represents the sea surface current velocity vector; These represent longitude, latitude coordinates, and time variables, respectively. For the land side, energy flux density includes thermal flux and kinetic flux, defined as follows: in, This represents the energy flow density on the land side; Indicates air density; This indicates the specific heat capacity of air at constant pressure. Indicates land surface temperature; Represents the land wind speed vector; S22: Define the energy flux conservation relationship on the land-sea boundary curve to construct the energy transfer process between the ocean and land. Let the boundary normal vector be... The energy flux across the boundary is then expressed by the following integral: in, Indicates at time Along the boundary Net energy flux; Indicates the density of the medium; Represents the land-sea boundary curve; Indicates specific heat capacity; Represents the temperature field; Represents the velocity field; A boundary flux constraint term is introduced, as follows: in, This indicates the energy flux output from the ocean side; This represents the energy flux input from the land side; Indicates the energy difference threshold; Let the ocean characteristic tensor be represented as The land feature tensor is represented as The two obtain the latent factor matrix through tensor decomposition; The optimization objective of the cross-domain energy flow tensor fusion algorithm can be expressed as: in, The latent factor matrix representing the ocean and land feature tensors; and Represents the measured ocean tensor and the reconstructed tensor; and Represents the measured tensor and the reconstructed tensor of the land; and Represents the regularization coefficient; Represents a distance metric based on optimal transmission; and Indicates the distribution of marine and terrestrial features; The optimization process of the cross-domain energy flow tensor fusion algorithm achieves rapid convergence through the Sinkhorn iterative algorithm; Introducing the Helmholtz–Hodge decomposition, the wind speed vector field is... It can be decomposed into two parts: irrotational potential flow and vortex flow. in, Represents the wind speed vector field; Represents the irrotational potential flow component; Represents the gradient operator; Represents the vector potential function; In the output phase of the fusion results, the land-sea energy coupling index is defined. + is used to quantify the degree of energy coherence and transmission intensity between land and sea, and its calculation formula is: in, Represents the natural exponential function; Indicates the activation function; Indicates temperature; The infinitesimal arc length representing the boundary curve; The larger the value, the stronger the energy transfer between land and sea and the higher the degree of coupling.

4. The method for identifying disaster chains in a combined land-sea extreme parallel meteorological event as described in claim 1, characterized in that, Step S3 includes: During the observation period Internal identification Set of disaster-related event nodes Each node This indicates a specific type of meteorological or power transmission monitoring event. ; Meteorological or power transmission monitoring events include: typhoon landfall.

1. Rainstorm intensifies 2. Ice accumulation 3. Conductor vibration amplification 4 and partial discharge 5; Each event node has an attribute feature vector, including: wind speed, temperature and humidity, electric field strength and vibration energy spectrum features; Constructing a hypergraph structure for disaster chains =( , ),in Let be the set of superedges, each superedge This indicates a causal relationship triggered by multiple parent events. In mathematical modeling, a labeled multidimensional Hawkes process is used to describe the event triggering intensity, and target event nodes are defined. The conditional strength function is: in Represents the target event At any moment Trigger strength; Indicates base strength; Represents the time decay kernel function; Represents the time constant; Indicates the function that triggers the effect; Represents the feature vector of the parent event; Represents the differentiation of the event counting process; Using the land-sea energy coupling index as a priori constraint for the hyperedge weights, the hyperedge weight function is defined as follows: in Indicates the superedge At any moment The weights; , and Indicates the weighting coefficient; Indicates time The energy coupling index; The correlation coefficient represents the correlation between features of parent events. Indicates the first Feature vectors of each parent event; Represents the natural exponential function; Indicates the spatial attenuation coefficient; Indicates the spatial distance between the parent event and the child event; A time attention mechanism is introduced to adaptively adjust the influence of the parent event at different time intervals. The time attention weight is defined as follows: in Indicates time Lower parent event Pair Events Attention weights; Represents the query vector; Represents the key vector; Indicates the vector dimension; Indicates sub-event The set of parent events; Represents the key vector; Design a soft intervention mechanism to estimate the counterfactual lift of a set of parent events to the target event. Its definition is as follows: in and Indicates the start and end times of the observation; Indicates the original trigger strength; Indicates the triggering intensity after soft intervention; This represents the smoothing constant.

5. The method for identifying disaster chains in a combined land-sea extreme parallel meteorological event as described in claim 1, characterized in that, Step S4 includes: Meteorological and monitoring data are divided into categories after bandpass filtering. Characteristic frequency bands To simultaneously capture high-frequency local disturbances and low-frequency regional variations; For each frequency band The instantaneous phase of the signal is obtained using the Hilbert transform. ; In phase synchronization analysis, an extreme value weighting mechanism is introduced. The weighted multi-source phase synchronization coefficient is defined as follows: in Indicates frequency band The weighted synchronization coefficient is below; Indicates the total number of signals; Represents the phase of a complex number; The normalized extremum weight function represents each signal; Introducing directional propagation constraints by defining the main propagation direction vector in geographic coordinates. Construct an anisotropic phase diffusion operator Its mathematical expression is: in Represents the directional loss function; Indicates a spatial region; This represents the gradient operator along the main propagation direction; Indicates frequency band The phase field below; Represents the anisotropy coefficient; This represents the gradient operator perpendicular to the main propagation direction; Topological persistent cohomology analysis is introduced, the phase similarity matrix is ​​thresholded, and the length of the persistent stripe of its cohomology group as a function of the threshold is calculated. ; Considering extreme value weighting, directional constraints, and topological stability, a comprehensive parallel synchronization index is defined. : in Indicates the total number of frequency bands; Indicates the direction-weighted synchronization coefficient; Indicates the frequency band weighting coefficient; when When the threshold is exceeded, it can be determined that multiple disasters are occurring in parallel extreme events.

6. The method for identifying disaster chains in a combined land-sea extreme parallel meteorological event as described in claim 1, characterized in that, Step S5 includes: Energy Coupling Index + Causal triggering strength Parallel synchronization metrics The risk values ​​of the units in the disaster chain risk matrix are formed by weighted superposition and defined as follows: in Represents longitude, latitude coordinates, and time variables; , and This represents the risk weighting coefficient.

7. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform the land-sea coupled extreme parallel meteorological disaster chain identification analysis method as described in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a computer, perform the land-sea coupled extreme parallel meteorological disaster chain identification method as described in any one of claims 1-6.

Images