Automatic identification methods, systems and media for storm surge intensity impact areas

Abstract

This invention provides a method, system, and medium for automatically identifying storm surge intensity impact areas, relating to the field of disaster emergency response technology. The method includes: collecting storm surge-related impact data, performing intensity impact analysis, and constructing a set of storm surge intensity impact factors; classifying, training, and fusing historical storm surge event samples to generate a storm surge intensity impact level discriminator; performing twin fusion to obtain a storm surge intensity prediction agent; performing water level prediction simulation to generate a multi-dimensional storm surge intensity field; performing cluster analysis and geographic matching identification to output a spatial distribution map of the intensity impact area, and conducting storm surge early warning and forecasting. This invention solves the technical problem that existing technologies for identifying storm surge intensity impact areas often rely on simple numerical simulation models or use only a single storm surge impact factor, failing to flexibly adapt to complex marine environmental changes, resulting in insufficient reliability and accuracy of the identification results.

Description

Technical Field

[0001] This invention relates to the field of disaster emergency response technology, specifically to a method, system, and medium for automatically identifying areas affected by storm surge intensity. Background Technology

[0002] Storm surges are abnormal rises in sea level caused by strong meteorological systems. They are usually accompanied by extreme weather and changes in the marine environment, causing significant impacts on the safety of life and property, the ecological environment, and economic activities in and around coastlines. The intensity and affected area of ​​storm surges often vary dramatically, making accurate prediction and real-time monitoring of them of significant scientific and applied value. However, traditional methods for identifying storm surge intensity and affected areas often rely on simple numerical simulation models or use only a single storm surge influencing factor, such as meteorological conditions. These methods fail to comprehensively consider the multiple influencing factors of storm surges and cannot flexibly adapt to complex changes in the marine environment. This results in insufficient reliability and accuracy of storm surge intensity predictions, which in turn affects the automatic identification and early warning of storm surge affected areas, posing challenges to disaster prevention and mitigation efforts. Summary of the Invention

[0003] This application provides an automatic identification method, system, and medium for storm surge intensity impact areas, aiming to solve the technical problem that existing technologies for identifying storm surge intensity impact areas often rely on simple numerical simulation models or use only a single storm surge influencing factor, which cannot flexibly adapt to complex marine environmental changes, resulting in insufficient reliability and accuracy of the identification results.

[0004] The first aspect disclosed in this application provides an automatic identification method for storm surge intensity impact areas. The method includes: collecting storm surge-related impact data, which integrates meteorological data, tide level monitoring data, and marine environmental element data; performing intensity impact analysis on the storm surge-related impact data to construct a storm surge intensity impact factor set; classifying, training, and fusing historical storm surge event samples according to the storm surge intensity impact factor set to generate a storm surge intensity impact level discriminator; constructing a storm surge impact forecasting model for the target area; performing a twin fusion of the storm surge impact forecasting model and the storm surge intensity impact level discriminator to obtain a storm surge intensity prediction agent; performing water level prediction simulation on current storm surge process data based on the storm surge intensity prediction agent to generate a multidimensional storm surge intensity field; performing cluster analysis and geographic matching identification on the multidimensional storm surge intensity field to output a spatial distribution map of the intensity impact area; and using the spatial distribution map of the intensity impact area to conduct storm surge early warning and forecasting.

[0005] The second aspect of this application discloses an automatic identification system for storm surge intensity impact areas. This system is used in the aforementioned automatic identification method for storm surge intensity impact areas. The system includes: an intensity impact analysis module for collecting storm surge-related impact data, which integrates meteorological data, tide level monitoring data, and marine environmental element data; performing intensity impact analysis on the storm surge-related impact data to construct a storm surge intensity impact factor set; and a classification training and fusion module for classifying and training historical storm surge event samples according to the storm surge intensity impact factor set to generate storm surge intensity impact... The system includes: a storm surge level discriminator; a twin fusion module for constructing a storm surge impact forecasting model for the target area, fusing the storm surge impact forecasting model and the storm surge intensity impact level discriminator to obtain a storm surge intensity prediction agent; a water level prediction simulation module for performing water level prediction simulation on current storm surge process data based on the storm surge intensity prediction agent to generate a multi-dimensional storm surge intensity field; and a storm surge early warning and forecasting module for performing cluster analysis and geographic matching identification on the multi-dimensional storm surge intensity field, outputting a spatial distribution map of the intensity impact area, and using the spatial distribution map of the intensity impact area to issue storm surge early warnings and forecasts.

[0006] The third aspect disclosed in this application provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for automatically identifying the storm surge intensity impact area in the first aspect.

[0007] One or more technical solutions provided in this application have at least the following beneficial effects:

[0008] By integrating meteorological data, tide monitoring data, and marine environmental data, the comprehensiveness and accuracy of storm surge impact analysis are ensured. Intensity impact analysis of different data sources constructs a set of storm surge intensity influencing factors, enabling the extraction of key factors affecting storm surges. This allows the model to more accurately capture the core driving factors during storm surge occurrence, aiding in subsequent intensity classification and prediction. Classification training on historical storm surge event samples generates a storm surge intensity impact level discriminator, accurately classifying the intensity of future storm surge events and ensuring the model's accurate predictive capabilities. It can effectively identify and judge the potential intensity level of future storm surges. By fusion of the storm surge impact forecasting model and the storm surge intensity impact level discriminator, a storm surge impact prediction model with multi-dimensional intelligent analysis capabilities is formed. The storm surge intensity prediction agent achieves synergy between forecasting and intensity assessment, thereby improving overall prediction reliability and real-time response capabilities. By simulating current storm surge data using this agent, a multi-dimensional storm surge intensity field is generated. This allows for detailed spatial quantification of the storm surge's impact, creating detailed spatial data charts that provide precise support for subsequent regional impact analysis and early warning. Through cluster analysis and geographic matching of the multi-dimensional storm surge intensity field, regions with different storm surge intensities can be identified, and a spatial distribution map of the storm surge's output intensity impact areas can be output. This helps to clarify which areas may be affected and can be further used for storm surge early warning and forecasting. This visualization of spatial data enables rapid identification of hazardous areas, providing decision support for storm surge emergency response and disaster prevention.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the automatic identification method for storm surge intensity impact area provided in an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of the structure of the automatic identification system for storm surge intensity impact area provided in an embodiment of this application.

[0012] Figure labeling: Intensity impact analysis module 10, classification training fusion module 20, twin fusion module 30, water level prediction simulation module 40, storm surge warning and forecasting module 50. Detailed Implementation

[0013] This application provides an automatic identification method, system, and medium for storm surge intensity impact areas, which solves the technical problem that existing storm surge intensity impact area identification often relies on simple numerical simulation models or only uses a single storm surge influencing factor, which cannot flexibly adapt to complex marine environmental changes, resulting in insufficient reliability and accuracy of the identification results.

[0014] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0015] Example 1, as Figure 1 As shown in the embodiments of this application, an automatic identification method for storm surge intensity impact areas is provided, the method comprising:

[0016] Storm surge-related impact data are collected, which integrates meteorological data, tide level monitoring data, and marine environmental element data. Intensity impact analysis is performed on the storm surge-related impact data to construct a set of storm surge intensity impact factors.

[0017] Real-time meteorological data includes information such as wind speed, air pressure, precipitation, and temperature. Storm surges are typically triggered by severe meteorological conditions, such as typhoons and hurricanes; therefore, meteorological data is crucial to understanding storm surges. Tidal monitoring data, including high and low tides and tidal amplitude, is an indicator of storm surge intensity, as changes in tide level directly affect storm surge performance. Marine environmental data includes ocean temperature, current velocity, and sea level changes, all of which influence the formation and evolution of storm surges. The collected data are analyzed to assess the impact of each factor on storm surges. Key factors influencing storm surge intensity are extracted from multiple data sources, forming a set of storm surge intensity influencing factors for subsequent prediction and analysis.

[0018] Based on the aforementioned set of storm surge intensity influencing factors, historical storm surge event samples are classified, trained, and fused to generate a storm surge intensity influencing level discriminator.

[0019] Historical storm surge event data is selected as samples of historical storm surge events. This data contains information such as the intensity, affected area, and time of actual storm surges, which can serve as the basis for model training. Features are extracted from these historical storm surge event samples, including various factors in the storm surge intensity influencing factor set. Then, they are classified according to storm surge intensity, and a classification model is used for training, such as random forest or neural network. The trained model can output the storm surge intensity level, serving as a discriminator for the storm surge intensity influence level.

[0020] A storm surge impact forecasting model for the target area is constructed, and the storm surge impact forecasting model and the storm surge intensity impact level discriminator are fused together to obtain a storm surge intensity prediction agent.

[0021] Acquiring relevant geographic information of the target area, such as topographic data and coastline data, helps the model understand the region's natural conditions, thus influencing storm surge simulation and prediction. For example, coastline morphology, tidal differences, and topographic relief all affect storm surge intensity and propagation paths. A storm surge impact forecasting model for the target area is constructed by combining this geographic information. The established storm surge impact forecasting model and a storm surge intensity impact level discriminator are then fused using a twin agent. The key to this twin agent fusion is combining the prediction results of both models. The storm surge intensity impact level discriminator is used to further classify and discriminate the simulation results. The twin agent, based on the interaction of the two models, not only outputs numerical simulation results of storm surges but also includes intensity level predictions, improving the accuracy and reliability of storm surge prediction.

[0022] Based on the storm surge intensity prediction agent, the current storm surge process data is used to perform water level prediction simulation and generate a multidimensional storm surge intensity field.

[0023] The system acquires data on ongoing storm surge events, including real-time meteorological information, tide level monitoring, and changes in the marine environment. Specifically, this includes data on wind speed, air pressure, and tide level variations. A storm surge intensity prediction agent is then used to simulate the current storm surge data and predict water level changes over a future period. This simulation combines the relationship between historical and current data with the characteristics of the geographical region to predict water level change trends at different locations. The results of the water level prediction simulation are used to generate a multi-dimensional storm surge intensity field.

[0024] Cluster analysis and geographic matching identification are performed on the multidimensional storm surge intensity field to output a spatial distribution map of the intensity influence area, and storm surge early warning and forecasting are carried out through the spatial distribution map of the intensity influence area.

[0025] Clustering algorithms, such as K-means and DBSCAN, are used to analyze the multidimensional storm surge intensity field. Different regions are grouped according to their storm surge intensity, impact range, and expected water level. Each cluster represents a region with similar storm surge characteristics. The clustering results are matched with actual geographic data, mapping the clustering analysis results to specific geographic space to ensure that the predicted storm surge impact area matches the real world. Through geographic matching identification, a spatial distribution map of the intensity impact area is generated, which can clearly identify the specific areas affected by storm surges, marking which areas will be affected by strong storm surges and which areas will have water level changes exceeding the warning line. This information is used for storm surge early warning and forecasting, and is provided to relevant departments and the public to help them make timely emergency responses and preventive measures.

[0026] Furthermore, the construction of the storm surge intensity influencing factor set includes:

[0027] Outlier cleaning and spatiotemporal alignment are performed on the storm surge-related impact data to obtain usable storm surge-related impact data; a set of storm surge intensity influencing factors is obtained, which includes meteorological factors, ocean dynamic factors, and topographic factors; in accordance with each influencing factor in the storm surge intensity influencing factor set, the usable storm surge-related impact data is sequentially processed to extract related influencing factors to obtain a preliminary set of intensity factor related influencing factors; the preliminary set of intensity factor related influencing factors is subjected to impact analysis and screening to construct a storm surge intensity influencing factor set.

[0028] Storm surge impact data may contain erroneous or outlier values, possibly due to sensor malfunctions, data transmission errors, or data entry mistakes. Statistical analysis or rule-based algorithms can be used to identify these outliers. Removing outliers ensures the accuracy of subsequent analysis and prevents them from interfering with model training and storm surge prediction. Storm surge impact data typically comes from multiple sources, such as meteorological observatories and marine observation stations. These data have different timestamps and collection frequencies. Timestamps are used to synchronize data from different data sources over time. If some data is missing, interpolation methods can be used to fill the gaps. After these processes, the resulting usable storm surge impact data is cleaned and aligned, providing accurate input for subsequent analysis and modeling.

[0029] Meteorological factors, including wind speed, air pressure, precipitation, temperature, and humidity, directly influence the formation and intensity of storm surges. For example, strong winds and low air pressure can cause sea levels to rise, resulting in storm surges. Ocean dynamic factors, including ocean currents, waves, sea surface temperature, and sea level changes, directly affect the propagation, expansion, and intensity of storm surges. For instance, warm water can enhance the intensity of storm surges, while strong ocean currents can alter their direction and propagation speed. Topographic factors, including coastline morphology, seabed topography, and coastal topography (such as slope and beach width), have a significant impact on storm surges because different topographic features can alter wave height and intrusion depth. Integrating these factors constitutes a comprehensive set of factors influencing storm surge intensity, providing a foundation for subsequent feature extraction and analysis.

[0030] Based on the various factors influencing storm surge intensity, including meteorological, ocean dynamic, and topographic factors, corresponding influencing factors are extracted from available storm surge correlation data. For example, wind speed and air pressure are extracted from meteorological data as meteorological factors; sea level change and wave height are extracted from ocean data as ocean dynamic factors; and slope and coastline morphology are extracted from topographic data as topographic factors. These extracted factors are then organized according to storm surge events to obtain a preliminary set of intensity factor correlation influencing factors.

[0031] The initial set of influencing factors related to intensity factors is screened, and the relationship between each factor and storm surge intensity is determined through correlation analysis, such as Pearson correlation coefficient. Based on the results of the influence analysis, factors that have a significant impact on the prediction of storm surge intensity are retained, while redundant or low-impact factors are removed. Finally, an optimized set of storm surge intensity influencing factors is constructed, which is the core input for storm surge intensity prediction and early warning.

[0032] Furthermore, the process of conducting impact analysis and screening on the preliminary intensity factor-related influencing factor set to construct a storm surge intensity influencing factor set includes:

[0033] The storm surge-related impact data are classified according to the preliminary intensity factor associated impact factor set to obtain preliminary impact factor feature data and storm surge intensity level data; random forest training is performed using the storm surge intensity level data as labels and the preliminary impact factor feature data as input features to extract the preliminary impact factor contribution set; an impact factor contribution threshold is set according to the intensity prediction accuracy target; the preliminary intensity factor associated impact factor set is then filtered based on the preliminary impact factor contribution set according to the impact factor contribution threshold to construct the storm surge intensity impact factor set.

[0034] Based on the preliminary intensity factor-related impact factor set, the storm surge-related impact data is classified to provide data for subsequent model training. Through classification, preliminary impact factor feature data for each storm surge event is extracted from the storm surge-related impact data. These data include meteorological factors, ocean dynamic factors, and topographic factors. These factors serve as input features, describing the different characteristics of different storm surge events. At the same time, based on the actual impact of storm surge events, storm surge intensity level data is extracted. This data is used to label the intensity level of each storm surge event, serving as label data for the machine learning model.

[0035] The Random Forest algorithm is used for training. Random Forest is an ensemble learning algorithm that builds multiple decision trees for classification or regression tasks. It can effectively handle the relationships between high-dimensional data and complex features. Here, it is used for classification tasks. Preliminary influencing factor feature data is used as input features, and storm surge intensity level data is used as labels. During the training process, the algorithm automatically learns the relationship between each input feature and the storm surge intensity level, and builds a series of decision trees. The final classification result is obtained through a voting mechanism. After training, Random Forest provides an importance score for each feature. The contribution of each influencing factor to the storm surge intensity prediction is quantified, resulting in a preliminary influencing factor contribution set. Factors with higher contributions are key factors for storm surge intensity prediction, while factors with lower contributions are redundant features that can be removed in subsequent screening stages.

[0036] Based on the target accuracy of intensity prediction, set a threshold for the contribution of influencing factors. For example, set factors with a contribution higher than a certain percentage to be retained factors. By setting the threshold for the contribution of influencing factors, factors that contribute less to the model can be removed, reducing model complexity, improving computational efficiency, and avoiding overfitting.

[0037] Based on the set threshold for the contribution of influencing factors, the factors that have the greatest impact on storm surge intensity prediction are selected. Factors with a contribution below the threshold are removed, and the remaining factors form a more concise and effective set of storm surge intensity influencing factors. These factors have been precisely selected to maximize the accuracy of storm surge intensity prediction.

[0038] Furthermore, the storm surge intensity impact level discriminator includes:

[0039] Historical storm surge event samples are correlated and classified according to the storm surge intensity influencing factor set to obtain a storm surge intensity factor sample set; a deep neural network structure is used to train the storm surge intensity factor sample set for intensity discrimination to obtain a branch intensity factor level discriminator; the intensity influencing factor decision coefficient set is determined according to the preliminary influencing factor contribution set; the branch intensity factor level discriminator is fused based on the intensity influencing factor decision coefficient set to generate the storm surge intensity influence level discriminator.

[0040] Based on the storm surge intensity influencing factor set, the influencing factor characteristics of each event are extracted from historical storm surge event samples, including meteorological factors, oceanographic factors, and topographic factors. These factors collectively describe the characteristics of each storm surge event. Simultaneously, the historical storm surge event samples are labeled according to their actual impact, for example, different levels such as weak, moderate, and strong. Through the classification and labeling of historical storm surge event samples, a storm surge intensity factor sample set is constructed, with each sample consisting of its characteristic data and corresponding intensity level label.

[0041] A deep neural network (DNN) was employed for training to determine storm surge intensity. DNNs learn complex relationships in data through multi-level nonlinear transformations, effectively handling high-dimensional and nonlinear features. The input features consist of influencing factor data from a storm surge intensity factor sample set. This data serves as the input to the neural network for training the model, and the output label is the storm surge intensity level label, which is the prediction target of the neural network. The DNN structure includes multiple hidden layers, each containing several neurons. Nonlinear features are introduced through activation functions. Finally, the DNN structure outputs a classification result, i.e., the storm surge intensity level, based on the input feature data. Backpropagation and gradient descent algorithms are used to train the DNN structure. By minimizing the loss function, the network's weights and biases are adjusted, enabling the network to accurately predict storm surge intensity levels. After training, a branch intensity factor level discriminator is generated, which can predict the intensity level of storm surge events based on the input storm surge influencing factor data.

[0042] The relative importance of each factor is determined using the preliminary set of influencing factors' contribution scores. The role of each factor in storm surge intensity prediction is quantified into an intensity influence factor decision coefficient, representing the degree of influence of each factor on the final decision. Factors with higher contribution scores are assigned higher decision coefficients, while those with lower contribution scores are assigned lower weights. By standardizing the contribution scores of each factor in the preliminary set of influencing factors' contribution scores, the intensity influence factor decision coefficient for each factor is calculated and used in the subsequent decision fusion process to accurately reflect the contribution of each factor to storm surge intensity determination in the final model.

[0043] Based on the decision coefficient set of intensity influence factors, the branch intensity factor level discriminators are fused. Specifically, the output of each branch intensity factor level discriminator is combined with its corresponding intensity influence factor decision coefficient and weighted. Factors with higher contributions will have a greater impact on the final prediction result, resulting in the final discrimination result. Finally, through decision fusion, a storm surge intensity influence level discriminator is obtained. This discriminator can integrate the contributions of all influence factors and predict the intensity level of storm surge based on the input storm surge influence factor data.

[0044] Furthermore, the storm surge impact forecasting model for the target area includes:

[0045] Acquire basic geographic data and historical storm surge impact data for the target area, wherein the basic geographic data includes topographic data and coastline data; construct grid division rules, and divide the target area into grids based on the basic geographic data according to the grid division rules to obtain the target grid area; select the ADCIRC+SWAN coupling model, and use the ADCIRC+SWAN coupling model to read the historical storm surge impact data to simulate storm surge operation in the target grid area, thereby constructing the storm surge impact forecast model.

[0046] Basic geographic data includes topographic data and coastline data. Topographic data includes elevation information, seabed topography, and land topography of the target area. This data is used to determine the propagation path, energy distribution, and potential impact area of ​​storm surges. Typically, topographic data is obtained through digital elevation models (DEMs). Coastline data refers to the precise location, shape, and variations of the coastline. It determines the entry, expansion, and ultimate impact of storm surges on different areas. Coastline data is usually derived from satellite imagery, marine survey data, or maps. Historical storm surge impact data includes detailed records of past storm surge events, specifically including the time, duration, location, and intensity of storm surges, as well as the maximum water level, elevation, and degree of damage.

[0047] Grid partitioning involves dividing the target area into multiple discrete grid cells. Each grid cell is used to store and process storm surge prediction data. The size of the grid cell is determined based on the characteristics of the target area and the required prediction accuracy. Smaller grid cells can provide higher spatial resolution, but also have a greater computational burden. The grid partitioning process is combined with the basic geographic data within the target area to ensure that the topographic changes in key areas are accurately represented. After grid partitioning, the target grid area obtained will serve as the calculation unit for the storm surge impact forecasting model.

[0048] ADCIRC is a numerical model used to simulate tides, storm surges, and ocean circulation. It is widely used in hydrodynamic simulations of coastal and nearshore areas. ADCIRC can handle the generation and propagation of storm surges, as well as their interactions with waves and tides. SWAN is a numerical model for simulating nearshore waves, used in storm surge research to analyze the interactions between waves and tides, and storm surges. SWAN can accurately calculate wave propagation, wave height, and wave energy. The coupling of these two models can provide more accurate predictions of storm surge levels, waves, and current velocities, especially in complex coastal topography and shallow water areas.

[0049] Historical storm surge impact data is used as input for the model. The ADCIRC+SWAN coupled model is used to simulate storm surges in the target area, calculating physical quantities such as water level and waves in different grid cells. The simulation process combines the complex interactions of storm surge intensity, path changes, marine environment, and topography. Finally, the predicted storm surge water level, wave, and current velocity data are output. Based on the model output data, the model is further adjusted and optimized to accurately predict the occurrence, development, and impact area of ​​storm surges. Through long-term historical data accumulation and model optimization, the accuracy of the model is gradually improved, ultimately forming a stable and reliable storm surge impact forecasting model.

[0050] Furthermore, the agent for predicting storm surge intensity includes:

[0051] A twin fusion framework for intelligent agents is constructed, comprising a forecast branch, a discriminator branch, and a collaborative fusion layer. The storm surge impact forecast model and the storm surge intensity impact level discriminator are embedded into the forecast branch and the discriminator branch, respectively, to obtain the storm surge impact forecast branch layer and the storm surge intensity discriminator branch layer. The storm surge impact forecast branch layer and the storm surge intensity discriminator branch layer are collaboratively fused and interacted through the collaborative fusion layer to construct a storm surge intensity prediction intelligent agent.

[0052] The intelligent agent twin fusion framework is a multi-module structure designed to achieve interactive fusion between different data sources and prediction tasks through collaborative work, resulting in more accurate storm surge intensity prediction. The forecasting branch is responsible for forecasting the impact of storm surges, primarily used to predict water levels and waves based on historical and real-time storm surge data. The discriminator branch is responsible for determining the storm surge intensity level, judging the impact intensity level of the storm surge based on the storm surge intensity factor and prediction data. The collaborative fusion layer integrates the outputs of the storm surge impact forecasting branch and the storm surge intensity discrimination branch to make the final storm surge intensity prediction.

[0053] A storm surge impact forecasting model is embedded into the forecasting branch to obtain the storm surge impact forecasting branch layer, which is used to calculate the physical characteristics of storm surge processes, such as water level changes and wave propagation. A storm surge intensity impact level discriminator is embedded into the discriminator branch to obtain the storm surge intensity discriminator branch layer, which is used to determine the storm surge intensity level based on the predicted storm surge characteristic data. These two branches will independently predict storm surge water level and intensity, providing data support for subsequent fusion.

[0054] The collaborative fusion layer receives storm surge level and wave prediction data from the storm surge impact forecasting branch layer and intensity level prediction data from the storm surge intensity discrimination branch layer. It then performs collaborative fusion on these data using different fusion algorithms, such as weighted average, attention mechanism, and deep learning fusion. Through collaborative fusion, the output data of the storm surge impact forecasting branch layer and the storm surge intensity discrimination branch layer are optimized and merged to form a comprehensive prediction result. Finally, the storm surge intensity prediction agent can output the storm surge intensity prediction for the target area and provide further early warning and forecast based on indicators such as intensity and water level changes.

[0055] Furthermore, the construction of the storm surge intensity prediction agent includes:

[0056] Based on the collaborative fusion layer, a branch layer data interaction mechanism and a decision fusion strategy are determined; based on the branch layer data interaction mechanism, the output data of the storm surge impact forecast branch layer and the storm surge intensity discrimination branch layer are transmitted and interacted to obtain storm surge impact distribution-intensity level data; an attention mechanism is introduced and the decision fusion strategy is used to perform fusion prediction simulation and dynamic feedback optimization on the storm surge impact distribution-intensity level data to construct the storm surge intensity prediction agent.

[0057] In the collaborative fusion layer, the branch layer data interaction mechanism refers to how different branch layers exchange information. The purpose of data interaction is to allow each branch layer to obtain the output information of the other, thereby influencing and complementing each other in the prediction process. For example, the prediction results output by each branch can be passed to another branch as additional input features, so that each branch can integrate the prediction results of the other branch when making predictions. The decision fusion strategy refers to how to merge the prediction results from different branches to generate the final storm surge intensity prediction. For example, different weights can be assigned according to the accuracy or importance of each branch, and the final result can be generated by weighted averaging.

[0058] Based on the data exchange mechanism between the branch layers, the storm surge impact forecasting branch layer and the storm surge intensity discrimination branch layer can exchange information. For example, the output data of the storm surge impact forecasting branch layer can be used as input features to pass to the storm surge intensity discrimination branch layer to more accurately determine the intensity level of the storm surge; conversely, the output data of the storm surge intensity discrimination branch layer can be used as additional information for the storm surge impact forecasting branch layer, affecting the water level prediction results. Through this two-way interaction, the model can gradually optimize the comprehensive assessment of storm surge impact and intensity, ultimately obtaining storm surge impact distribution-intensity level data.

[0059] Attention mechanisms help the model focus on key features and ignore less important ones by calculating the importance of different input features and assigning them different weights. Specifically, based on the importance of the prediction task, the model can adaptively adjust its focus according to the features of storm surge impact data in different regions. For example, the intensity of certain regions has a greater impact on water level changes, and the attention mechanism will automatically assign higher weights to these regions. By introducing attention mechanisms and adopting decision fusion strategies, the model can more intelligently fuse and predict storm surge impact distribution and intensity level data. This means that the model not only merges data but also performs deep feature learning, using data from various regions to make detailed storm surge intensity predictions. Dynamic feedback optimization refers to adjusting the model's prediction strategy by analyzing the error of each prediction result. By introducing feedback mechanisms, the model can gradually learn and optimize its decisions, ensuring continuous evolution of predictions. By integrating attention mechanisms and decision fusion strategies, the constructed storm surge intensity prediction agent can make predictions in multiple dimensions and dynamically optimize based on real-time data.

[0060] Furthermore, the spatial distribution map of the output intensity influence area includes:

[0061] Based on the multidimensional storm surge intensity field, the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution are determined; cluster analysis is performed on the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution to obtain a set of storm surge intensity affected areas; geographical matching and identification are then performed on the set of storm surge intensity affected areas to output the spatial distribution map of the intensity affected areas.

[0062] The storm surge impact area refers to the geographical range that a storm surge will affect. In a multidimensional storm surge intensity field, based on changes in intensity and water level data, it is possible to determine which areas will be affected by the storm surge, such as coastal areas, islands, and estuaries. Within the storm surge impact area, based on the predicted storm surge intensity, these areas are divided into different regional impact intensity levels, such as mild, moderate, and severe. Specific methods typically involve setting thresholds based on factors such as the maximum storm surge water level, wave height, and wind speed to determine the intensity level for each region. Regional water level spatial distribution refers to the water level changes in different regions during a storm surge. Based on storm surge water level prediction data, water level distribution maps can be drawn, displaying the water level height and its changes in each region.

[0063] Cluster analysis divides the storm surge-affected area into different regional clusters based on the intensity level of the regional impact and the spatial distribution of regional water levels. For example, using K-means clustering, after cluster analysis, the storm surge-affected area is divided into multiple storm surge intensity-affected area clusters. Each storm surge intensity-affected area cluster represents a region with similar storm surge characteristics. For example, one cluster represents a high-intensity storm surge region, another cluster represents a medium-intensity region, and a third cluster represents a region with weaker impact.

[0064] For each storm surge intensity impact area cluster obtained from cluster analysis, it is matched and identified with the actual geographic region. This process involves mapping the regions in the storm surge intensity impact area cluster to actual geographic coordinates or features. For example, satellite imagery or high-resolution maps can be used to align the clustered regions with the actual geographic space. Through geographic matching and identification, the regional distribution, intensity level, and water level changes of each storm surge intensity impact area cluster are accurately plotted on the actual geographic space, resulting in a spatial distribution map of the intensity impact area. This map shows the impact area of ​​the storm surge and uses colors, shapes, and other methods to indicate the intensity level and water level distribution of different regions.

[0065] In summary, the automatic identification method for storm surge intensity impact areas provided in this application has the following technical effects:

[0066] By integrating meteorological data, tide monitoring data, and marine environmental data, the comprehensiveness and accuracy of storm surge impact analysis are ensured. Intensity impact analysis of different data sources constructs a set of storm surge intensity influencing factors, enabling the extraction of key factors affecting storm surges. This allows the model to more accurately capture the core driving factors during storm surge occurrence, aiding in subsequent intensity classification and prediction. Classification training on historical storm surge event samples generates a storm surge intensity impact level discriminator, accurately classifying the intensity of future storm surge events and ensuring the model's accurate predictive capabilities. It can effectively identify and judge the potential intensity level of future storm surges. By fusion of the storm surge impact forecasting model and the storm surge intensity impact level discriminator, a storm surge impact prediction model with multi-dimensional intelligent analysis capabilities is formed. The storm surge intensity prediction agent achieves synergy between forecasting and intensity assessment, thereby improving overall prediction reliability and real-time response capabilities. By simulating current storm surge data using this agent, a multi-dimensional storm surge intensity field is generated. This allows for detailed spatial quantification of the storm surge's impact, creating detailed spatial data charts that provide precise support for subsequent regional impact analysis and early warning. Through cluster analysis and geographic matching of the multi-dimensional storm surge intensity field, regions with different storm surge intensities can be identified, and a spatial distribution map of the storm surge's output intensity impact areas can be output. This helps to clarify which areas may be affected and can be further used for storm surge early warning and forecasting. This visualization of spatial data enables rapid identification of hazardous areas, providing decision support for storm surge emergency response and disaster prevention.

[0067] Example 2, based on the same inventive concept as the automatic identification method for storm surge intensity impact areas in the foregoing examples, such as... Figure 2 As shown in the embodiment of this application, an automatic identification system for storm surge intensity impact areas is provided, the system comprising:

[0068] The intensity impact analysis module 10 is used to collect storm surge-related impact data, which integrates meteorological data, tide level monitoring data, and marine environmental element data. Intensity impact analysis is performed on the storm surge-related impact data to construct a set of storm surge intensity impact factors. The classification training and fusion module 20 is used to classify, train, and fuse historical storm surge event samples according to the storm surge intensity impact factor set to generate a storm surge intensity impact level discriminator. The twin fusion module 30 is used to construct a storm surge impact forecasting model for the target area, and to fuse the storm surge impact forecasting model and the storm surge intensity impact level discriminator into a twin fusion to obtain a storm surge intensity prediction agent. The water level prediction simulation module 40 is used to perform water level prediction simulation based on the storm surge intensity prediction agent on current storm surge process data to generate a multi-dimensional storm surge intensity field. The storm surge early warning and forecasting module 50 is used to perform cluster analysis and geographic matching identification on the multi-dimensional storm surge intensity field, output a spatial distribution map of the intensity impact area, and conduct storm surge early warning and forecasting based on the spatial distribution map of the intensity impact area.

[0069] Furthermore, the intensity influence analysis module 10 is used to perform the following operation steps:

[0070] Outlier cleaning and spatiotemporal alignment are performed on the storm surge-related impact data to obtain usable storm surge-related impact data; a set of storm surge intensity influencing factors is obtained, which includes meteorological factors, ocean dynamic factors, and topographic factors; in accordance with each influencing factor in the storm surge intensity influencing factor set, the usable storm surge-related impact data is sequentially processed to extract related influencing factors to obtain a preliminary set of intensity factor related influencing factors; the preliminary set of intensity factor related influencing factors is subjected to impact analysis and screening to construct a storm surge intensity influencing factor set.

[0071] Furthermore, the intensity influence analysis module 10 is used to perform the following operation steps:

[0072] The storm surge-related impact data are classified according to the preliminary intensity factor associated impact factor set to obtain preliminary impact factor feature data and storm surge intensity level data; random forest training is performed using the storm surge intensity level data as labels and the preliminary impact factor feature data as input features to extract the preliminary impact factor contribution set; an impact factor contribution threshold is set according to the intensity prediction accuracy target; the preliminary intensity factor associated impact factor set is then filtered based on the preliminary impact factor contribution set according to the impact factor contribution threshold to construct the storm surge intensity impact factor set.

[0073] Furthermore, the classification training fusion module 20 is used to perform the following operation steps:

[0074] Historical storm surge event samples are correlated and classified according to the storm surge intensity influencing factor set to obtain a storm surge intensity factor sample set; a deep neural network structure is used to train the storm surge intensity factor sample set for intensity discrimination to obtain a branch intensity factor level discriminator; the intensity influencing factor decision coefficient set is determined according to the preliminary influencing factor contribution set; the branch intensity factor level discriminator is fused based on the intensity influencing factor decision coefficient set to generate the storm surge intensity influence level discriminator.

[0075] Furthermore, the twin fusion module 30 is used to perform the following operation steps:

[0076] Acquire basic geographic data and historical storm surge impact data for the target area, wherein the basic geographic data includes topographic data and coastline data; construct grid division rules, and divide the target area into grids based on the basic geographic data according to the grid division rules to obtain the target grid area; select the ADCIRC+SWAN coupling model, and use the ADCIRC+SWAN coupling model to read the historical storm surge impact data to simulate storm surge operation in the target grid area, thereby constructing the storm surge impact forecast model.

[0077] Furthermore, the twin fusion module 30 is used to perform the following operation steps:

[0078] A twin fusion framework for intelligent agents is constructed, comprising a forecast branch, a discriminator branch, and a collaborative fusion layer. The storm surge impact forecast model and the storm surge intensity impact level discriminator are embedded into the forecast branch and the discriminator branch, respectively, to obtain the storm surge impact forecast branch layer and the storm surge intensity discriminator branch layer. The storm surge impact forecast branch layer and the storm surge intensity discriminator branch layer are collaboratively fused and interacted through the collaborative fusion layer to construct a storm surge intensity prediction intelligent agent.

[0079] Furthermore, the twin fusion module 30 is used to perform the following operation steps:

[0080] Based on the collaborative fusion layer, a branch layer data interaction mechanism and a decision fusion strategy are determined; based on the branch layer data interaction mechanism, the output data of the storm surge impact forecast branch layer and the storm surge intensity discrimination branch layer are transmitted and interacted to obtain storm surge impact distribution-intensity level data; an attention mechanism is introduced and the decision fusion strategy is used to perform fusion prediction simulation and dynamic feedback optimization on the storm surge impact distribution-intensity level data to construct the storm surge intensity prediction agent.

[0081] Furthermore, the storm surge warning and forecasting module 50 is used to perform the following operational steps:

[0082] Based on the multidimensional storm surge intensity field, the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution are determined; cluster analysis is performed on the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution to obtain a set of storm surge intensity affected areas; geographical matching and identification are then performed on the set of storm surge intensity affected areas to output the spatial distribution map of the intensity affected areas.

[0083] Through the foregoing detailed description of the automatic identification method for storm surge intensity impact areas, those skilled in the art can clearly understand the automatic identification system for storm surge intensity impact areas in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0084] Example 3 provides a storage medium on which a computer program is stored, which, when executed by a processor, implements any step of Example 1.

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automatically identifying the area affected by storm surge intensity, characterized in that, The method includes: Collect storm surge-related impact data, which integrates meteorological data, tide level monitoring data, and marine environmental element data. Perform intensity impact analysis on the storm surge-related impact data to construct a set of storm surge intensity impact factors. Based on the set of storm surge intensity influencing factors, historical storm surge event samples are classified, trained, and fused to generate a storm surge intensity influencing level discriminator. A storm surge impact forecasting model for the target area is constructed, and the storm surge impact forecasting model and the storm surge intensity impact level discriminator are fused together to obtain a storm surge intensity prediction agent; The agent for predicting storm surge intensity includes: A smart agent twin fusion framework is constructed, which includes a prediction branch, a discriminator branch, and a collaborative fusion layer; The storm surge impact forecast model and the storm surge intensity impact level discriminator are embedded into the forecast branch and the discriminator branch, respectively, to obtain the storm surge impact forecast branch layer and the storm surge intensity discriminator branch layer; The storm surge impact forecasting branch layer and the storm surge intensity discrimination branch layer are collaboratively fused and interacted through the collaborative fusion layer to construct a storm surge intensity prediction agent. The construction of the storm surge intensity prediction agent includes: Based on the collaborative fusion layer, the data interaction mechanism and decision fusion strategy for the branch layer are determined; Based on the branch layer data interaction mechanism, the output data of the storm surge impact forecasting branch layer and the storm surge intensity discrimination branch layer are transmitted and interacted to obtain storm surge impact distribution-intensity level data. An attention mechanism is introduced to employ the aforementioned decision fusion strategy to perform fusion prediction simulation and dynamic feedback optimization on the storm surge impact distribution-intensity level data, thereby constructing the storm surge intensity prediction agent. Based on the storm surge intensity prediction agent, the water level prediction simulation is performed on the current storm surge process data to generate a multi-dimensional storm surge intensity field. Cluster analysis and geographic matching identification are performed on the multidimensional storm surge intensity field to output a spatial distribution map of the intensity influence area, and storm surge early warning and forecasting are carried out through the spatial distribution map of the intensity influence area; The output intensity influence area spatial distribution map includes: Based on the multidimensional storm surge intensity field, the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution are determined. Cluster analysis was performed based on the storm surge impact area, the regional impact intensity level, and the regional water level spatial distribution to obtain a set of storm surge intensity impact area clusters; Geographic matching and identification are performed on the clusters of areas affected by storm surge intensity in sequence, and a spatial distribution map of the areas affected by intensity is output.

2. The method for automatic identification of storm surge intensity impact areas as described in claim 1, characterized in that, The set of storm surge intensity influencing factors includes: The storm surge-related impact data is cleaned of outliers and spatiotemporally aligned to obtain usable storm surge-related impact data. Obtain a set of factors influencing storm surge intensity, which includes meteorological factors, ocean dynamic factors, and topographic factors; According to each influencing factor in the storm surge intensity influencing factor set, the available storm surge related influencing data are sequentially processed to extract the associated influencing factors, thereby obtaining a preliminary set of intensity factor associated influencing factors. An impact analysis and screening of the preliminary intensity factor correlation influencing factor set was conducted to construct a storm surge intensity influencing factor set.

3. The method for automatic identification of storm surge intensity impact areas as described in claim 2, characterized in that, The process involves conducting an impact analysis and screening of the preliminary intensity factor correlation influencing factor set to construct a storm surge intensity influencing factor set, including: The storm surge associated impact data are classified according to the preliminary intensity factor associated impact factor set to obtain preliminary impact factor characteristic data and storm surge intensity level data. Using the storm surge intensity level data as labels and the preliminary impact factor feature data as input features, random forest training is performed to extract the preliminary impact factor contribution set. Based on the target accuracy of intensity prediction, set the threshold for the contribution of impact factors; Based on the contribution threshold of the impact factors, the set of impact factors associated with the preliminary intensity factors is screened to construct the set of storm surge intensity impact factors.

4. The method for automatic identification of storm surge intensity impact areas as described in claim 3, characterized in that, The storm surge intensity impact level discriminator includes: Based on the storm surge intensity influencing factor set, the historical storm surge event samples are correlated and classified to obtain the storm surge intensity factor sample set; A deep neural network structure was used to train the storm surge intensity factor sample set to determine the intensity, thereby obtaining a branch intensity factor level discriminant. Based on the preliminary set of contribution factors, determine the set of decision coefficients for intensity impact factors; Based on the set of decision coefficients for the intensity influence factors, the branch intensity factor level discriminator is fused to generate the storm surge intensity influence level discriminator.

5. The method for automatically identifying the storm surge intensity impact area as described in claim 1, characterized in that, The storm surge impact forecasting model for the target area includes: Acquire basic geographic data and historical storm surge impact data for the target area, wherein the basic geographic data includes topographic data and coastline data; Construct grid division rules, and divide the target area into grids based on the basic geographic data according to the grid division rules to obtain the target grid area; Select the ADCIRC+SWAN coupled model, use the ADCIRC+SWAN coupled model to read the historical data of storm surge impact to simulate the operation of storm surge in the target grid area, and construct the storm surge impact forecast model.

6. An automatic identification system for storm surge intensity impact areas, characterized in that, The system for implementing the automatic identification method for storm surge intensity impact areas according to any one of claims 1-5, the system comprising: The intensity impact analysis module is used to collect storm surge-related impact data, which integrates meteorological data, tide level monitoring data, and marine environmental element data. The module performs intensity impact analysis on the storm surge-related impact data and constructs a set of storm surge intensity impact factors. The classification training and fusion module is used to classify, train and fuse historical storm surge event samples according to the storm surge intensity influencing factor set to generate a storm surge intensity influencing level discriminator. The twin fusion module is used to construct a storm surge impact forecast model for the target area, and to fuse the storm surge impact forecast model and the storm surge intensity impact level discriminator into a twin fusion to obtain a storm surge intensity prediction agent; The water level prediction simulation module is used to perform water level prediction simulation based on the storm surge intensity prediction agent on the current storm surge process data, and generate a multi-dimensional storm surge intensity field. The storm surge warning and forecasting module is used to perform cluster analysis and geographic matching identification on the multidimensional storm surge intensity field, output a spatial distribution map of the intensity influence area, and perform storm surge warning and forecasting based on the spatial distribution map of the intensity influence area; The output intensity influence area spatial distribution map includes: Based on the multidimensional storm surge intensity field, the storm surge affected area, the regional impact intensity level, and the regional water level spatial distribution are determined. Cluster analysis was performed based on the storm surge impact area, the regional impact intensity level, and the regional water level spatial distribution to obtain a set of storm surge intensity impact area clusters; Geographic matching and identification are performed on the clusters of areas affected by storm surge intensity in sequence, and a spatial distribution map of the areas affected by intensity is output.

7. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for automatically identifying the storm surge intensity impact area as described in any one of claims 1 to 5.

Images