Storm surge intelligent early warning method and system based on multi-model coupling, and medium

Abstract

The application discloses a storm surge intelligent early warning method and system based on multi-model coupling and a medium, and relates to the technical field of disaster early warning.The method comprises the following steps: establishing a basic data set; acquiring real-time meteorological data and marine data, performing fuzzy path fitting, and establishing a fitting result; performing regional grid segmentation by using the basic data set and the fitting result, and establishing a grid segmentation result; inputting the real-time data and the basic data into a simulation unit, performing simulation operation, and establishing node simulation data; acquiring a node preset safety response measure, performing joint early warning identification according to the response measure and the simulation data, and reporting an early warning signal, so as to solve the technical problems that the existing storm surge early warning method has strong subjectivity in experience prediction, inherent errors of a single model lead to insufficient simulation accuracy, and early warning timeliness and refinement degree are insufficient, and achieve the technical effect that the accuracy, timeliness and refinement degree of storm surge early warning are improved through fuzzy path fitting and multi-model coupling simulation operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of disaster warning, in particular to a storm surge intelligent warning method and system based on multi-model coupling and a medium. BACKGROUND

[0002] With the global climate change and the increase of extreme weather events, storm surge, as an important marine disaster, has become a major problem threatening the safety of life and property and the stability of the ecological environment in coastal areas. Storm surge is a phenomenon of rapid rise of sea level caused by cyclones, typhoons, tropical storms, etc., and its intensity and influence range are affected by the complex interaction of various factors such as ocean and meteorology. When storm surge occurs, it may cause seawater to submerge low-lying areas, damage infrastructure, and even cause casualties. Therefore, how to effectively predict storm surge in advance and timely issue warnings has become an important technical requirement in the current storm surge disaster prevention and mitigation work.

[0003] Traditional storm surge warning methods mainly rely on experience prediction and numerical simulation prediction, but both methods have certain limitations. Although experience prediction can provide a certain warning, it relies too much on historical data and the personal experience of forecasters, and has the problems of incomplete data, poor timeliness and strong subjectivity; while numerical prediction, although it uses mathematical models and computer technology for simulation, can accurately reflect the changes of meteorology and ocean, but due to the high requirement of computing resources, model simplification error and single model use, it is still difficult to achieve all-round and fine-grained warning. SUMMARY

[0004] The present application provides a storm surge intelligent warning method and system based on multi-model coupling and a medium, which solves the technical problems of insufficient simulation accuracy, insufficient warning timeliness and fine-grained degree caused by strong subjectivity of experience prediction and inherent error of single model in existing storm surge warning methods.

[0005] In view of the above problems, the present application provides a storm surge intelligent warning method and system based on multi-model coupling and a medium.

[0006] In a first aspect, the application provides a storm surge intelligent early warning method based on multi-model coupling, which comprises: collecting geographical space data, population and infrastructure data to establish a basic data set; acquiring real-time meteorological data and real-time ocean data, performing fuzzy path fitting according to the real-time meteorological data and the real-time ocean data to establish a fuzzy path fitting result; performing regional grid segmentation using the basic data set and the fuzzy path fitting result to establish a concerned grid segmentation result; inputting the real-time meteorological data, the real-time ocean data and the basic data set into a multi-model coupling simulation unit, performing a multi-model coupling simulation unit simulation operation based on the concerned grid segmentation result to establish node simulation data of the concerned grid segmentation result; acquiring a preset safety response measure of a node in the concerned grid segmentation result, performing joint early warning identification according to the preset safety response measure and the node simulation data, and reporting an early warning signal.

[0007] In a second aspect, the application provides a storm surge intelligent early warning system based on multi-model coupling, which comprises: a basic data acquisition module that collects geographical space data, population and infrastructure data to establish a basic data set; a fuzzy path fitting module that acquires real-time meteorological data and real-time ocean data, performs fuzzy path fitting according to the real-time meteorological data and the real-time ocean data, and establishes a fuzzy path fitting result; a regional grid segmentation module that performs regional grid segmentation using the basic data set and the fuzzy path fitting result to establish a concerned grid segmentation result; a simulation operation module that inputs the real-time meteorological data, the real-time ocean data and the basic data set into a multi-model coupling simulation unit, performs a multi-model coupling simulation unit simulation operation based on the concerned grid segmentation result, and establishes node simulation data of the concerned grid segmentation result; and a joint early warning identification module that acquires a preset safety response measure of a node in the concerned grid segmentation result, performs joint early warning identification according to the preset safety response measure and the node simulation data, and reports an early warning signal.

[0008] In a third aspect, the application provides a computer readable medium storing a computer program, which, when executed by a processor, implements the storm surge intelligent early warning method based on multi-model coupling.

[0009] The one or more technical solutions provided in the application have at least the following technical effects or advantages:

[0010] Firstly, geographical space data, population and infrastructure data are collected to establish a basic data set; subsequently, real-time meteorological data and real-time marine data are obtained, and fuzzy path fitting is performed according to the real-time meteorological data and the real-time marine data to establish a fuzzy path fitting result; further, the basic data set and the fuzzy path fitting result are used to perform regional grid segmentation to establish a concerned grid segmentation result; then, the real-time meteorological data, the real-time marine data and the basic data set are input into a multi-model coupling simulation unit to perform a multi-model coupling simulation unit simulation operation based on the concerned grid segmentation result to establish node simulation data of the concerned grid segmentation result; finally, a preset safety response measure of a node in the concerned grid segmentation result is obtained, and joint early warning identification is performed according to the preset safety response measure and the node simulation data to report an early warning signal. The technical problems of insufficient simulation accuracy, insufficient timeliness and insufficient refinement degree of the existing storm surge early warning method due to strong subjectivity of experience prediction and inherent error of a single model are solved, and the technical effects of improving the accuracy, timeliness and refinement degree of storm surge early warning by performing fuzzy path fitting and multi-model coupling simulation operation are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0012] Figure 1 A flowchart of a storm surge intelligent early warning method based on multi-model coupling provided by the embodiments of the present application is shown.

[0013] Figure 2 A structure diagram of a storm surge intelligent early warning system based on multi-model coupling provided by the embodiments of the present application is shown.

[0014] Legend: basic data acquisition module 11, fuzzy path fitting module 12, regional grid segmentation module 13, simulation operation module 14, joint early warning identification module 15. DETAILED DESCRIPTION

[0015] The present application provides a storm surge intelligent early warning method, system and medium based on multi-model coupling, which solves the technical problems of insufficient simulation accuracy, insufficient timeliness and insufficient refinement degree of the existing storm surge early warning method due to strong subjectivity of experience prediction and inherent error of a single model.

[0016] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0017] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units need not be limited to only those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices.

[0018] Embodiment one, as shown in the present application provides a storm surge intelligent early warning method based on multi-model coupling, wherein the method comprises: Figure 1

[0019] Collecting geographic space data, population and infrastructure data, and establishing a basic data set.

[0020] In one embodiment, before the target area is warned of storm surge, geographic space data, population and infrastructure data of the target area need to be collected, wherein the geographic space data can be obtained through geographic information system (GIS) or remote sensing technology, including but not limited to water depth, terrain, coastline, geographic coordinate system information, which will provide accurate geographic information for storm surge warning, helping the model to divide the grid and define the region in space. Population data is usually obtained from the National Bureau of Statistics, local government or special population survey agencies, including but not limited to population density, distribution of residential areas, which helps to assess the impact of storm surge disaster on different areas of the population. Infrastructure data can be obtained from government departments, public data platforms, etc., including but not limited to power facilities, water conservancy facilities, transportation facilities, building distribution, etc., which plays an important role in assessing the impact of storm surge on the local. By integrating these different sources of data, a unified basic data set is established to provide necessary input for the model, support subsequent regional grid segmentation, simulation operation and early warning identification steps, and ensure that storm surge simulation and early warning can cover all aspects of risk.

[0021] Obtaining real-time meteorological data and real-time ocean data, performing fuzzy path fitting according to the real-time meteorological data and the real-time ocean data, and establishing a fuzzy path fitting result.

[0022] ​In one embodiment, first, real-time meteorological data of the target area is obtained through ground meteorological observation stations, including but not limited to the center position of a typhoon or tropical storm, wind speed, wind direction, and pressure field. Then, real-time ocean data is obtained from ocean monitoring stations, tide observation stations, etc., including but not limited to tide level, sea water temperature, sea current speed, and sea surface pressure. These ocean data and meteorological data are interrelated and are of great significance to predicting the development path, intensity, and affected area of a storm surge. Subsequently, historical meteorological data and historical ocean data of the target area are obtained, fuzzy subsets are generated using the fuzzy C-means algorithm, and Takagi-Sugeno type fuzzy inference rules are established to complete the construction of the fuzzy rule base. After that, the fuzzy controller parameters are initialized, real-time meteorological data and real-time ocean data are input into the fuzzification interface to calculate the activation degree of each fuzzy subset, and a typhoon moving direction probability density function (PDF) is generated by defuzzification through weighted average method. Then, an adaptive neuro-fuzzy inference system (ANFIS) is used to dynamically correct the path probability at a preset period (e.g., every 6 hours). Then, based on the PDF, multiple Monte Carlo sampling paths are generated and assigned with probability weights, and the particle filter algorithm is used to select effective paths to calculate the probability weighted average center line (longitude and latitude coordinate sequence), 2σ confidence ellipse (long axis / short axis / rotation angle) determined by the covariance matrix, and path and coastline intersection time window (±3 hour error range) statistical characteristics. Finally, these statistical characteristics are stored according to the geometric field (such as the center line and uncertainty ellipse) and the attribute field (such as the path probability and key node timestamp) to construct a fuzzy path fitting result. This fuzzy path fitting result contains the center path of the storm surge and the uncertainty band of the path, which can reflect the most likely path of the storm surge and the deviation range of the path, providing reliable path prediction and uncertainty analysis for subsequent storm surge simulation and warning, so that the entire warning can cope with complex and variable meteorological and ocean conditions, improving the accuracy and timeliness of storm surge warning.

[0023] The basic data set and the fuzzy path fitting result are used to perform regional grid segmentation to establish a focus grid segmentation result.

[0024] In one embodiment, after obtaining the basic data set and the fuzzy path fitting result, the center line and uncertainty band in the fuzzy path fitting result are combined with the geographical features in the basic data set, and a distance decay factor is used to divide the region into grids. During the grid segmentation process, the population density and infrastructure importance of the region are also considered to make the grid division more accurate and accurate to ensure accurate simulation of the impact of the storm surge on these areas. In addition, historical disaster data is also included to further refine the grid distribution. Finally, by considering these factors comprehensively, a focus grid segmentation result is generated, which provides accurate spatial division for subsequent simulation operations and warning, ensuring comprehensive and accurate monitoring and warning of areas that may be affected by the storm surge.

[0025] Further, the area grid segmentation using the basic data set and the fuzzy path fitting result includes:

[0026] The path center line and the uncertainty band of the fuzzy path fitting result are obtained, and a storm intensity index is configured according to the real-time meteorological data. After setting a distance decay factor using the storm intensity index, the grid density evaluation under distance decay is performed according to the path center line, the uncertainty band and the distance decay factor to establish a first segmentation constraint. The terrain feature extraction is performed on the geographic spatial data in the basic data set to establish local elevation features, slope features and coastal proximity features, and the terrain feature extraction result is used to establish a second segmentation constraint. After extracting the population and infrastructure data in the basic data set, the third segmentation constraint is established according to the population density features and the facility criticality features. The area grid segmentation is performed using the first segmentation constraint, the second segmentation constraint and the third segmentation constraint to establish the attention grid segmentation result.

[0027] Preferably, the path center line and the uncertainty band are first extracted from the fuzzy path fitting result, the path center line represents the most possible development path of the storm, and the uncertainty band represents the possible deviation range of the storm path. The maximum wind speed, the central pressure difference, the sea surface temperature and the typhoon moving speed are obtained from the real-time meteorological data, and the maximum-minimum normalization method is used to process these data. After the normalization is completed, the standard data is weighted to obtain the storm intensity index, which is used to reflect the potential destructive power of the storm. Then, the storm intensity index is input into the formula: to calculate the distance decay factor, wherein D(r) is the distance decay factor at the vertical distance r from the path center line, λ is a reference decay coefficient, and is usually 0.05 km -1, S is a storm intensity index. After setting the distance attenuation factor, the density of the grid in the target area is evaluated based on the path centerline, uncertainty band, and attenuation factor. In this process, a polar coordinate grid is established in the target area. For each grid node, the distance from the node to each vertex of the path centerline is multiplied by the weight of the path segment (adaptively assigned based on path curvature), and then the results are accumulated to obtain the grid density index. By matching the grid density index of each grid with the density threshold, the grid size corresponding to different grid density indexes is matched, for example, when the grid density index is less than 0.3, the grid size is 5km, and when the grid density index is between 0.3 and 0.6, the grid size is 2km. These grid sizes form the first segmentation constraint. Through this density adjustment, the grid segmentation of the target area can be more refined, and the warning needs of different intensity storms can be met. Then, all geographic spatial data of the target area is extracted from the basic data set, and the local elevation, slope, and coastal proximity features are obtained by matching according to the preset feature key name. The local elevation feature reflects the terrain undulation of the area, the slope feature helps to evaluate the change of water flow, and the coastal proximity feature helps to evaluate the risk of sea flooding. Then, grid refinement rules are defined based on these features, such as elevation less than 5m and slope less than 5°, forcing the grid size to be less than or equal to 1km; coastal proximity feature less than or equal to 5km, encrypted to 0.5km; slope greater than 15°, relaxed to 5km (due to terrain blocking effect). By combining these grid refinement rules, the second segmentation constraint is constructed to simulate more accurately. Then, the population distribution and infrastructure data of the target area are extracted from the basic data set. These data usually include population density and facility criticality (such as the functional importance of power, transportation, water conservancy and other infrastructure). By inputting the population density and facility criticality into the corresponding mapping table for matching, the grid size corresponding to different population density and facility criticality can be determined, and the third segmentation constraint is established. Generally, in areas with high population density or high facility criticality, the grid should be encrypted first to accurately predict the specific impact of storm surge on these areas. Finally, the first segmentation constraint, the second segmentation constraint, and the third segmentation constraint are considered comprehensively, and these segmentation constraints are combined through weighted fusion to form the final attention grid segmentation result. This attention grid segmentation result includes detailed information of each grid unit, such as location, size, population density, facility impact, etc. These information will provide fine spatial data support for subsequent multi-model coupling simulation, disaster loss assessment and warning release, and further improve the accuracy and timeliness of the warning.

[0028] Further, the use of the first segmentation constraint, the second segmentation constraint, the third segmentation constraint for regional grid segmentation to establish an attention grid segmentation result includes:

[0029] The historical disaster data set of the target area is acquired, data analysis of the historical disaster data set is performed, disaster frequency characteristics and disaster severity characteristics are established, the fourth segmentation constraint is configured using the disaster frequency characteristics and the disaster severity characteristics, and after the fourth segmentation constraint is added to the first segmentation constraint, the second segmentation constraint and the third segmentation constraint, the regional grid segmentation is performed.

[0030] Optionally, when the target area has a disaster situation in the past time period, the historical disaster data set of the target area is acquired, including the time, location, intensity of the disaster, economic loss caused by the disaster, casualties, etc. Subsequently, data analysis is performed on the historical disaster data set, the frequency of storm surges occurring in the target area within a certain time range is counted, the disaster frequency characteristics are obtained, and the normalized inundation depth, duration, economic loss, and casualty rate are weighted to obtain the disaster severity characteristics. Then, the disaster frequency characteristics and the disaster severity characteristics are input into the mapping table for comparison to determine the grid size corresponding to different disaster frequency characteristics and disaster severity characteristics, and the fourth segmentation constraint is constructed. Then, the fourth segmentation constraint is combined with the first segmentation constraint, the second segmentation constraint, and the third segmentation constraint obtained above, the attention grid segmentation result is determined by weighting, and the target area is finely divided according to the attention grid segmentation result, to ensure that higher density grids are used in areas with severe storms, close distances, high disaster frequencies, and severe disasters, so as to more finely simulate and predict storm surges in these areas, and improve the accuracy of early warning and the timeliness of response.

[0031] The real-time meteorological data, the real-time marine data, and the basic data set are input into the multi-model coupling simulation unit, a multi-model coupling simulation unit simulation operation based on the attention grid segmentation result is performed, and node simulation data of the attention grid segmentation result is established.

[0032] In one embodiment, real-time meteorological data, real-time ocean data and basic data sets are first input as input data into a pre-constructed multi-model coupled simulation unit. After the multi-model coupled simulation unit receives these data, it uses the geographic spatial data in the basic data set and the attention grid segmentation result to initialize the model network inside the multi-model coupled simulation unit. This model network is an unstructured triangular network used to accurately describe the geographic spatial characteristics of the target area. On this basis, the multi-model coupled simulation unit will simulate the wind field according to the data related to typhoons or tropical storms in the received real-time meteorological data and real-time ocean data, such as center position, actual path information, wind speed, wind direction, tidal level, current speed, sea surface pressure, etc. The simulation of the wind field simulates the dynamic change process of the tidal level change, current speed, and wind wave propagation of storm surge at different time nodes and different spatial positions to describe the meteorological influence on the target area and generate simulation data of the wind field. By combining these wind field simulation data with real-time meteorological data and real-time ocean data, the final coupled simulation data is generated, and these coupled simulation data are divided into corresponding grid nodes to form node simulation data for each grid node. These node simulation data will be used for subsequent warning processing to provide timely and accurate data support for disaster prevention and mitigation decision-making.

[0033] Further, the execution of the multi-model coupled simulation unit simulation operation based on the attention grid segmentation result establishes node simulation data of the attention grid segmentation result, including:

[0034] The attention grid segmentation result is subjected to grid granularity similar clustering to establish a clustering result, and the clustering result is subjected to precision level identification. The multi-model coupled simulation unit simulation operation of the clustering result is performed in order from high to low precision level, wherein the processing scale of the multi-model coupled simulation unit is set based on the precision level, and the multi-model coupled simulation unit includes an ADCIRC+SWAN coupled model. A linkage simulation rule is configured, the linkage simulation rule is a data correlation compensation rule of high-precision node simulation data as low-precision node simulation data, simulation compensation is performed using the linkage simulation rule, and node simulation data is established.

[0035] Preferably, before the simulation operation is performed, firstly, all the grids in the attention grid segmentation result are clustered according to the granularity similarity, that is, the grids with similar spatial distribution and storm surge influence are converged, and in this process, the similarity between the grids is calculated according to the size, density, vertical distance from the path center line, local water depth gradient, facility criticality and other data of the grids, and the calculation result is compared with the spatial distance threshold in the spatial continuity constraint, and the grids meeting the spatial distance threshold requirement are converged together to form a clustering result. Subsequently, for each clustering cluster in the clustering result, it is counted whether there is a key facility such as a nuclear power plant, a port, a hospital, etc. in each clustering cluster, and if there is, the clustering cluster is marked as a high-precision area, and then the average population density of each clustering cluster is counted, and according to the interval in which the average population density is located, the corresponding precision level is allocated, such as a high-precision area and a low-precision area. In addition, for the clustering cluster with more than a preset coastal proximity, it is marked as a low-precision area, because these areas represent open sea areas. Subsequently, according to the determined precision level, a multi-model coupled simulation unit is used in turn from high to low to perform the simulation operation, and the multi-model coupled simulation unit includes an ocean circulation model (ADCIRC) and a wave model (SWAN), which can simulate the changes of storm surge, tidal level, sea current velocity, wave propagation and other phenomena. The multi-model coupled simulation unit simulates the occurrence, development and propagation of storm surges according to the processing scale corresponding to the different precision levels of each clustering cluster, and compensates according to the linkage simulation rule, that is, the distance between the high-precision node and the low-precision node is calculated according to the Euclidean distance, and at most three high-precision nodes closest to the low-precision node are selected. It should be noted that the distance between these high-precision nodes and the low-precision node is less than the preset proximity distance. Then, the distance between the high-precision node and the low-precision node is quantified as the weight of the high-precision node using the inverse distance method, and the simulation data of the high-precision node and the simulation data of the low-precision node are weighted and fused according to the weights to form the final node simulation data. Through this series of steps, the evolution process of the storm surge can be more accurately simulated to provide timely and accurate storm surge warning information for coastal areas and reduce disaster losses.

[0036] Further, the grid granularity similarity clustering of the attention grid segmentation result and the establishment of the clustering result include:

[0037] The simulation history data calling is performed, the simulation boundary fragmentation identification under the spatial distance influence is performed according to the calling result, and the calibration spatial distance threshold is established. The spatial continuity constraint is set by using the calibration spatial distance threshold. In the process of performing the grid granularity similarity clustering, the clustering constraint is performed through the spatial continuity constraint to establish the clustering result.

[0038] Optionally, when performing grid granularity similar clustering of the grid segmentation result of interest, simulation history data is first extracted from past storm surge simulation results, including the evolution of spatial distribution, tidal level, current, wind speed and other variables of past storm surge events. Then, the historical data is identified to obtain the position where the difference of adjacent grid nodes' hydrodynamic parameters (such as flow rate) exceeds the threshold value, and the similarity distance of the size, density, and vertical distance from the path center line of the grids on both sides of the position is calculated using the Euclidean distance. By summarizing all the calculated similarity distances, a histogram is constructed, and the similarity distance with a cumulative probability exceeding the probability threshold value is extracted from the histogram as the calibrated spatial distance threshold value, and this calibrated spatial distance threshold value is set as the spatial continuity constraint. Then, the similarity distance of each two grids in the grid segmentation result of interest is calculated using the Euclidean distance, and the calculated similarity distance is compared with the spatial continuity constraint. If the similarity distance is less than the spatial continuity constraint, it means that the two grids are similar, and at this time, the two grids will be divided into a cluster. This process is repeated until all grids are divided into corresponding cluster clusters. For each cluster, the similarity distance between the grids inside the cluster is less than the spatial continuity constraint. After clustering is completed, a clustering result can be obtained, which reasonably divides the grids and ensures the consistency and continuity of the influence of adjacent areas during simulation.

[0039] Further, the multi-model coupling simulation unit simulation operation of the clustering result is performed in order from high to low according to the precision level, including:

[0040] According to the real-time weather data and the real-time ocean data, a weather evolution trend is predicted, an evolution trend level is established, a load identification of the multi-model coupling simulation unit is performed, a basic model processing scale is set according to the load identification result and the evolution trend level, and after the scale compensation of the basic model processing scale is performed according to the precision level, the multi-model coupling simulation unit simulation operation of the clustering result is performed in order from high to low according to the precision level.

[0041] Optionally, first, the real-time meteorological data and real-time ocean data are input into a pre-constructed long short-term memory network (LSTM) to predict the future weather evolution and generate an evolution trend level, such as a high level (strong storm), indicating that the storm is strong and has a wide impact, which may cause a serious storm surge; a medium level (medium storm), indicating that the storm is moderate and has a relatively local impact, which may cause a moderate storm surge; and a low level (weak storm), indicating that the storm is weak and has a small impact, which may cause a limited storm surge. The long short-term memory network is obtained by iterative training according to historical meteorological data sequences and historical ocean data sequences through forward propagation, loss calculation, back propagation (such as a cross-entropy loss function), and parameter optimization (such as an Adam optimizer). Subsequently, the Prometheus+Grafana tool is used to collect the computing resource usage of the multi-model coupled simulation unit, including GPU memory occupancy, CPU load, and memory occupancy, and the load identification result is obtained by normalizing and weighting these load indicators. Then, the load identification result is combined with the evolution trend level, and the basic model processing scale is obtained by matching with a mapping table. The basic model processing scale is then compensated using the precision level, i.e., the precision level is multiplied by the basic model processing scale to obtain the compensated model processing scale of each grid. Finally, according to the size order of the precision level, the multi-model coupled simulation unit simulates the grid according to the corresponding compensated model processing scale. In this process, the ADCIRC model is responsible for simulating ocean circulation and tidal processes, calculating the flow velocity and tidal level changes in the calculation area, and the SWAN model simulates wave propagation and the interaction between waves and water flow. Both models share data through mutual feedback and perform coupled calculations at each time step to update the tidal level, flow velocity, and wave state of each grid node in real time. During the simulation process, as the storm evolves, the two models dynamically update the calculation results to ensure accurate prediction of the storm surge, while adjusting the precision and compensating the data to ensure consistency between different precision areas, providing a basis for disaster warning.

[0042] The preset safety response measures of the nodes in the attention grid segmentation result are obtained, joint warning identification is performed according to the preset safety response measures and the node simulation data, and a warning signal is reported.

[0043] In one embodiment, after obtaining the node simulation data, the preset safety response measures of each node in the grid segmentation result of interest are obtained. These safety response measures are set in advance according to the geographic information, historical disaster data, population density, infrastructure importance and other factors of each grid node, including but not limited to evacuation measures, facility protection, alarm release, etc. When a storm surge occurs, whether the early warning needs to be started is judged according to the simulation data of each node combined with the preset safety response measures. If it is needed, the corresponding early warning signal is generated according to the risk level, and the early warning information is timely released through the ways such as short message, broadcast, network platform, etc. to provide guidance for emergency response for relevant personnel.

[0044] Further, the obtaining of the preset safety response measures of the nodes in the grid segmentation result of interest, the joint early warning identification according to the preset safety response measures and the node simulation data, and the reporting of the early warning signal comprise:

[0045] The preset safety response measures are analyzed to obtain a response level, a response action operation and a response time node. The safety response measure adaptation evaluation of the simulation data is performed based on the response level, the response action operation and the response time node, an abnormal adaptation value is established, the joint early warning identification is completed according to the abnormal adaptation value, and the early warning signal is reported.

[0046] Preferably, firstly, the preset safety response measures corresponding to each grid node are analyzed, and the response level, response action operation and response time node are extracted from the preset key name, wherein the response level is set according to the severity of storm surge, the influence on local community and facilities, etc., including the threshold of various indicators; the response action operation refers to the specific countermeasures under each response level, such as emergency evacuation, road closure, emergency repair start, etc., different response levels correspond to different action steps; the response time node is the time node of the start of the countermeasures, which helps to start the necessary measures in time in the early warning process. Subsequently, the preset safety response measures are adapted and analyzed according to the simulation data of each node, that is, the tidal level in the simulation data is compared with the critical tidal level of the preset response level, the response level adaptation degree is determined by calculating the absolute difference of the tidal level and the ratio of the critical tidal level of the preset response level; the inundation depth in the simulation data is compared with the critical inundation depth, the inundation depth feasibility factor is calculated, the flow rate is compared with the critical flow rate, the flow rate feasibility factor is calculated, the response action operation adaptation degree is obtained by weighting the inundation depth feasibility factor and the flow rate feasibility factor; the predicted time of storm surge arrival is compared with the response time node to determine the response time node adaptation degree. Then, the response level adaptation degree, the response action operation adaptation degree and the response time node adaptation degree are weighted, and then 1 is subtracted from the weighted result to obtain the abnormal adaptation value, which represents the degree of inconsistency between the response measures and the simulation data. Finally, whether the response measures need to be adjusted or optimized to complete the joint early warning identification is determined according to the abnormal adaptation value of each node, if the abnormal adaptation value is too high, it means that the response measures may not be suitable for the current storm surge situation, at this time, the early warning signal will be generated according to the abnormal adaptation value, indicating that the response measures need to be re-determined, in addition, the corresponding early warning level (such as blue, yellow, orange, red, etc.) will also be generated. On the contrary, the early warning level and the response measures will be packaged together as an early warning signal for reporting, thereby providing reliable protection for disaster prevention and personnel safety.

[0047] Further, after the early warning signal is reported, it includes:

[0048] The verification time interval is set according to the early warning level of the early warning signal; the execution state verification of the real safety response measures is performed in the verification time interval, and the verification anomaly is established; the early warning superposition analysis is performed according to the verification anomaly and the verification time interval, and the early warning signal is reconstructed.

[0049] Preferably, according to the early warning level in the early warning signal, the verification time interval is set to ensure the verification of the safety response measures within a certain time range. For different early warning levels, the length of the verification time interval is different. High-level early warning (such as red early warning) will set a shorter verification time interval, usually a few hours or a few minutes, while low-level early warning (such as blue early warning) allows longer response preparation. Subsequently, within the verification time interval, the execution state of the real safety response measures of each node is verified to check whether the response measures are executed as planned, for example, whether evacuation, facility reinforcement, emergency repair and other measures are started in time. If it is found that a response measure is executed late or does not meet the expectation during the verification process, a verification exception identifier will be established, for example, if the evacuation measure is delayed for more than the specified time, or the facility protection work is not completed, a higher abnormality degree will be recorded, indicating that the response measure is not executed on time. Then, according to the severity of the verification exception and the verification time interval, early warning superposition analysis will be performed, that is, the original early warning signal will be weighted and adjusted. At this time, the influence of the verification exception and the early warning level will be combined to determine whether the early warning level needs to be raised or the response measures need to be adjusted. If the verification exception is more serious, it may lead to the promotion of the early warning level, for example, the yellow early warning is promoted to orange or red early warning. If the exception is lighter, the original level may be maintained or a moderate adjustment may be made. Finally, based on the superposition analysis result, the early warning signal will be reconstructed, and the updated early warning signal will be published through various communication channels, including SMS, broadcast, mobile application, etc., to ensure that the residents, government departments and emergency response units in the affected areas obtain accurate early warning information and emergency response suggestions in time.

[0050] In summary, the embodiments of the present application have at least the following technical effects:

[0051] Firstly, geographical space data, population and infrastructure data are collected to establish a basic data set; then, real-time meteorological data and real-time marine data are obtained, and fuzzy path fitting is performed according to the real-time meteorological data and real-time marine data to establish a fuzzy path fitting result; further, the basic data set and the fuzzy path fitting result are used for regional grid segmentation to establish a concerned grid segmentation result; thereafter, the real-time meteorological data, the real-time marine data and the basic data set are input into a multi-model coupling simulation unit to perform a multi-model coupling simulation unit simulation operation based on the concerned grid segmentation result, and node simulation data of the concerned grid segmentation result are established; finally, a preset safety response measure of a node in the concerned grid segmentation result is obtained, and joint early warning identification is performed according to the preset safety response measure and the node simulation data, and a warning signal is reported. The technical problems of insufficient simulation accuracy, insufficient timeliness and refinement degree of early warning caused by strong subjectivity of experience prediction and inherent error of single model in the prior storm surge early warning method are solved, and the technical effects of improving the accuracy, timeliness and refinement degree of storm surge early warning by performing fuzzy path fitting and multi-model coupling simulation operation are achieved.

[0052] In the second embodiment, based on the same inventive concept as the storm surge intelligent early warning method based on multi-model coupling in the foregoing embodiments, as shown in the accompanying drawings, the present application provides a storm surge intelligent early warning system based on multi-model coupling, wherein the system comprises: Figure 2

[0053] The basic data acquisition module 11 collects geographical space data, population and infrastructure data to establish a basic data set; the fuzzy path fitting module 12 obtains real-time meteorological data and real-time marine data, and performs fuzzy path fitting according to the real-time meteorological data and real-time marine data to establish a fuzzy path fitting result; the regional grid segmentation module 13 uses the basic data set and the fuzzy path fitting result for regional grid segmentation to establish a concerned grid segmentation result; the simulation operation module 14 inputs the real-time meteorological data, the real-time marine data and the basic data set into a multi-model coupling simulation unit, performs a multi-model coupling simulation unit simulation operation based on the concerned grid segmentation result, and establishes node simulation data of the concerned grid segmentation result; and the joint early warning identification module 15 obtains a preset safety response measure of a node in the concerned grid segmentation result, and performs joint early warning identification according to the preset safety response measure and the node simulation data, and reports a warning signal.

[0054] Further, the regional grid segmentation module 13 is used to perform the following method:

[0055] ​The path center line and the uncertainty band of the fuzzy path fitting result are acquired, and a storm intensity index is configured according to the real-time meteorological data; after a distance decay factor is set by using the storm intensity index, grid density evaluation under distance decay is performed according to the path center line, the uncertainty band and the distance decay factor, a first segmentation constraint is established; terrain feature extraction is performed on the geographic spatial data in the basic data set, local elevation features, slope features and coastal proximity features are established, and a second segmentation constraint is established by using the terrain feature extraction result; after population and infrastructure data in the basic data set are extracted, a third segmentation constraint is established according to population density features and facility criticality features; regional grid segmentation is performed by using the first segmentation constraint, the second segmentation constraint and the third segmentation constraint, and a concerned grid segmentation result is established.

[0056] Further, the regional grid segmentation module 13 is configured to perform the following method:

[0057] A historical disaster data set of a target region is acquired, data analysis of the historical disaster data set is performed, disaster frequency features and disaster severity features are established, a fourth segmentation constraint is configured by using the disaster frequency features and the disaster severity features, and regional grid segmentation is performed after the fourth segmentation constraint is added to the first segmentation constraint, the second segmentation constraint and the third segmentation constraint.

[0058] Further, the simulation operation module 14 is configured to perform the following method:

[0059] Grid granularity similar clustering is performed on the concerned grid segmentation result, a clustering result is established, and precision level identification is performed on the clustering result; multi-model coupled simulation unit simulation operation is performed on the clustering result in order from high to low precision level, wherein the processing scale of the multi-model coupled simulation unit is set based on the precision level, and the multi-model coupled simulation unit includes an ADCIRC+SWAN coupled model; a linkage simulation rule is configured, the linkage simulation rule is a data association compensation rule of high-precision node simulation data as low-precision node simulation data, simulation compensation is performed by using the linkage simulation rule, and node simulation data is established.

[0060] Further, the simulation operation module 14 is configured to perform the following method:

[0061] Simulation history data calling is performed, simulation boundary fragmentation identification under spatial distance influence is performed according to the calling result, a calibrated spatial distance threshold is established, a spatial continuity constraint is set by using the calibrated spatial distance threshold, and clustering constraint is performed by using the spatial continuity constraint in the process of grid granularity similar clustering, so as to establish a clustering result.

[0062] Further, the simulation operation module 14 is configured to execute the following method:

[0063] According to the real-time weather data and the real-time ocean data, a weather evolution trend is predicted, an evolution trend level is established, load identification of a multi-model coupling simulation unit is performed, a basic model processing scale is set according to the load identification result and the evolution trend level, and after scale compensation of the basic model processing scale is performed through an accuracy level, the multi-model coupling simulation unit simulation operation of a clustering result is performed in order from high to low according to the accuracy level.

[0064] Further, the joint early warning identification module 15 is configured to execute the following method:

[0065] The preset safety response measures are analyzed to obtain a response level, a response action operation and a response time node, safety response measure adaptive evaluation of simulation data is performed based on the response level, the response action operation and the response time node, an abnormal adaptive value is established, joint early warning identification is completed according to the abnormal adaptive value, and a warning signal is reported.

[0066] Further, the joint early warning identification module 15 is configured to execute the following method:

[0067] A verification time interval is set according to a warning level of the warning signal, an execution state verification of a real safety response measure is performed in the verification time interval, a verification anomaly is established, early warning superposition analysis is performed according to the verification anomaly and the verification time interval, and the warning signal is reconstructed.

[0068] In the third embodiment, based on the same inventive concept as the storm surge intelligent early warning method based on multi-model coupling in the foregoing embodiments, the application provides a medium, and the medium stores a computer program. When the processor executes the computer program, the following steps are implemented: geographical space data and population and infrastructure data are collected and obtained to establish a basic data set; real-time weather data and real-time ocean data are obtained, fuzzy path fitting is performed according to the real-time weather data and the real-time ocean data, and a fuzzy path fitting result is established; the basic data set and the fuzzy path fitting result are used to perform regional grid segmentation to establish a grid segmentation result of interest; the real-time weather data, the real-time ocean data and the basic data set are input into a multi-model coupling simulation unit, multi-model coupling simulation unit simulation operation based on the grid segmentation result of interest is performed, and node simulation data of the grid segmentation result of interest are established; preset safety response measures of nodes in the grid segmentation result of interest are obtained, joint early warning identification is performed according to the preset safety response measures and the node simulation data, and a warning signal is reported.

[0069] It should be noted that the above-mentioned embodiment sequences of the present application are merely for description only, but not for representing the advantages and disadvantages of the embodiments. And the above-mentioned embodiments of the present specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0070] The above only describes the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0071] The specification and drawings are merely exemplary of the present application, and any and all modifications, variations, combinations or equivalents that are within the scope of the present application should be included. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A storm surge intelligent early warning method based on multi-model coupling, characterized in that, The method comprises: Collecting geographic space data, population and infrastructure data, and establishing a basic data set; Obtaining real-time meteorological data and real-time marine data, and performing fuzzy path fitting according to the real-time meteorological data and the real-time marine data to establish a fuzzy path fitting result; Using the basic data set and the fuzzy path fitting result to perform regional grid segmentation to establish a focus grid segmentation result; Inputting the real-time meteorological data, the real-time marine data, and the basic data set into a multi-model coupling simulation unit, performing multi-model coupling simulation unit simulation operation based on the focus grid segmentation result, and establishing node simulation data of the focus grid segmentation result; Obtaining a preset safety response measure of a node in the focus grid segmentation result, and performing joint early warning identification according to the preset safety response measure and the node simulation data to report an early warning signal; The execution of the multi-model coupling simulation unit simulation operation based on the focus grid segmentation result and the establishment of the node simulation data of the focus grid segmentation result comprise: Performing grid granularity similar clustering on the focus grid segmentation result to establish a clustering result, and performing precision level identification on the clustering result; Performing multi-model coupling simulation unit simulation operation on the clustering result in order from high to low according to the precision level, wherein the processing scale of the multi-model coupling simulation unit is set based on the precision level, and the multi-model coupling simulation unit comprises an ADCIRC+SWAN coupling model; Configuring a linkage simulation rule, wherein the linkage simulation rule is a data correlation compensation rule that high-precision node simulation data is used as low-precision node simulation data, performing simulation compensation on the simulation operation by using the linkage simulation rule to establish node simulation data; The grid granularity similar clustering on the focus grid segmentation result to establish a clustering result comprises: Performing simulation history data calling, and performing simulation boundary fragmentation identification under the influence of spatial distance according to the calling result to establish a calibrated spatial distance threshold; Setting a spatial continuity constraint by using the calibrated spatial distance threshold; In the process of performing grid granularity similar clustering, the spatial continuity constraint is used for clustering constraint to establish a clustering result; The multi-model coupling simulation unit simulation operation on the clustering result in order from high to low according to the precision level comprises: Performing weather evolution trend prediction according to the real-time meteorological data and the real-time marine data to establish an evolution trend level; Performing load identification of the multi-model coupling simulation unit, and setting a basic model processing scale according to the load identification result and the evolution trend level; After the scale compensation of the basic model processing scale by the precision level, the multi-model coupling simulation unit simulation operation on the clustering result is performed in order from high to low according to the precision level.

2. The storm surge intelligent early warning method based on multi-model coupling of claim 1, wherein, The regional grid segmentation using the basic data set and the fuzzy path fitting result to establish a focus grid segmentation result comprises: Obtaining a path center line and an uncertainty band of the fuzzy path fitting result, and configuring a storm intensity index according to the real-time meteorological data; After setting the distance decay factor by using the storm intensity indicator, a grid density evaluation under distance decay is performed according to the path centerline, the uncertainty band, and the distance decay factor, and a first segmentation constraint is established; Terrain feature extraction is performed on the geographic spatial data in the basic data set to establish local elevation features, slope features, and coastal proximity features, and a second segmentation constraint is established by using the terrain feature extraction results; After extracting the population and infrastructure data in the basic data set, a third segmentation constraint is established according to the population density features and the facility criticality features; The first segmentation constraint, the second segmentation constraint, and the third segmentation constraint are used to perform regional grid segmentation, and a focus grid segmentation result is established. 3.The storm surge intelligent early warning method based on multi-model coupling of claim 2, wherein, The use of the first segmentation constraint, the second segmentation constraint, and the third segmentation constraint to perform regional grid segmentation and establish a focus grid segmentation result includes: A historical disaster data set of a target region is obtained, data analysis of the historical disaster data set is performed, and disaster frequency features and disaster severity features are established; The fourth segmentation constraint is configured by using the disaster frequency features and the disaster severity features; After adding the fourth segmentation constraint to the first segmentation constraint, the second segmentation constraint, and the third segmentation constraint, regional grid segmentation is performed.

4. The storm surge intelligent early warning method based on multi-model coupling of claim 1, wherein, The preset safety response measures of the nodes in the focus grid segmentation result are obtained, joint early warning identification is performed according to the preset safety response measures and the node simulation data, and a warning signal is reported, including: The preset safety response measures are analyzed to obtain response levels, response action operations, and response time nodes; Based on the response levels, response action operations, and response time nodes, a safety response measure adaptation evaluation of the simulation data is performed, and an abnormal adaptation value is established; Joint early warning identification is completed according to the abnormal adaptation value, and a warning signal is reported.

5. The storm surge intelligent early warning method based on multi-model coupling according to claim 1, characterized in that, After the warning signal is reported, it includes: A verification time interval is set according to the warning level of the warning signal; The execution state of the real safety response measure is verified in the verification time interval, and a verification anomaly is established; A warning superposition analysis is performed according to the verification anomaly and the verification time interval, and the warning signal is reconstructed.

6. The storm surge intelligent early warning system based on multi-model coupling, characterized in that, The system for implementing the storm surge intelligent early warning method based on multi-model coupling of any one of claims 1-5, the system comprising: a basic data acquisition module: acquiring geographic spatial data and population and infrastructure data to establish a basic data set; a fuzzy path fitting module: obtaining real-time meteorological data and real-time ocean data, performing fuzzy path fitting according to the real-time meteorological data and the real-time ocean data, and establishing a fuzzy path fitting result; a regional grid segmentation module: using the basic data set and the fuzzy path fitting result to perform regional grid segmentation and establish a focus grid segmentation result; a simulation operation module: inputting the real-time meteorological data, the real-time ocean data, and the basic data set into a multi-model coupling simulation unit, performing multi-model coupling simulation unit simulation operation based on the focus grid segmentation result, and establishing node simulation data of the focus grid segmentation result; The joint early warning identification module obtains preset safety response measures of the nodes in the attention grid segmentation result, performs joint early warning identification according to the preset safety response measures and the node simulation data, and reports an early warning signal.

7. A computer readable medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the storm surge intelligent early warning method based on multi-model coupling in any one of claims 1-5.

Images