A long-term evolution prediction method and system for beach shorelines under the influence of typhoons

Abstract

The present invention application provides a long-term evolution prediction method and system for beach shorelines under the influence of typhoons. The method includes: obtaining a preset probability distribution model; obtaining historical typhoon data and generating multiple sets of typhoon scenario data; dividing an initial area into several grid areas; using the typhoon scenario data to conduct simulations in each grid area to obtain simulation results; extracting hydrodynamic field data of the beach shoreline area from the simulation results, extracting feature vectors and identifying target areas, and then using a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline. This application describes the intensity and movement path of typhoons through a probability distribution model, and then generates multiple sets of typhoon scenario data in combination with historical typhoon data and uses them for simulations in each grid area, which can take into account the differences at the spatial level between grid areas, that is, the influence of typhoons on different grids in the initial area, and effectively improve the evolution prediction effect of beach shorelines under the influence of typhoons.

Description

Technical Field

[0001] This invention application relates to the field of image analysis, and particularly to a method and system for predicting the long-term evolution of beach shorelines under the influence of typhoons. Background Art

[0002] On a time scale of years, beach shorelines are eroded and thus evolve under the influence of external factors such as typhoons and ocean waves. When studying the long-term evolution prediction of beach shorelines under the influence of typhoons, there are multiple difficulties. For example, a typhoon is a complex meteorological process, and its intensity, movement path, and duration all affect the evolution of beach shorelines; secondly, the ocean dynamic processes caused by typhoons are also very complex. For example, it may cause phenomena such as wind-driven currents, storm surges, and waves, and the differences in these ocean dynamic phenomena in terms of space-time scale are relatively significant; furthermore, beach shorelines are comprehensively affected by factors such as ocean currents and sediment movement, and these factors interact with each other. In the case where the feedback mechanism is not clear, it is difficult to quantitatively simulate the response of the shoreline to typhoons, resulting in certain uncertainties. Existing technologies mainly simulate the intensity of typhoons in a fixed area at the probability level when deducing and predicting the evolution of beach shorelines, while ignoring the change factors at the spatial level, such as the influence of typhoons on different areas of beach shorelines. Therefore, the prediction effect is generally average. Summary of the Invention

[0003] This invention application provides a method and system for predicting the long-term evolution of beach shorelines under the influence of typhoons to solve the technical problem of how to improve the prediction effect of the evolution of beach shorelines under the influence of typhoons.

[0004] To solve the above technical problem, this invention application provides a method for predicting the long-term evolution of beach shorelines under the influence of typhoons, including:

[0005] Obtain a preset probability distribution model, where the probability distribution model is used to describe the intensity and movement path of typhoons;

[0006] Obtain historical typhoon data, and generate multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model;

[0007] Adopt the nested grid technology to divide the initial area into several grid areas;

[0008] Use the multiple sets of typhoon scenario data to conduct simulations in each of the grid areas to obtain the simulation results of each grid data, where the simulation results include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data;

[0009] Extract the hydrodynamic field data of the beach shoreline area from the simulation results, extract the eigenvectors of the hydrodynamic field data through the principal component analysis method, identify the target area from the hydrodynamic field data according to the eigenvectors, and then use the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, the spatio-temporal distribution data of the waves in the target area and the preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline.

[0010] As a preferred solution, the step of using the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, the spatio-temporal distribution data of the waves in the target area and the preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline includes:

[0011] Calculate the frequency distribution of the beach shoreline according to the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, and the spatio-temporal distribution data of the waves in the target area;

[0012] Perform fitting processing on the frequency distribution to obtain a fitting model of the beach shoreline, and the fitting model is used to characterize the random process of the change of the beach shoreline;

[0013] Generate a number of random samples by the Monte Carlo method using the fitting model;

[0014] Generate at least one beach shoreline evolution path based on the random samples using the time scale.

[0015] As a preferred solution, the step of generating at least one beach shoreline evolution path based on the random samples using the time scale includes:

[0016] Calculate the position and range of the target area according to the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, and the spatio-temporal distribution data of the waves in the target area to obtain a boundary condition data set;

[0017] Input the boundary condition data set into a preset sediment movement model, calculate the sediment transport parameters and sedimentation process parameters of the target area to obtain the spatio-temporal variation characteristics of sediment, and use the spatio-temporal variation characteristics of sediment as the first shoreline response characteristic;

[0018] Obtain a second shoreline response characteristic based on the time scale and the random samples; wherein, the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic;

[0019] Generate at least one beach shoreline evolution path based on the first shoreline response characteristic and the second shoreline response characteristic.

[0020] As a preferred solution, obtaining historical typhoon data, and generating multiple sets of typhoon scenario data according to the historical typhoon data and probability distribution model, including:

[0021] Obtaining the historical typhoon data, where the historical typhoon data includes the position data, wind speed data, and air pressure data of the typhoon;

[0022] Through the probability distribution model, obtaining the correlation between the position data, wind speed data, and air pressure data, and then generating multiple sets of typhoon scenario data, where the typhoon scenario data includes parameter changes during the typhoon life cycle.

[0023] As a preferred solution, the typhoon scenario data includes the wind stress, pressure gradient force, and Coriolis force of the typhoon;

[0024] Using the multiple sets of typhoon scenario data to perform simulations in each of the grid regions to obtain simulation results of each grid data, including:

[0025] Using the multiple sets of typhoon scenario data to perform simulations in each of the grid regions, constructing a parameterized wind field model based on gradient wind balance and boundary layer theory, and obtaining wind field spatio-temporal distribution data;

[0026] Inputting the wind stress, pressure gradient force, and Coriolis force into a preset FVCOM model to simulate a sea current model in each of the grid regions, and then obtaining flow field spatio-temporal distribution data;

[0027] Inputting the multiple sets of typhoon scenario data into a preset SWAN model to obtain wave spatio-temporal distribution data;

[0028] Coupling the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data, and updating the coupling result at preset time intervals to obtain the simulation results of each grid data.

[0029] As a preferred solution, the long-term beach shoreline evolution prediction method further includes:

[0030] Constructing an evolution data set according to multiple beach shoreline evolution paths; dividing the target area into shoreline units, and respectively calculating the vulnerability of each shoreline unit according to the geological type, terrain slope, and vegetation cover of each shoreline unit;

[0031] Using the evolution data set to train the preset long short-term memory networks corresponding to each shoreline unit;

[0032] Using the vulnerability parameters, iteratively optimize the model structures and training parameters of each of the long short-term memory networks. When the performance parameters of the long short-term memory networks meet the preset convergence requirements, weight and fuse the long short-term memory networks that meet the convergence requirements according to the preset weights of each shoreline unit to obtain a shoreline prediction model.

[0033] Correspondingly, the present invention application also provides a long-term evolution prediction system for beach shorelines under the influence of typhoons, including an acquisition module, a generation module, a division module, a simulation module, and a prediction module; wherein,

[0034] The acquisition module is used to acquire a preset probability distribution model, and the probability distribution model is used to describe the intensity and movement path of typhoons;

[0035] The generation module is used to acquire historical typhoon data, and generate multiple groups of typhoon scenario data according to the historical typhoon data and the probability distribution model;

[0036] The division module is used to divide the initial area into several grid areas by using the nested grid technology;

[0037] The simulation module is used to perform simulations in each of the grid areas by using the multiple groups of typhoon scenario data, and obtain the simulation results of each grid data, and the simulation results include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data;

[0038] The prediction module is used to extract the hydrodynamic field data of the beach shoreline area from the simulation results, extract the eigenvectors of the hydrodynamic field data by using the principal component analysis method, identify the target area from the hydrodynamic field data according to the eigenvectors, and then use the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area and a preset time scale to simulate the change process of the beach shoreline and predict the beach shoreline evolution path.

[0039] As a preferred solution, the prediction module uses the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area and a preset time scale to simulate the change process of the beach shoreline and predict the beach shoreline evolution path, including:

[0040] The prediction module calculates the frequency distribution of the beach shoreline according to the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area;

[0041] Perform fitting processing on the frequency distribution to obtain a fitting model of the beach shoreline, and the fitting model is used to characterize the random process of the change of the beach shoreline;

[0042] Generate a number of random samples using the Monte Carlo method with the fitting model;

[0043] Generate at least one beach shoreline evolution path based on the random samples using the time scale.

[0044] As a preferred solution, the prediction module generates at least one beach shoreline evolution path based on the random samples using the time scale, including:

[0045] The prediction module calculates the position and range of the target area based on the spatio-temporal distribution data of the wind field, flow field, and wave field in the target area to obtain a boundary condition data set;

[0046] Input the boundary condition data set into a preset sediment movement model, calculate the sediment transport parameters and sedimentation process parameters of the target area to obtain the spatio-temporal variation characteristics of sediment, and use the spatio-temporal variation characteristics of sediment as the first shoreline response characteristic;

[0047] Based on the time scale and random samples, obtain a second shoreline response characteristic; wherein, the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic;

[0048] Generate at least one beach shoreline evolution path based on the first shoreline response characteristic and the second shoreline response characteristic.

[0049] As a preferred solution, the generation module obtains historical typhoon data and generates multiple sets of typhoon scenario data according to the historical typhoon data and probability distribution model, including:

[0050] The generation module obtains the historical typhoon data, and the historical typhoon data includes the position data, wind speed data, and air pressure data of the typhoon;

[0051] Through the probability distribution model, obtain the correlation between the position data, wind speed data, and air pressure data, and then generate multiple sets of typhoon scenario data, and the typhoon scenario data includes parameter changes during the typhoon life cycle.

[0052] As a preferred solution, the typhoon scenario data includes the wind stress, pressure gradient force, and Coriolis force of the typhoon;

[0053] The simulation module uses the multiple sets of typhoon scenario data to perform simulations in each grid area to obtain the simulation results of each grid data, including:

[0054] The simulation module uses the multiple sets of typhoon scenario data to perform simulations in each grid area, constructs a parameterized wind field model based on the gradient wind balance and boundary layer theory, and obtains the spatio-temporal distribution data of the wind field;

[0055] Input the wind stress, barometric gradient force, and Coriolis force into a preset FVCOM model, simulate a sea current model in each of the grid regions, and then obtain the spatio-temporal distribution data of the flow field;

[0056] Input the multiple sets of typhoon scenario data into a preset SWAN model to obtain the spatio-temporal distribution data of the waves;

[0057] Couple the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, and the spatio-temporal distribution data of the waves, and update the coupling result at preset time intervals to obtain the simulation results of the grid data.

[0058] As a preferred solution, the long-term evolution prediction system of the beach shoreline further includes a model construction module, and the model construction module is used for:

[0059] Construct an evolution data set according to multiple beach shoreline evolution paths; divide the shoreline units of the target area, and calculate the vulnerability of each shoreline unit respectively according to the geological type, terrain slope, and vegetation coverage of each shoreline unit;

[0060] Use the evolution data set to train the preset long short-term memory networks corresponding to each shoreline unit respectively;

[0061] Use the vulnerability parameters to iteratively optimize the model structure and training parameters of each long short-term memory network respectively. When the performance parameters of the long short-term memory network meet the preset convergence requirements, weight and fuse the long short-term memory networks that meet the convergence requirements according to the preset weights of each shoreline unit to obtain a shoreline prediction model.

[0062] Compared with the prior art, the present invention application has the following beneficial effects:

[0063] The present invention application provides a method and system for predicting the long-term evolution of a beach shoreline under the influence of typhoons. The method for predicting the long-term evolution of the beach shoreline includes: obtaining a preset probability distribution model, which is used to describe the intensity and movement path of typhoons; obtaining historical typhoon data, and generating multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model; using the nested grid technology to divide the initial area into several grid areas; using the multiple sets of typhoon scenario data to perform simulations in each of the grid areas to obtain the simulation results of each grid data, where the simulation results include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data; extracting the hydrodynamic field data of the beach shoreline area from the simulation results, and extracting the eigenvectors of the hydrodynamic field data through the principal component analysis method, identifying the target area from the hydrodynamic field data according to the eigenvectors, and then using the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline. The present invention application describes the intensity and movement path of typhoons through a probability distribution model, and then generates multiple sets of typhoon scenario data in combination with historical typhoon data and uses them for simulations in each grid area, which can take into account the spatial differences between different grid areas, that is, the influence of typhoons on different grids in the initial area, and effectively improve the prediction effect of the evolution of the beach shoreline under the influence of typhoons; in addition, the simulation results of the grid data in this application include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data. By extracting the hydrodynamic field data of the beach shoreline area, which may include the simulation results of one or more grids, the significance of the prediction method in this application at the spatial level is further improved; furthermore, the target area is identified by extracting the eigenvectors of the hydrodynamic field data through the principal component analysis method. Using the hydrodynamic field data and eigenvectors, the beach shoreline area and its spatio-temporal distribution are locked, so as to be used for simulating the evolution of the beach shoreline. Through the preset time period, multi-scale prediction of long-term and short-term can be effectively realized, ensuring the accuracy of the prediction method at different time scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 : A flowchart showing an embodiment of the method for predicting the long-term evolution of a beach shoreline under the influence of typhoons provided by the present invention application.

[0065] Figure 2 : A structural diagram showing an embodiment of the system for predicting the long-term evolution of a beach shoreline under the influence of typhoons provided by the present invention application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0067] Embodiment 1

[0068] Please refer to Figure 1 , Figure 1 , which is a long-term evolution prediction method for beach shorelines under the influence of typhoons provided by this application of the present invention, including steps S101 to S105; among them, each step is described in detail as follows:

[0069] Step S101, obtain a preset probability distribution model.

[0070] In this step, the probability distribution model is used to describe the intensity and movement path of typhoons. It can, for example, describe the correlation between parameters such as the position of the typhoon center, movement speed, central pressure, and maximum wind speed through a joint probability distribution, and is used to generate typhoon scenario data in subsequent steps.

[0071] Step S102, obtain historical typhoon data, and generate multiple groups of typhoon scenario data according to the historical typhoon data and the probability distribution model.

[0072] In this step, the obtaining of historical typhoon data and generating multiple groups of typhoon scenario data according to the historical typhoon data and the probability distribution model includes: obtaining the historical typhoon data, where the historical typhoon data includes position data, wind speed data, and pressure data of typhoons (such as the position of the typhoon center, movement speed, central pressure, and maximum wind speed, etc.); through the probability distribution model, obtain the correlation between the position data, wind speed data, and pressure data, and then generate multiple groups of typhoon scenario data, where the typhoon scenario data includes parameter changes during the typhoon life cycle.

[0073] Exemplarily, the above historical typhoon data can be, for example, the typhoon best track dataset every 6 hours in the western North Pacific from 1979 to 2022, including parameters such as the center position (latitude and longitude), maximum wind speed, and minimum pressure. Through its fitting to the probability distribution model, multiple groups of typhoon scenario data (such as 1000 groups) are further generated. Each group of typhoon scenario data contains parameter changes during the complete typhoon life cycle to ensure the integrity of the typhoon scenario data and facilitate the complete and accurate simulation of the typhoon scenario in step S104.

[0074] Step S103, adopt the nested grid technology to divide the initial area into several grid areas.

[0075] In this step, the nested grid technique can adopt a three-layer nested grid. Exemplarily, the outer grid covers the entire study sea area (initial area), and the grid resolution corresponds to approximately 9 km; the middle grid covers the main affected area, and the grid resolution corresponds to approximately 3 km; the inner grid covers the key nearshore area, and the grid resolution corresponds to approximately 1 km.

[0076] Moreover, data exchange and boundary condition transfer between different grids can be achieved through the bilinear interpolation method.

[0077] This embodiment can achieve different-scale divisions by dividing different grids, including dividing the initial area into an outer grid, a middle grid, and an inner grid. Further, the outer grid, the middle grid, and the inner grid are each further divided into multiple grids, which can realize different-scale divisions, that is, achieve adaptive scale changes for different sub-regions of the initial area, facilitating the accurate simulation of typhoon scenario data in each grid area in the subsequent step S104.

[0078] Step S104, using the multiple groups of typhoon scenario data to perform simulations in each of the grid areas to obtain simulation results of each grid data.

[0079] In this step, the simulation results include wind field spatio-temporal distribution data, current field spatio-temporal distribution data, and wave spatio-temporal distribution data.

[0080] Furthermore, the typhoon scenario data includes mechanical data of the typhoon, specifically including but not limited to wind stress, pressure gradient force, and Coriolis force of the typhoon, etc.

[0081] The using the multiple groups of typhoon scenario data to perform simulations in each of the grid areas to obtain simulation results of each grid data includes:

[0082] Using the above 1000 groups of typhoon scenario data to perform simulations in each of the grid areas, based on the gradient wind balance and boundary layer theory, constructing a parameterized wind field model to obtain wind field spatio-temporal distribution data (such as wind speed and wind direction at each grid point).

[0083] Inputting the wind stress, pressure gradient force, and Coriolis force into a preset FVCOM model (the FVCOM model (Finite Volume Coastal Ocean Model) is an ocean numerical model based on the finite volume method), simulating a sea current model in each of the grid areas, and then obtaining current field spatio-temporal distribution data.

[0084] Input the 1000 sets of typhoon scenario data into a preset SWAN model (the SWAN model (Simulating Waves Nearshore) is a numerical model for simulating nearshore wave propagation, refraction, shoaling, breaking, and nonlinear interactions), and obtain wave spatio-temporal distribution data (such as wave height, wave period, and wave direction at each grid point).

[0085] Couple the wind field spatio-temporal distribution data, the current field spatio-temporal distribution data, and the wave spatio-temporal distribution data (which can be achieved through a preset coupling module), and update the coupling result every preset time (for example, perform data exchange every hour to update the coupling result. Specifically, since the wind field drives ocean currents and waves, the ocean current affects the bottom flow velocity and then affects the wind stress, and the wave affects the ocean current and storm surge water level rise through radiation stress), and determine the latest coupling result as the simulation result of the grid data.

[0086] The simulation results can be used for statistical analysis to obtain distribution maps of extreme water levels and extreme wave heights in different grid regions, etc.

[0087] Step S105: Extract the hydrodynamic field data of the beach shoreline area from the simulation results, extract the eigenvectors of the hydrodynamic field data through the principal component analysis method, identify the target area from the hydrodynamic field data according to the eigenvectors, and then use the wind field spatio-temporal distribution data, current field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area, and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline.

[0088] In this step, extract the hydrodynamic field data of the beach shoreline area from the simulation results, and preprocess the hydrodynamic field data to remove noise and outliers to obtain standardized data.

[0089] Then, the standardized data can be reduced in dimension through the principal component analysis method to extract the main features of the hydrodynamic field, form the above eigenvectors, and identify the target area from the eigenvectors.

[0090] For example, identify the hydrodynamic field through a pre-trained support vector machine model to obtain the target area.

[0091] Furthermore, the use of the wind field spatio-temporal distribution data, current field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area, and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline includes:

[0092] Calculate the frequency distribution of the beach shoreline according to the wind field spatio-temporal distribution data, current field spatio-temporal distribution data, and wave spatio-temporal distribution data of the target area.

[0093] Perform a fitting process on the frequency distribution to obtain a fitting model of the beach shoreline, where the fitting model is used to characterize the random process of the beach shoreline change;

[0094] Generate a number of random samples by the Monte Carlo method using the fitting model;

[0095] Generate at least one beach shoreline evolution path based on the random samples using the time scale.

[0096] Exemplarily, considering that the shoreline change may be affected by various factors and is usually an asymmetric distribution, a lognormal distribution can be considered for the fitting process of the frequency distribution to better describe the skewed data and avoid the unreasonable situation of the shoreline change being negative infinity.

[0097] Based on the fitting situation, the maximum likelihood estimation method can be used, and statistical tools of MATLAB or Python can be used to estimate the lognormal distribution. Then, the Monte Carlo method is used to generate random samples, and at least one beach shoreline evolution path is generated based on the random samples using a preset time scale (generally a long-term time scale, such as fifty years). These paths reflect various possible situations of the shoreline evolution in the next 50 years under the current climate conditions and can be used in specific application scenarios such as risk assessment and planning design of coastal protection projects. For example, if a certain path shows that the shoreline erosion exceeds 30 meters after 50 years, it can be considered that there is a high erosion risk in this area and corresponding protection measures need to be taken.

[0098] To further improve the prediction accuracy, the short-term scale evolution situation can be further considered on the basis of considering the long-term time scale. Generally, the short-term time scale evolution is related to sediment transport and deposition.

[0099] For example, using the time scale, at least one beach shoreline evolution path is generated based on the random samples, including: calculating the position and range of the target area according to the spatio-temporal distribution data of the wind field, flow field, and wave field in the target area to obtain a boundary condition data set; inputting the boundary condition data set into a preset sediment movement model to calculate the sediment transport parameters and deposition process parameters in the target area, obtaining the spatio-temporal variation characteristics of sediment, and using the spatio-temporal variation characteristics of sediment as the first shoreline response characteristic; obtaining a second shoreline response characteristic based on the time scale and random samples; wherein, the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic; based on the first shoreline response characteristic on a short-term time scale and the second shoreline response characteristic on a long-term time scale, at least one beach shoreline evolution path can be accurately generated to consider the complex non-linear relationship between dynamic processes and shoreline morphology at various scales, and to construct a unified prediction model framework from short-term event scale to long-term scale.

[0100] In a preferred embodiment, in order to facilitate the invocation of the results of the long-term evolution prediction method of the beach shoreline under typhoon influence in this application to ensure that the accuracy of the prediction method in this application can be transferred to other application scenarios, an artificial intelligence large model can be trained using the prediction results of the prediction method in this application.

[0101] Exemplarily, using the long short-term memory network LSTM as the basic model, the long-term evolution prediction method of the beach shoreline further includes:

[0102] Constructing an evolution data set according to multiple beach shoreline evolution paths; dividing the shoreline units of the target area, and respectively calculating the vulnerability of each shoreline unit according to the geological type, terrain slope, and vegetation cover of each shoreline unit;

[0103] Training the preset long short-term memory network corresponding to each shoreline unit using the evolution data set; evaluating the training using the vulnerability parameter, iteratively optimizing the model structure and training parameters of each long short-term memory network respectively, and when the performance parameters of the long short-term memory network meet the preset convergence requirements, performing weighted fusion on the long short-term memory networks that meet the convergence requirements according to the preset weights of each shoreline unit to obtain a shoreline prediction model.

[0104] This preferred embodiment constructs an evolution data set and uses the shoreline prediction model to learn the beach shoreline evolution law contained in the evolution data set, extracts the prediction features of the prediction method in this application, which can be called by other application scenarios for the shoreline prediction model, and transfers the accuracy of the prediction method in this application to other sea areas or other prediction scenarios to realize the further utilization of the prediction results in this application.

[0105] Correspondingly, as Figure 2 shown, the present invention application also provides a long-term evolution prediction system 200 for a beach shoreline under the influence of typhoons, including an acquisition module 201, a generation module 202, a division module 203, a simulation module 204, and a prediction module 205; wherein,

[0106] the acquisition module 201 is configured to acquire a preset probability distribution model, and the probability distribution model is used to describe the intensity and movement path of typhoons;

[0107] the generation module 202 is configured to acquire historical typhoon data, and generate multiple groups of typhoon scenario data according to the historical typhoon data and the probability distribution model;

[0108] the division module 203 is configured to divide an initial area into several grid areas by using a nested grid technology;

[0109] the simulation module 204 is configured to perform simulations in each of the grid areas by using the multiple groups of typhoon scenario data to obtain simulation results of each grid data, and the simulation results include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data;

[0110] the prediction module 205 is configured to extract hydrodynamic field data of the beach shoreline area from the simulation results, extract eigenvectors of the hydrodynamic field data by using a principal component analysis method, identify a target area from the hydrodynamic field data according to the eigenvectors, and then use the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area, and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline.

[0111] As a preferred solution, the prediction module 205 uses the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area, and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline, including:

[0112] the prediction module 205 calculates the frequency distribution of the beach shoreline according to the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area;

[0113] performs fitting processing on the frequency distribution to obtain a fitting model of the beach shoreline, and the fitting model is used to characterize the random process of the change of the beach shoreline;

[0114] generates a number of random samples by using the Monte Carlo method and the fitting model;

[0115] generates at least one beach shoreline evolution path based on the random samples by using the time scale.

[0116] As a preferred solution, the prediction module 205 uses the time scale to generate at least one beach shoreline evolution path based on the random samples, including:

[0117] The prediction module 205 calculates the position and range of the target area according to the spatio-temporal distribution data of the wind field, the spatio-temporal distribution data of the flow field, and the spatio-temporal distribution data of the waves in the target area, and obtains a boundary condition data set;

[0118] Input the boundary condition data set into a preset sediment movement model, calculate the sediment transport parameters and sedimentation process parameters of the target area, obtain the spatio-temporal variation characteristics of the sediment, and use the spatio-temporal variation characteristics of the sediment as the first shoreline response characteristic;

[0119] Based on the time scale and the random samples, obtain a second shoreline response characteristic; wherein, the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic;

[0120] Generate at least one beach shoreline evolution path based on the first shoreline response characteristic and the second shoreline response characteristic.

[0121] As a preferred solution, the generation module 202 obtains historical typhoon data and generates multiple groups of typhoon scenario data according to the historical typhoon data and the probability distribution model, including:

[0122] The generation module 202 obtains the historical typhoon data, and the historical typhoon data includes the position data, wind speed data, and air pressure data of the typhoon;

[0123] Through the probability distribution model, obtain the correlation between the position data, wind speed data, and air pressure data, and then generate multiple groups of typhoon scenario data, and the typhoon scenario data includes the parameter changes during the typhoon life cycle.

[0124] As a preferred solution, the typhoon scenario data includes the wind stress, air pressure gradient force, and Coriolis force of the typhoon;

[0125] The simulation module 204 uses the multiple groups of typhoon scenario data to perform simulations in each of the grid regions, and obtains the simulation results of each grid data, including:

[0126] The simulation module 204 uses the multiple groups of typhoon scenario data to perform simulations in each of the grid regions, constructs a parameterized wind field model based on the gradient wind balance and boundary layer theory, and obtains the spatio-temporal distribution data of the wind field;

[0127] Input the wind stress, air pressure gradient force, and Coriolis force into a preset FVCOM model, simulate a sea current model in each of the grid regions, and then obtain the spatio-temporal distribution data of the flow field;

[0128] Input the multi - group typhoon scenario data into a preset SWAN model to obtain wave spatio - temporal distribution data;

[0129] Couple the wind field spatio - temporal distribution data, the flow field spatio - temporal distribution data and the wave spatio - temporal distribution data, and update the coupling result every preset time to obtain the simulation results of the grid data.

[0130] As a preferred solution, the long - term beach shoreline evolution prediction system 200 further includes a model construction module, and the model construction module is used for:

[0131] Construct an evolution data set according to multiple beach shoreline evolution paths; divide the shoreline units of the target area, and calculate the vulnerability of each shoreline unit respectively according to the geological type, terrain slope and vegetation cover of each shoreline unit;

[0132] Use the evolution data set to train the preset long - short - term memory networks corresponding to each shoreline unit respectively;

[0133] Use the vulnerability parameters to iteratively optimize the model structure and training parameters of each long - short - term memory network. When the performance parameters of the long - short - term memory network meet the preset convergence requirements, weight - fuse the long - short - term memory networks that meet the convergence requirements according to the preset weights of each shoreline unit to obtain a shoreline prediction model.

[0134] Compared with the prior art, the present invention application has the following beneficial effects:

[0135] The present invention application provides a method and system for predicting the long-term evolution of a beach shoreline under the influence of typhoons. The method for predicting the long-term evolution of the beach shoreline includes: obtaining a preset probability distribution model, which is used to describe the intensity and movement path of typhoons; obtaining historical typhoon data, and generating multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model; using the nested grid technology to divide the initial area into several grid areas; using the multiple sets of typhoon scenario data to perform simulations in each of the grid areas to obtain the simulation results of each grid data, where the simulation results include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data; extracting the hydrodynamic field data of the beach shoreline area from the simulation results, and extracting the eigenvectors of the hydrodynamic field data through the principal component analysis method, identifying the target area from the hydrodynamic field data according to the eigenvectors, and then using the wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, wave spatio-temporal distribution data of the target area and a preset time scale to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline. The present invention application describes the intensity and movement path of typhoons through a probability distribution model, and then combines historical typhoon data to generate multiple sets of typhoon scenario data and uses them for simulations in each grid area, which can take into account the spatial differences between different grid areas, that is, the influence of typhoons on different grids in the initial area, and effectively improve the prediction effect of the evolution of the beach shoreline under the influence of typhoons; in addition, the simulation results of the grid data in this application include wind field spatio-temporal distribution data, flow field spatio-temporal distribution data, and wave spatio-temporal distribution data. By extracting the hydrodynamic field data of the beach shoreline area, the beach shoreline area may include the simulation results of one or more grids, further improving the significance of the prediction method in this application at the spatial level; furthermore, the eigenvectors of the hydrodynamic field data are extracted through the principal component analysis method to identify the target area. Using the hydrodynamic field data and the eigenvectors, the beach shoreline area and its spatio-temporal distribution are locked, so as to be used for simulating the evolution of the beach shoreline. Through a preset time period, multi-scale predictions of long-term and short-term can be effectively realized, ensuring the accuracy of the prediction method at different time scales.

[0136] In the specific embodiments described above, the purpose, technical solutions, and beneficial effects of the present invention have been further described in detail. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. In particular, it is pointed out that for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the long-term evolution of beach shoreline under the influence of typhoon, characterized in that: include: Obtaining a preset probability distribution model, where the probability distribution model is used to describe the intensity and movement path of a typhoon; Acquire historical typhoon data, and generate multiple sets of typhoon scenario data based on the historical typhoon data and a probability distribution model; The nested grid technique is used to divide the initial area into several grid areas; Using the multiple sets of typhoon scenario data to perform simulation in each of the grid areas, obtaining simulation results of each grid data, wherein the simulation results include wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, and wave spatiotemporal distribution data; Extracting the hydrodynamic field data of the beach shoreline area from the simulation results, and extracting the characteristic vector of the hydrodynamic field data by principal component analysis, identifying the target area from the hydrodynamic field data according to the characteristic vector, and then using the wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, wave spatiotemporal distribution data and a preset time scale of the target area to simulate the change process of the beach shoreline and predict the evolution path of the beach shoreline; The method utilizes the wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, wave spatiotemporal distribution data and a preset time scale of the target area to simulate the change process of the beach shoreline and predict the beach shoreline evolution path, including: Calculate the frequency distribution of the beach coastline according to the wind field spatiotemporal distribution data, the flow field spatiotemporal distribution data, and the wave spatiotemporal distribution data of the target area; Performing fitting processing on the frequency distribution to obtain a fitting model of the beach coastline, wherein the fitting model is used to characterize the random process of the beach coastline change; Generate a number of random samples using the fitted model through a Monte Carlo method; At least one beach shoreline evolution path is generated based on the random sample using the time scale.

2. The method for predicting the long-term evolution of beach shoreline under the influence of typhoon as claimed in claim 1, characterized in that: The step of utilizing the time scale to generate at least one beach shoreline evolution path based on the random sample comprises: Calculate the position and range of the target area according to the wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, and wave spatiotemporal distribution data of the target area to obtain a boundary condition data set; Inputting the boundary condition data set into a preset sediment movement model, calculating the sediment transport parameters and deposition process parameters of the target area, obtaining the sediment temporal and spatial variation characteristics, and using the sediment temporal and spatial variation characteristics as the first shoreline response characteristics; Based on the time scale and the random sample, a second shoreline response characteristic is obtained; wherein the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic; At least one beach shoreline evolution path is generated based on the first shoreline response characteristic and the second shoreline response characteristic.

3. The method for predicting the long-term evolution of beach shoreline under the influence of typhoon as claimed in claim 1, characterized in that: The acquiring of historical typhoon data and generating multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model include: Acquire the historical typhoon data, wherein the historical typhoon data includes typhoon location data, wind speed data, and air pressure data; The probability distribution model is used to obtain the correlation between the position data, wind speed data and air pressure data, thereby generating multiple sets of typhoon scenario data, wherein the typhoon scenario data includes parameter changes within the life cycle of a typhoon.

4. The method for predicting the long-term evolution of beach shoreline under the influence of typhoon as claimed in claim 1, characterized in that: The typhoon scenario data include the wind stress, pressure gradient force and Coriolis force of the typhoon; The using the multiple groups of typhoon scenario data to simulate in each of the grid areas to obtain simulation results of each grid data includes: Using the multiple sets of typhoon scenario data to perform simulations in each of the grid areas, constructing a parameterized wind field model based on gradient wind balance and boundary layer theory, and obtaining wind field spatiotemporal distribution data; Input the wind stress, pressure gradient force and Coriolis force into a preset FVCOM model, simulate the ocean current model in each grid area, and then obtain the spatiotemporal distribution data of the flow field; Inputting the plurality of sets of typhoon scenario data into a preset SWAN model to obtain wave spatiotemporal distribution data; The wind field spatiotemporal distribution data, the flow field spatiotemporal distribution data and the wave spatiotemporal distribution data are coupled, and the coupling results are updated at preset time intervals to obtain simulation results of the grid data.

5. A long-term beach shoreline evolution prediction system under the influence of typhoons, characterized in that: It includes acquisition module, generation module, division module, simulation module and prediction module; among them, The acquisition module is used to acquire a preset probability distribution model, where the probability distribution model is used to describe the intensity and movement path of the typhoon; The generating module is used to obtain historical typhoon data and generate multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model; The division module is used to divide the initial area into a number of grid areas by using a nested grid technology; The simulation module is used to use the multiple groups of typhoon scenario data to perform simulation in each of the grid areas to obtain simulation results of each grid data, wherein the simulation results include wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, and wave spatiotemporal distribution data; The prediction module is used to extract the hydrodynamic field data of the beach coastline area from the simulation results, and extract the characteristic vector of the hydrodynamic field data by principal component analysis, identify the target area from the hydrodynamic field data according to the characteristic vector, and then use the wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, wave spatiotemporal distribution data and a preset time scale of the target area to simulate the change process of the beach coastline and predict the evolution path of the beach coastline; The prediction module uses the wind field spatiotemporal distribution data, flow field spatiotemporal distribution data, wave spatiotemporal distribution data and a preset time scale of the target area to simulate the change process of the beach coastline and predict the beach coastline evolution path, including: The prediction module calculates the frequency distribution of the beach shoreline according to the wind field spatiotemporal distribution data, the flow field spatiotemporal distribution data, and the wave spatiotemporal distribution data of the target area; Performing fitting processing on the frequency distribution to obtain a fitting model of the beach coastline, wherein the fitting model is used to characterize the random process of the beach coastline change; Generate a number of random samples using the fitted model through a Monte Carlo method; At least one beach shoreline evolution path is generated based on the random sample using the time scale.

6. The long-term evolution prediction system of beach shoreline under typhoon influence as claimed in claim 5, characterized in that: The prediction module generates at least one beach shoreline evolution path based on the random sample using the time scale, including: The prediction module calculates the position and range of the target area according to the wind field spatiotemporal distribution data, the flow field spatiotemporal distribution data, and the wave spatiotemporal distribution data of the target area to obtain a boundary condition data set; Inputting the boundary condition data set into a preset sediment movement model, calculating the sediment transport parameters and deposition process parameters of the target area, obtaining the sediment temporal and spatial variation characteristics, and using the sediment temporal and spatial variation characteristics as the first shoreline response characteristics; Based on the time scale and the random sample, a second shoreline response characteristic is obtained; wherein the time scale corresponding to the second shoreline response characteristic is greater than that of the first shoreline response characteristic; At least one beach shoreline evolution path is generated based on the first shoreline response characteristic and the second shoreline response characteristic.

7. The long-term evolution prediction system of beach shoreline under typhoon influence as claimed in claim 5, characterized in that: The generation module acquires historical typhoon data, and generates multiple sets of typhoon scenario data according to the historical typhoon data and the probability distribution model, including: The generating module acquires the historical typhoon data, wherein the historical typhoon data includes the location data, wind speed data and air pressure data of the typhoon; The probability distribution model is used to obtain the correlation between the position data, wind speed data and air pressure data, thereby generating multiple sets of typhoon scenario data, wherein the typhoon scenario data includes parameter changes within the life cycle of a typhoon.

8. The long-term evolution prediction system of beach shoreline under typhoon influence as claimed in claim 5, characterized in that: The typhoon scenario data include the wind stress, pressure gradient force and Coriolis force of the typhoon; The simulation module uses the multiple sets of typhoon scenario data to perform simulation in each of the grid areas to obtain simulation results of each grid data, including: The simulation module uses the multiple sets of typhoon scenario data to perform simulations in each of the grid areas, constructs a parameterized wind field model based on gradient wind balance and boundary layer theory, and obtains wind field spatiotemporal distribution data; Input the wind stress, pressure gradient force and Coriolis force into a preset FVCOM model, simulate the ocean current model in each grid area, and then obtain the spatiotemporal distribution data of the flow field; Inputting the plurality of sets of typhoon scenario data into a preset SWAN model to obtain wave spatiotemporal distribution data; The wind field spatiotemporal distribution data, the flow field spatiotemporal distribution data and the wave spatiotemporal distribution data are coupled, and the coupling results are updated at preset time intervals to obtain simulation results of the grid data.

Images