Three-dimensional island terrain maximum wave height spatial distribution forecasting method and system

Abstract

The invention belongs to the technical field of ocean hydrodynamics, and discloses a three-dimensional island topography maximum wave height spatial distribution forecasting method and system, and the method comprises the steps: calculating the average wave height of each measurement point according to the wave surface change of each measurement point of the island topography, obtaining the three-dimensional island topography maximum wave height spatial distribution, constructing a random forest model, and calculating the maximum wave height spatial distribution of the three-dimensional island topography; optimizing the random forest model according to the three-dimensional island terrain maximum wave height spatial distribution data; according to a control variable method, separately changing incident wave height, water depth and wave direction data of the open sea, and establishing a corresponding wave height spatial distribution map and a wave height spatial distribution map predicted by a random forest model; and comparing the wave height spatial distribution diagrams to obtain data characteristics which have relatively great influence on the maximum wave height spatial distribution, thereby obtaining the maximum wave height spatial distribution. The method realizes high-precision simulation and prediction of propagation and evolution of waves on the three-dimensional island terrain, and has the advantages of high precision, high flexibility, wide application prospect and the like.

Description

Technical Field

[0001] The invention belongs to the technical field of ocean hydrodynamics and relates to a method and system for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain. Background Art

[0002] With the development of economy and society and the need of resource development, the construction of buildings on islands and reefs is increasing. Rational development and utilization of marine resources and scientific protection of marine environment are important foundations for realizing my country's sustainable economic development strategy. Therefore, it is of great practical significance to study the wave propagation characteristics on islands and reefs, especially the propagation characteristics of three-dimensional waves.

[0003] At present, there have been many studies on the propagation characteristics of waves on island and reef terrain under two-dimensional conditions. However, the actual waves are composed of waves of different directions and frequencies. Therefore, there are still relatively few studies on the propagation characteristics of three-dimensional waves on island and reef terrain. Existing research methods include physical model tests and numerical simulations. Among them, physical model tests can intuitively simulate the propagation process of waves on island and reef terrain, but the cost is high and time-consuming. Although numerical simulation methods are widely used in the study of three-dimensional wave propagation characteristics, the accuracy and reliability of the model still need to be continuously verified and optimized. Especially for complex island and reef terrain and extreme wave conditions, the applicability and accuracy of the model need to be further studied and improved. At the same time, the forecasting method for the spatial distribution of the maximum wave height of three-dimensional island and reef terrain is still imperfect; the lack of systematic forecasting methods and standards makes it difficult to guarantee the accuracy and reliability of the forecast results. Summary of the invention

[0004] The purpose of the present invention is to solve the problems that the accuracy and reliability of numerical simulation methods in the prior art still need to be continuously verified and optimized, and the forecasting method for the spatial distribution of maximum wave heights in three-dimensional island and reef terrain is still imperfect; there is a lack of systematic forecasting methods and standards, which makes it difficult to ensure the accuracy and reliability of forecast results. A method and system for forecasting the spatial distribution of maximum wave heights in three-dimensional island and reef terrain are provided.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for predicting the spatial distribution of maximum wave height in three-dimensional island and reef terrain, comprising:

[0007] Step 1, extracting the topographic characteristics of the three-dimensional island and reef terrain and collecting wave data in the open sea;

[0008] Step 2: Based on the characteristics of offshore waves, the propagation and evolution of waves in three-dimensional island and reef terrain are simulated according to a fully nonlinear wave numerical model;

[0009] Step 3, according to the wave surface changes at each measuring point of the island and reef terrain, the average wave height of each measuring point is calculated according to the upper crossing zero point method, and the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain is obtained;

[0010] Step 4, based on the control variable method, respectively change the incident wave height, water depth and wave direction of the open sea, and repeat steps 2 to 3 based on the changed data, so as to establish a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea;

[0011] Step 5, construct a random forest model, and optimize the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain the optimal random forest model;

[0012] Step 6, input the modified offshore incident wave height, water depth and wave direction data into the random forest model to obtain a spatial distribution map of wave height predicted by the random forest model;

[0013] Step 7, compare the wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyze the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtain the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtain the maximum wave height spatial distribution.

[0014] A further improvement of the present invention is:

[0015] Furthermore, wave data are collected in the open sea, including: several groups of water depths at different depths, the submerged water depth of the corresponding reef flats, the wave direction and incident wave height.

[0016] Furthermore, according to the wave surface changes at each measuring point of the island and reef terrain, the average wave height of each measuring point is calculated according to the upper crossing zero point method, and the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain is obtained, specifically: the wave surface data of each measuring point is preprocessed to remove outliers; and the position of the wave surface data crossing the zero point from positive to negative is found; for each upper crossing zero point, the height from the previous wave trough to the next wave peak is calculated as the wave height; the wave heights of all measuring points are averaged to obtain the average wave height of the measuring point; if the number of measuring points is limited, the average wave heights of each measuring point are interpolated to obtain the spatial distribution of the average wave height of the entire island and reef terrain; in the spatial distribution obtained by interpolation, the maximum wave height value of each point is extracted to obtain the spatial distribution of the maximum wave height.

[0017] Furthermore, based on the control variable method, the incident wave height, water depth and wave direction of the open sea are changed respectively, specifically: the water depth and wave direction are controlled to remain unchanged, only the incident wave height is changed, and the changed incident wave height value does not appear in the wave data collected in the open sea; the wave direction and incident wave height are controlled to remain unchanged, only the water depth is changed, and the changed water depth value does not appear in the wave data collected in the open sea; the water depth and incident wave height are controlled to remain unchanged, only the wave direction is changed, and the changed wave direction value does not appear in the wave data collected in the open sea.

[0018] Furthermore, a wave height spatial distribution map drawn by numerical simulation including the changed offshore incident wave height, water depth and wave direction is established, specifically:

[0019] The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the fully nonlinear wave numerical model, and the simulation data is compared to obtain the maximum wave height value of each measuring point in each group of examples in the simulation data, and the maximum wave height value is the maximum wave peak value, and the wave height spatial distribution diagram after the incident wave height data is changed is constructed;

[0020] Repeat the above steps, input the water depth data after the wave direction and incident wave height remain unchanged and the water depth and incident wave height remain unchanged but the wave direction data is changed into the fully nonlinear wave numerical model; respectively construct the spatial distribution diagram of the wave height after the water depth data is changed and the spatial distribution diagram of the wave height after the wave direction data is changed.

[0021] Furthermore, the random forest model is optimized according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain the optimal random forest model, which is as follows:

[0022] The spatial distribution data of the maximum wave height of three-dimensional island and reef terrain are divided into training sets and test sets; a suitable machine learning algorithm is selected as the benchmark model, each candidate feature set in the training set is trained based on the benchmark model, and the performance of the benchmark model is evaluated; based on the evaluation results, the feature set with the best performance is selected; on the selected feature set, the hyperparameters of the random forest model are optimized based on grid search;

[0023] The training set is randomly divided into several subsets, and the random forest model is trained on the divided subsets through the five-fold cross-validation method; the performance of the model is evaluated based on the root mean square error (RMSE) as the objective function; the hyperparameter combination is continuously adjusted to obtain the optimal hyperparameter combination that minimizes the RMSE;

[0024] Substitute the samples in the test set into the trained random forest model, and test the accuracy of the model by comparing the predicted results with the actual values ​​in the test set; based on the test results, optimize and adjust the random forest model until the random forest model with the best performance is found.

[0025] Furthermore, the modified offshore incident wave height, water depth and wave direction data were input into the optimal random forest model to obtain the spatial distribution map of wave height predicted by the random forest model, specifically:

[0026] The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the optimal random forest model to obtain the spatial distribution map of wave height predicted by the random forest model of the changed incident wave height data;

[0027] Repeat the above steps and input the water depth data after the wave direction and incident wave height remain unchanged and the water depth and incident wave height remain unchanged but the wave direction data is changed into the fully nonlinear wave numerical model; respectively construct the spatial distribution map of wave height predicted by the random forest model after the water depth data is changed and the spatial distribution map of wave height predicted by the random forest model after the wave direction data is changed.

[0028] Furthermore, the wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction is compared with the wave height spatial distribution map predicted by the random forest model. The prediction effect and the error spatial distribution are analyzed based on the relative error spatial distribution map, and the data characteristics that have a greater impact on the spatial distribution of the maximum wave height are obtained, and then the maximum wave height spatial distribution is obtained, which is specifically:

[0029] The wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction and the wave height spatial distribution map predicted by the random forest model are compared to determine whether the relative error of the wave height exceeds the preset threshold. If not, it is determined that the offshore wave data corresponding to the wave height relative error closest to the threshold has the greatest influence on the spatial distribution of the maximum wave height, and the wave height spatial distribution predicted by the random forest model for the offshore wave data is the maximum wave height spatial distribution; if it exceeds, the random forest model is continuously optimized until the maximum wave height spatial distribution is obtained.

[0030] A three-dimensional island and reef terrain maximum wave height spatial distribution prediction system, comprising:

[0031] A collection module, which extracts the topographic characteristics of the three-dimensional island and reef terrain and collects wave data in the open sea;

[0032] A simulation module, wherein the simulation module simulates the propagation evolution of waves in three-dimensional island and reef terrain according to a fully nonlinear wave numerical model based on the characteristics of offshore waves;

[0033] A calculation module, wherein the calculation module calculates the average wave height of each measuring point according to the wave surface change of each measuring point of the island and reef terrain according to the upper crossing zero point method, and obtains the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain;

[0034] A construction module, wherein the construction module changes the incident wave height, water depth and wave direction of the open sea respectively based on the control variable method, and establishes a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea;

[0035] An optimization module, wherein the optimization module constructs a random forest model and optimizes the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain an optimal random forest model;

[0036] An acquisition module, wherein the acquisition module inputs the modified offshore incident wave height, water depth and wave direction data into a random forest model to obtain a wave height spatial distribution map predicted by the random forest model;

[0037] A comparison module compares the wave height spatial distribution map drawn by the numerical simulation of the changed open sea incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyzes the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtains the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtains the maximum wave height spatial distribution.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention simulates the propagation and evolution of waves in three-dimensional island and reef terrain through offshore wave data and a completely nonlinear wave numerical model, and optimizes the constructed random deep forest model through the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain; according to the control variable method, the offshore incident wave height, water depth and wave direction data are changed respectively, and a wave height spatial distribution map drawn by numerical simulation of the changed offshore incident wave height, water depth and wave direction, and a wave height spatial distribution map predicted by the random forest model are established; and the above wave height spatial distribution map is compared, and the prediction effect and error spatial distribution are analyzed based on the relative error spatial distribution map, and the data characteristics that have a greater impact on the maximum wave height spatial distribution are obtained, and then the maximum wave height spatial distribution is obtained. The present invention realizes high-precision simulation and prediction of wave propagation evolution on three-dimensional island and reef terrain by integrating numerical simulation technology and random forest model, and has the advantages of high precision, flexibility, high efficiency, data-driven optimization and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0041] Figure 1It is a schematic flow chart of the method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain of the present invention;

[0042] Figure 2 It is a structural schematic diagram of the spatial distribution forecasting system of the maximum wave height of three-dimensional island and reef terrain of the present invention;

[0043] Figure 3 This is a schematic diagram of the topographic structure of three-dimensional islands and reefs;

[0044] Figure 4 The data diagram of each measuring point in the example is shown in Figure 2.

[0045] Figure 5 Schematic diagram for building a random forest model;

[0046] Figure 6 This is a schematic diagram of the comparison before and after the hyperparameters of the random forest model are tuned;

[0047] Figure 7 The spatial distribution diagram of wave height of numerical simulation results;

[0048] Figure 8 Spatial distribution map of wave height based on random forest prediction results;

[0049] Fig. 9 It is the spatial distribution map of wave height of numerical simulation results and the spatial distribution map of relative error of wave height based on random forest prediction results. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0051] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0053] In the description of the embodiments of the present invention, it should be noted that if the terms "upper", "lower", "horizontal", "inner", etc. indicate an orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the invention is usually placed when in use, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0054] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", which does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0055] In the description of the embodiments of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0056] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0057] See also Figure 1 The present invention discloses a method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain, comprising:

[0058] S101, extracting the topographic characteristics of the three-dimensional island and reef terrain, and collecting wave data in the open sea;

[0059] Collect wave data in the open sea, including: several groups of water depths at different depths, the corresponding submerged water depth of the reef flat, the wave direction and incident wave height.

[0060] S102, based on the characteristics of offshore waves, simulates the propagation and evolution of waves on three-dimensional island and reef terrain according to a fully nonlinear wave numerical model;

[0061] S103, according to the wave surface changes at each measuring point of the island and reef terrain, the average wave height of each measuring point is calculated according to the upper crossing zero point method, and the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain is obtained;

[0062] Preprocess the wave surface data of each measuring point to remove outliers; find the position of the wave surface data crossing the zero point from positive to negative; for each zero-crossing point, calculate the height from the previous trough to the next crest as the wave height; average the wave heights of all measuring points to obtain the average wave height of the measuring point; if the number of measuring points is limited, interpolate the average wave heights of each measuring point to obtain the average wave height spatial distribution of the entire island and reef terrain; in the spatial distribution obtained by interpolation, extract the maximum wave height value of each point to obtain the maximum wave height spatial distribution.

[0063] S104, based on the control variable method, respectively changing the incident wave height, water depth and wave direction of the open sea, and repeating S102 to S103 based on the changed data, so as to establish a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea;

[0064] Based on the control variable method, the incident wave height, water depth and wave direction of the open sea are changed respectively, specifically: the water depth and wave direction are controlled to remain unchanged, only the incident wave height is changed, and the changed incident wave height value does not appear in the wave data collected in the open sea; the wave direction and incident wave height are controlled to remain unchanged, only the water depth is changed, and the changed water depth value does not appear in the wave data collected in the open sea; the water depth and incident wave height are controlled to remain unchanged, only the wave direction is changed, and the changed wave direction value does not appear in the wave data collected in the open sea.

[0065] A wave height spatial distribution map drawn by numerical simulation including the changed offshore incident wave height, water depth and wave direction is established, specifically:

[0066] The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the fully nonlinear wave numerical model, and the simulation data is compared to obtain the maximum wave height value of each measuring point in each group of examples in the simulation data, and the maximum wave height value is the maximum wave peak value, and the wave height spatial distribution diagram after the incident wave height data is changed is constructed;

[0067] Repeat the above steps, input the water depth data after the wave direction and incident wave height remain unchanged and the water depth and incident wave height remain unchanged but the wave direction data is changed into the fully nonlinear wave numerical model; respectively construct the spatial distribution diagram of the wave height after the water depth data is changed and the spatial distribution diagram of the wave height after the wave direction data is changed.

[0068] S105, constructing a random forest model, and optimizing the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain the optimal random forest model;

[0069] The spatial distribution data of the maximum wave height of three-dimensional island and reef terrain are divided into training sets and test sets; a suitable machine learning algorithm is selected as the benchmark model, each candidate feature set in the training set is trained based on the benchmark model, and the performance of the benchmark model is evaluated; based on the evaluation results, the feature set with the best performance is selected; on the selected feature set, the hyperparameters of the random forest model are optimized based on grid search;

[0070] The training set is randomly divided into several subsets, and the random forest model is trained on the divided subsets through the five-fold cross-validation method; the performance of the model is evaluated based on the root mean square error (RMSE) as the objective function; the hyperparameter combination is continuously adjusted to obtain the optimal hyperparameter combination that minimizes the RMSE;

[0071] Substitute the samples in the test set into the trained random forest model, and test the accuracy of the model by comparing the predicted results with the actual values ​​in the test set; based on the test results, optimize and adjust the random forest model until the random forest model with the best performance is found.

[0072] S106, inputting the modified offshore incident wave height, water depth and wave direction data into the random forest model to obtain a wave height spatial distribution map predicted by the random forest model;

[0073] The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the optimal random forest model to obtain the spatial distribution map of wave height predicted by the random forest model of the changed incident wave height data;

[0074] Repeat the above steps and input the water depth data after the wave direction and incident wave height remain unchanged and the water depth and incident wave height remain unchanged but the wave direction data is changed into the fully nonlinear wave numerical model; respectively construct the spatial distribution map of wave height predicted by the random forest model after the water depth data is changed and the spatial distribution map of wave height predicted by the random forest model after the wave direction data is changed.

[0075] S107, compare the wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyze the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtain the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtain the maximum wave height spatial distribution.

[0076] The wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction and the wave height spatial distribution map predicted by the random forest model are compared to determine whether the relative error of the wave height exceeds the preset threshold. If not, it is determined that the offshore wave data corresponding to the wave height relative error closest to the threshold has the greatest influence on the spatial distribution of the maximum wave height, and the wave height spatial distribution predicted by the random forest model for the offshore wave data is the maximum wave height spatial distribution; if it exceeds, the random forest model is continuously optimized until the maximum wave height spatial distribution is obtained.

[0077] See also Figure 2 The present invention discloses a three-dimensional island and reef terrain maximum wave height spatial distribution prediction system, comprising:

[0078] A collection module, which extracts the topographic characteristics of the three-dimensional island and reef terrain and collects wave data in the open sea;

[0079] A simulation module, wherein the simulation module simulates the propagation evolution of waves in three-dimensional island and reef terrain according to a fully nonlinear wave numerical model based on the characteristics of offshore waves;

[0080] A calculation module, wherein the calculation module calculates the average wave height of each measuring point according to the wave surface change of each measuring point of the island and reef terrain according to the upper crossing zero point method, and obtains the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain;

[0081] A construction module, wherein the construction module changes the incident wave height, water depth and wave direction of the open sea respectively based on the control variable method, and establishes a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea;

[0082] An optimization module, wherein the optimization module constructs a random forest model and optimizes the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain an optimal random forest model;

[0083] An acquisition module, wherein the acquisition module inputs the modified offshore incident wave height, water depth and wave direction data into a random forest model to obtain a wave height spatial distribution map predicted by the random forest model;

[0084] A comparison module compares the wave height spatial distribution map drawn by the numerical simulation of the changed open sea incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyzes the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtains the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtains the maximum wave height spatial distribution.

[0085] Embodiment: The present invention discloses a method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain, the steps of which are as follows:

[0086] Step 1: Determine the measured island and reef terrain data, collect offshore wave data, and perform numerical simulation on the offshore wave data:

[0087] Simulate the evolution of wave propagation on island and reef terrain and arrange measuring points. In order to obtain the wave height distribution along the way, a total of 56 measuring points were arranged. The numerical simulation conditions used to construct the training set this time include water depths h of 0.30m, 0.325m, 0.35m, 0.375m, and 0.40m, corresponding to the submerged water depth h of the reef flat. r They are 0m, 0.025m, 0.05m, 0.075m, and 0.1m respectively, and the wave directions θ are 0°, 22.5°, 45°, 67.5°, and 90° respectively. The incident wave heights are 0.04m, 0.05m, 0.06m, 0.07m, and 0.08m respectively. The incident wave type is a random wave. Different water depths, different wave directions, and different wave heights are simulated in the simulation, totaling 5×5×5=125 groups. A large number of examples are calculated using the verified fully nonlinear wave numerical model. Each group of examples outputs the changes in the water level of 56 measuring points, and the output frequency is 50Hz, providing training data for the random forest.

[0088] Step 2: Construct random forest training set:

[0089] The results of the simulation of 125 examples were processed and analyzed to obtain the maximum wave height value of each measuring point in each example. Finally, the training data set was obtained, as shown in Figure 4 , the characteristics are: offshore water depth h, wave direction θ, incident wave height H0, measuring point number, and the corresponding label is the maximum wave height.

[0090] Step 3: Establishment of the maximum wave height prediction model based on random forest:

[0091] Construct a random forest algorithm, such as Figure 5 , 1) Data preparation. All the data in this study are provided by numerical simulation. As mentioned above, in the same data set, 80% of the samples are used for training and 20% of the samples are used for testing. 2) Hyperparameter optimization. First, the benchmark model is used for preliminary screening to determine the optimal combination of input variables; then, on the training set of the selected feature set, grid search (Grid Search CV) is used to optimize the hyperparameters of the machine learning algorithm. 3) Model construction. During the training process, the 5-fold cross-validation method is used after randomly dividing the training data, and the root mean square error is used as the objective function to find the optimal hyperparameter combination. 4) Model evaluation. The data samples in the test set are used for prediction evaluation, and the samples in the test set are substituted into the random forest model after training to make predictions and test its accuracy.

[0092] Note: For the hyperparameter optimization part, according to the actual situation of this study, the number of decision trees (n-estimators), the maximum depth of each decision tree (max-depth), the minimum number of separation samples (min-samples-split) and the minimum number of leaf node samples (min-samples-leaf) were selected for hyperparameter optimization. Figure 6 shown.

[0093] Step 6: Optimize prediction results:

[0094] Evaluate the prediction effect of the prediction model established in step 3. Generally, for the evaluation of machine learning models, the model is tested using 20% ​​of the samples divided from the training set, namely the test set, and then the coefficient of determination (R2), mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE) and other indicators are used to evaluate the model. According to the actual situation of this study, this section not only adopts the general evaluation method, but also proposes a new evaluation method. It includes four categories: water depth, wave direction, incident wave height, and measuring point. The prediction effect of the random forest model is evaluated by the control variable method. Since the measuring point variable is a discrete variable, it only represents the various locations of the islands and reefs. Each set of examples contains measuring points 1 to 56 and is fixed. We regard the islands and reefs as a whole and evaluate the model from the following three perspectives: First, control the water depth and wave direction to change only the incident wave height, and the changed incident wave height value does not appear in the wave data collected in the open sea; second, control the wave direction and incident wave height to change only the water depth, and the changed water depth value does not appear in the wave data collected in the open sea; third, control the water depth and incident wave height to change only the wave direction, and the changed wave direction value does not appear in the wave data collected in the open sea. Then, these three types of data are input into the fully nonlinear wave numerical model respectively, and three types of numerical simulations are used to draw the spatial distribution map of wave height. And the three types of data are input into the optimal random forest model respectively, and three types of wave height spatial distribution maps predicted by the random forest model are obtained;

[0095] The three types of numerical simulations of wave height spatial distribution maps are compared with the three types of wave height spatial distribution maps predicted by the random forest model to determine which data, water depth, wave direction and incident wave height, has a greater impact on the predicted maximum wave height spatial distribution. Then, the wave height spatial distribution obtained by using the random forest model to predict the wave data corresponding to the data type with the greatest impact on the maximum wave height spatial distribution is the maximum wave height spatial distribution.

[0096] In this embodiment, a set of working conditions AS36055-34 is randomly selected, that is, the input offshore water depth is 0.36m, the input wave direction is 34°, the input wave height is 0.055m, and the input measuring points 1 to 56 are input to obtain the maximum wave height. The wave height spatial distribution diagram based on the numerical simulation results is drawn as shown in FIG. Figure 7 , the spatial distribution of wave height based on random forest prediction results is shown in Figure 8 and two The relative error spatial distribution diagram of Fig. 9 The average error is only about 5%. This method can effectively predict the spatial distribution of the maximum wave height of three-dimensional island and reef terrain. The timely prediction of the maximum wave height of island and reef terrain can effectively avoid the huge damage to life safety and island and reef buildings.

[0097] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, 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 spatial distribution of maximum wave height of three-dimensional island and reef terrain, characterized in that: include: Step 1, extracting the topographic characteristics of the three-dimensional island and reef terrain and collecting wave data in the open sea; Step 2: Based on the characteristics of offshore waves, the propagation and evolution of waves in three-dimensional island and reef terrain is simulated according to a fully nonlinear wave numerical model; Step 3, according to the wave surface changes at each measuring point of the island and reef terrain, the average wave height of each measuring point is calculated according to the upper crossing zero point method, and the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain is obtained; Step 4, based on the control variable method, respectively change the incident wave height, water depth and wave direction of the open sea, and repeat steps 2 to 3 based on the changed data, so as to establish a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea; Step 5, construct a random forest model, and optimize the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain the optimal random forest model; Step 6, input the modified offshore incident wave height, water depth and wave direction data into the random forest model to obtain a spatial distribution map of wave height predicted by the random forest model; Step 7, compare the wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyze the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtain the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtain the maximum wave height spatial distribution.

2. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 1 is characterized in that: The wave data collected at the offshore location include: a plurality of groups of water depths at different depths, the submerged water depths of the corresponding reef flats, the wave directions and incident wave heights.

3. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 2 is characterized in that: According to the wave surface changes at each measuring point of the island and reef terrain, the average wave height of each measuring point is calculated according to the upper crossing zero point method, and the spatial distribution of the maximum wave height of the three-dimensional island and reef terrain is obtained, specifically: the wave surface data of each measuring point is preprocessed to remove abnormal values; And find the position of the zero point from positive to negative in the wave surface data; for each zero point, calculate the height from the previous trough to the next crest as the wave height; average the wave heights of all measuring points to get the average wave height of the measuring point; If the number of measuring points is limited, the average wave height of each measuring point is interpolated to obtain the spatial distribution of the average wave height of the entire island and reef terrain; in the interpolated spatial distribution, the maximum wave height value of each point is extracted to obtain the spatial distribution of the maximum wave height.

4. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 3 is characterized in that: The control variable method is based on respectively changing the incident wave height, water depth and wave direction of the open sea, specifically: controlling the water depth and wave direction to remain unchanged, only changing the incident wave height, and the changed incident wave height value does not appear in the wave data collected in the open sea; controlling the wave direction and incident wave height to remain unchanged, only changing the water depth, and the changed water depth value does not appear in the wave data collected in the open sea; controlling the water depth and incident wave height to remain unchanged, only changing the wave direction, and the changed wave direction value does not appear in the wave data collected in the open sea.

5. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 4 is characterized in that: The establishment of a wave height spatial distribution map drawn by numerical simulation including the changed offshore incident wave height, water depth and wave direction is specifically as follows: The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the fully nonlinear wave numerical model, and the simulation data is compared to obtain the maximum wave height value of each measuring point in each group of examples in the simulation data, and the maximum wave height value is the maximum wave peak value, and the wave height spatial distribution diagram after the incident wave height data is changed is constructed; Repeat the above steps, and input the changed water depth data with the wave direction and incident wave height unchanged and the changed wave direction data with the water depth and incident wave height unchanged into the fully nonlinear wave numerical model; The spatial distribution diagram of wave height after changing the water depth data and the spatial distribution diagram of wave height after changing the wave direction data are constructed respectively.

6. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 5 is characterized in that: The random forest model is optimized according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain the optimal random forest model, which is specifically: The spatial distribution data of the maximum wave height of three-dimensional island and reef terrain are divided into training sets and test sets; a suitable machine learning algorithm is selected as the benchmark model, each candidate feature set in the training set is trained based on the benchmark model, and the performance of the benchmark model is evaluated; based on the evaluation results, the feature set with the best performance is selected; Optimize the hyperparameters of the random forest model based on a grid search on the selected feature set; The training set is randomly divided into several subsets, and the random forest model is trained on the divided subsets through the five-fold cross-validation method; the performance of the model is evaluated based on the root mean square error (RMSE) as the objective function; the hyperparameter combination is continuously adjusted to obtain the optimal hyperparameter combination that minimizes the RMSE; Substitute the samples in the test set into the trained random forest model, and test the accuracy of the model by comparing the predicted results with the actual values ​​in the test set; based on the test results, optimize and adjust the random forest model until the random forest model with the best performance is found.

7. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 6 is characterized in that: The modified offshore incident wave height, water depth and wave direction data are input into the optimal random forest model to obtain a wave height spatial distribution map predicted by the random forest model, specifically: The water depth and wave direction remain unchanged, and the changed incident wave height data is input into the optimal random forest model to obtain the spatial distribution map of wave height predicted by the random forest model of the changed incident wave height data; Repeat the above steps and input the water depth data after the wave direction and incident wave height remain unchanged and the water depth and incident wave height remain unchanged but the wave direction data is changed into the fully nonlinear wave numerical model; respectively construct the spatial distribution map of wave height predicted by the random forest model after the water depth data is changed and the spatial distribution map of wave height predicted by the random forest model after the wave direction data is changed.

8. The method for predicting the spatial distribution of maximum wave height of three-dimensional island and reef terrain according to claim 7 is characterized in that: The wave height spatial distribution map drawn by the numerical simulation of the modified offshore incident wave height, water depth and wave direction and the wave height spatial distribution map predicted by the random forest model are compared, and the prediction effect and the error spatial distribution are analyzed based on the relative error spatial distribution map, and the data characteristics that have a greater impact on the maximum wave height spatial distribution are obtained, and then the maximum wave height spatial distribution is obtained, which is specifically: The wave height spatial distribution map drawn by the numerical simulation of the changed offshore incident wave height, water depth and wave direction and the wave height spatial distribution map predicted by the random forest model are compared to determine whether the relative error of the wave height exceeds the preset threshold. If not, it is determined that the offshore wave data corresponding to the wave height relative error closest to the threshold has the greatest influence on the spatial distribution of the maximum wave height, and the wave height spatial distribution predicted by the random forest model for the offshore wave data is the maximum wave height spatial distribution; if it exceeds, the random forest model is continuously optimized until the maximum wave height spatial distribution is obtained.

9. A three-dimensional island and reef terrain maximum wave height spatial distribution forecasting system, characterized in that: include: A collection module, which extracts the topographic characteristics of the three-dimensional island and reef terrain and collects wave data in the open sea; A simulation module, wherein the simulation module simulates the propagation evolution of waves in three-dimensional island and reef terrain based on the characteristics of offshore waves and a fully nonlinear wave numerical model; A calculation module, which calculates the average wave height of each measuring point according to the wave surface change of each measuring point of the island and reef terrain according to the upper crossing zero point method, and obtains the maximum wave height spatial distribution of the three-dimensional island and reef terrain; A construction module, wherein the construction module changes the incident wave height, water depth and wave direction of the open sea respectively based on the control variable method, and establishes a wave height spatial distribution map drawn by numerical simulation of the changed incident wave height, water depth and wave direction of the open sea; An optimization module, wherein the optimization module constructs a random forest model and optimizes the random forest model according to the spatial distribution data of the maximum wave height of the three-dimensional island and reef terrain to obtain an optimal random forest model; An acquisition module, wherein the acquisition module inputs the modified offshore incident wave height, water depth and wave direction data into a random forest model to obtain a wave height spatial distribution map predicted by the random forest model; A comparison module compares the wave height spatial distribution map drawn by the numerical simulation of the changed open sea incident wave height, water depth and wave direction with the wave height spatial distribution map predicted by the random forest model, analyzes the prediction effect and the error spatial distribution based on the relative error spatial distribution map, obtains the data characteristics that have a greater impact on the maximum wave height spatial distribution, and then obtains the maximum wave height spatial distribution.

Images