Wave height prediction method and device, equipment and storage medium

Abstract

The invention provides a wave height prediction method and device, equipment and a storage medium, and relates to the technical field of ocean resource development and application. The method comprises the following steps: constructing a first prediction model and a second prediction model according to wave sample data; adjusting model parameters in the first prediction model and the second prediction model according to the fitness of wave height prediction, and obtaining a first target prediction model and a second target prediction model under the condition that the fitness of the first prediction model and the second prediction model is highest; and coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting a wave height prediction result through the integrated prediction model. According to the method, model diversity can be introduced, feature information in wave data can be captured, the advantages of the model are further fused, and therefore the accuracy of predicting the significant wave height of the sea waves can be improved.

Description

Technical Field

[0001] The present application relates to the field of marine resource development and application technology, and specifically to a wave height prediction method, device, equipment and storage medium. Background Art

[0002] In recent years, as the country pays more and more attention to the development and protection of marine resources, wave energy as a renewable clean energy has received widespread attention. Wave height is proportional to wave energy. The greater the wave height, the greater the wave energy. Therefore, analyzing the changing characteristics of wave height is very important for the study of wave energy.

[0003] Currently, the commonly used methods for predicting ocean effective wave height include spectral analysis, experimental modeling, field measurement, etc. However, ocean wave height has the characteristics of nonlinearity, non-smoothness, randomness, etc. It is difficult to capture the important characteristic information in the wave data when predicting wave height through the above methods, so there is a problem of low accuracy in wave height prediction. Summary of the invention

[0004] The present application provides a wave height prediction method, device, equipment and storage medium, which can introduce model diversity by constructing two prediction models, and this diversity can help capture different characteristics and laws of wave data, so that it is easier to capture important characteristic information from the wave data. The integrated prediction model obtained by coupling the first target prediction model and the second target prediction model can further integrate the advantages of the two models, and then predict the wave height through the integrated prediction model, so as to improve the accuracy of predicting the effective wave height of the waves.

[0005] In a first aspect, an embodiment of the present application provides a wave height prediction method, which may include: constructing a first prediction model and a second prediction model based on wave sample data; adjusting model parameters in the first prediction model and the second prediction model according to fitness for wave height prediction, and obtaining a first target prediction model and a second target prediction model when the fitness of the first prediction model and the second prediction model is the highest; coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting the wave height prediction result through the integrated prediction model.

[0006] In the above implementation, model diversity is introduced by constructing two prediction models, and this diversity can help capture different characteristics and laws of wave data, thereby making it easier to capture important characteristic information from wave data. The integrated prediction model obtained by coupling the first target prediction model with the second target prediction model can further integrate the advantages of the two models, and then the wave height is predicted by the integrated prediction model, thereby improving the accuracy of predicting the effective wave height of the waves.

[0007] In some embodiments, adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest, may include: The fitness of the first prediction model when the model parameter of the first prediction model is determined to be a first parameter; wherein the first parameter is any model parameter of the first prediction model; the fitness of the second prediction model when the model parameter of the second prediction model is determined to be a second parameter; wherein the second parameter is any model parameter of the second prediction model.

[0008] In some embodiments, when the model parameter of the first prediction model is determined to be the first parameter, the fitness of the first prediction model may include: Determine a first parameter interval of the first parameter according to the wave sample data; determine the fitness of the first prediction model for wave height prediction when any first parameter is selected from the first parameter interval; when the fitness of the first prediction model for wave height prediction is the highest, determine the current first parameter as the first target parameter; wherein the first target prediction model is a model whose model parameter is the first target parameter.

[0009] In the case where the model parameter of the second prediction model is determined to be the second parameter, the fitness of the second prediction model may include: Determine a second parameter interval of the second parameter according to the wave sample data; determine the fitness of the second prediction model for wave height prediction when any second parameter is selected from the second parameter interval; when the fitness of the second prediction model for wave height prediction is the highest, determine the current second parameter as the second target parameter; wherein the second target prediction model is a model whose model parameter is the second target parameter.

[0010] In the above implementation process, by adjusting the model parameters, the optimized model parameters can reduce the error between the predicted value and the actual value, make the model better fit the wave height data, make the prediction result closer to the actual situation, and thus improve the accuracy of the prediction. By searching for the first parameter and the second parameter that make the prediction model most adaptable in the first parameter interval of the first parameter and the second parameter interval of the second parameter, the parameter search range can be narrowed, the efficiency of model parameter optimization can be improved, and the calculation cost can be reduced.

[0011] In some embodiments, coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting a wave height prediction result through the integrated prediction model may include: Determine the first weighting parameter of the first target prediction model and the second weighting parameter of the second target prediction model respectively; determine the integrated prediction model based on the output result of the first target prediction model, the output result of the second target prediction model, the first weighting parameter and the second weighting parameter based on the linear weighting method, and output the wave height prediction result through the integrated prediction model.

[0012] In some embodiments, the determining the first weighted parameter of the first target prediction model and the second weighted parameter of the second target prediction model respectively may include: A first weighting parameter of the first target prediction model and a second weighting parameter of the second target prediction model are determined according to the adaptability to the wave height prediction.

[0013] In the above implementation, the integrated prediction model obtained by coupling the first target prediction model with the second target prediction model can further integrate the advantages of the two models, and then the wave height is predicted by the integrated prediction model, thereby improving the accuracy of predicting the effective wave height of the waves. The introduction of weighted parameters can make the integrated prediction model correct the prediction errors of different models when fusing the output results of the two models, thereby reducing the overall prediction error and improving the prediction accuracy.

[0014] In some embodiments, the adaptability to wave height prediction includes one or more of the determination coefficient, mean absolute error, root mean square error and mean absolute percentage error of the first target prediction model and the second target prediction.

[0015] In the above implementation, by introducing multiple evaluation indicators such as determination coefficient, mean absolute error, root mean square error and mean absolute percentage error, the performance of the prediction model can be comprehensively evaluated from different angles. These indicators can respectively reflect the degree of fit of the first target prediction model and the second target prediction model, the size of the prediction error, the degree of discreteness of the error and the ratio of the error to the actual value, so as to more comprehensively characterize the prediction accuracy of the model. And the combination of multiple evaluation indicators can also complement each other, so as to more accurately evaluate the performance of the model.

[0016] In some embodiments, the first direction is a downward direction relative to the energy mean, and the first envelope point is an upper envelope point of the first one-dimensional spectrum data.

[0017] In some embodiments, the first prediction model is an extreme learning machine model, and the second prediction model is a random forest model.

[0018] In a second aspect, an embodiment of the present application provides a wave height prediction device, which may include: A model building module, used for building a first prediction model and a second prediction model according to wave sample data; A parameter adjustment module, used for adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest; A coupling module is used to couple the first target prediction model and the second target prediction model to obtain an integrated prediction model, and output a wave height prediction result through the integrated prediction model.

[0019] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the above description when executing the computer program.

[0020] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the muscle function assessment method as described in the above description is implemented.

[0021] Compared with the prior art, the beneficial effect of the present application is that: by constructing two prediction models, model diversity is introduced, and this diversity can help capture different characteristics and laws of wave data, thereby making it easier to capture important characteristic information from wave data. The advantages of the two models can be further integrated by coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and then the wave height is predicted by the integrated prediction model, thereby improving the accuracy of predicting the effective wave height of the waves. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of the steps of the wave height prediction method provided in an embodiment of the present application.

[0023] Figure 2 A graph showing iterative optimization of the parameters of the ELM model provided in an embodiment of the present application.

[0024] Figure 3 A graph showing iterative optimization of parameters of an RF model provided in an embodiment of the present application.

[0025] Figure 4 A schematic diagram of a wave height prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The present application is further described in detail below in conjunction with test examples and specific implementation methods. However, this should not be understood as the scope of the above subject matter of the present application being limited to the following embodiments, and all technologies implemented based on the content of the present application belong to the scope of protection of the present application.

[0027] Unless otherwise specified, in the description of the specific embodiments of the present application, the terms indicating the orientation or position relationship such as "up", "down", "left", "right", "center", "inside", "outside", "side", etc. are all expressions based on the orientation or position relationship shown in the drawings, or the orientation or position relationship when the product / equipment / device is usually used. These terms of orientation or position relationship are only for the convenience of describing the scheme of the present application or simplifying the description in the specific embodiments to facilitate the technicians to quickly understand the scheme, rather than indicating or implying that a specific device / component / element must have a specific orientation, or be constructed and operated in a specific position relationship, and therefore cannot be understood as a limitation on the present application.

[0028] In the description of the embodiments of the present application, the technical terms "first", "second", etc. only distinguish one entity or operation from another entity or operation, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] Example 1 Please see Figure 1 , Figure 1 A schematic diagram of the steps of the wave height prediction method provided in an embodiment of the present application. The wave height prediction method may include: S1. Construct a first prediction model and a second prediction model based on wave sample data.

[0031] S2. Adjust model parameters in the first prediction model and the second prediction model according to the fitness of wave height prediction, and obtain the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest.

[0032] S3. Coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting the wave height prediction result through the integrated prediction model.

[0033] In an embodiment of the present application, the wave sample data may include wind speed, wind direction, main wave period of sea waves, average wave period, average wave direction of sea waves, average air pressure at sea level, sea surface temperature (sea temperature), etc. The input data of the first prediction model and the second prediction model are wave sample data, and the output data are wave height prediction results. The first prediction model and the second prediction model are different prediction models. The first prediction model and the second prediction model can be neural network models, such as Extreme Learning Machine (ELM), Support Vector Machine (SVM), Random Forest Model (RF), Group Method of Data Handling (GMDH), etc.

[0034] Model parameters refer to the configuration variables within the model, which can be automatically learned based on data and driven and adjusted by the model itself based on data. Model parameters can include input weights, hidden layer biases, the number of features of the decision tree, the number of decision trees, etc. The wave height prediction result refers to the effective wave height of the waves, which means arranging the observed wave data from large to small according to the wave height, and taking the average wave height of the first 1 / 3 of the large waves. The effective wave height reflects the height of the significant part of the waves and is a key reference indicator in the fields of navigation, port design, and marine engineering. Coupling refers to associating the outputs of the first target prediction model and the second target prediction model through a coupling method to obtain an integrated prediction model.

[0035] In the above implementation process, model diversity is introduced by constructing two prediction models. This diversity can help capture different characteristics and laws of wave data, making it easier to capture important characteristic information from wave data. The integrated prediction model obtained by coupling the first target prediction model with the second target prediction model can further integrate the advantages of the two models, and then the wave height is predicted by the integrated prediction model, thereby improving the accuracy of predicting the effective wave height of the waves.

[0036] Example 2 This embodiment is a specific implementation plan for obtaining the first target prediction model and the second target prediction model by adjusting the model parameters in the first prediction model and the second prediction model according to the adaptability to wave height prediction in the above embodiment 1.

[0037] The step of adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest, may include: S21. Determine the fitness of the first prediction model when the model parameter of the first prediction model is the first parameter.

[0038] Among them, the first parameter is any model parameter of the first prediction model. The first prediction model can have one first parameter or multiple first parameters. The number of first parameters depends on the type of the selected first prediction model. The following determination of the first parameter of the first prediction model is exemplary. If there are multiple first parameters, the fitness corresponding to each first parameter of the first prediction model can be determined respectively by repeating the following steps, and the first parameter with the highest fitness is determined as the first target parameter. S21 can specifically include: S211. Determine a first parameter interval of a first parameter according to the wave sample data.

[0039] S212: Determine the adaptability of the first prediction model to wave height prediction when any first parameter is selected from the first parameter interval.

[0040] S213: When the first prediction model has the highest adaptability to wave height prediction, determine the current first parameter as the first target parameter.

[0041] Among them, the first target prediction model is a model whose model parameters are the first target parameters. In the embodiment of the present application, the applicant has found through research that the extreme learning machine, as a kind of neural network, is good at processing nonlinear relationships and complex patterns in data, and the random forest can also capture the nonlinear features in the data by integrating multiple decision trees, and the two are complementary in processing nonlinear relationships. The extreme learning machine can avoid overfitting to a certain extent by randomly initializing the hidden layer weights and biases. The random forest can further reduce the risk of overfitting by constructing multiple decision trees and combining their prediction results. The integrated prediction model obtained by combining the two can have a higher generalization ability. By coupling the outputs of the extreme learning machine and the random forest model, the advantages of the two models can be fully utilized to improve the accuracy of wave height prediction. The extreme learning machine and the random forest model each have their own advantages in processing data of different types and characteristics. By coupling the outputs of the two models, the complexity and diversity of wave height data can be better adapted.

[0042] Therefore, the embodiment of the present application uses an extreme learning machine model as the first prediction model and a random forest model as the second prediction model for illustration: For a given wave sample data , the output value of the extreme learning machine model after the simulation process can be expressed as: in, represents the output value of the model, T represents the number of hidden layers, represents the bias of the output layer, represents the activation function, represents the feature vector and label of the i-th wave sample, represents the input weight of the hidden layer, represents the bias of the hidden layer.

[0043] For ease of calculation, the output of the extreme learning machine model in the embodiment of the present application can be simplified as: O=Hc Among them, O represents the output matrix of the model, H represents the output matrix of the hidden layer, and c represents the output weight matrix of the hidden layer. The matrix c can be solved based on the Moore-Penrose generalized inverse.

[0044] The first parameter interval can be set according to the number of collected samples and prior knowledge. For example, in the embodiment of the present application, the first parameter of the ELM model can be the input weight and the deviation of the hidden layer, the first parameter interval can be the preset period of the input weight and the preset interval of the deviation of the hidden layer, the first parameter interval can be specifically {0, 1, 2, 3, 4, 5}, and the number of neurons in the hidden layer of the ELM model can be preset to 102.

[0045] In an embodiment of the present application, the Golden Jackal Optimization (GJO) algorithm can be used to perform parameter optimization on the ELM model, and the first parameter of the iterative model is continuously optimized in a set first parameter interval, so as to determine the first target parameter of the ELM model when the fitness of the model is the highest.

[0046] The process of optimizing the parameters of the extreme learning machine model through GJO can include: Initializing the population: randomly generating the initial population within the preset interval of the input weight and the hidden layer bias, and each individual represents a set of parameters of the ELM model. Evaluating fitness: using the ELM model to train the training data and calculate the fitness value of each individual. Determining the male and female golden jackal pair: according to the fitness value, selecting the individual with the best fitness as the male golden jackal and the second best individual as the female golden jackal. Prey search phase: the male golden jackal updates its position according to the mathematical model in the GJO algorithm based on the current position, the position of the female golden jackal and the position of the prey (i.e. the current parameter solution). The female golden jackal also updates its position. After the update, the fitness value of each individual is recalculated. Encirclement and attack prey phase: on the basis of searching for prey, further narrow the search range, and encircle and attack the prey. Update the positions of the male and female golden jackals and recalculate the fitness value. Iterative update: Repeat the prey search phase and the prey siege phase until the preset number of iterations is reached or the stop condition is met. Select the optimal parameter: After the iteration, select the individual with the best fitness as the optimal parameter solution. Verify with the test data set: Apply the optimal parameter solution to the ELM model and verify it with the test data set. Calculate fitness indicators such as the coefficient of determination, mean absolute error, root mean square error, and mean absolute percentage error to evaluate the performance of the model, determine the parameter with the highest performance as the first target parameter, and determine the first target prediction model.

[0047] The fitness of the first prediction model for wave height prediction can be judged by the statistical indicators of fitness, which can include determination coefficient, mean absolute error, root mean square error and mean absolute percentage error.

[0048] Among them, the coefficient of determination R² represents the percentage of variance of the dependent variable that can be explained by the independent variable. R² shows the degree of fit of the data to the regression model, that is, the model's ability to explain the data. The closer the value of R² is to 1, the better the degree of fit of the model, the higher the variability that the model can explain, and the higher the fitness value. Mean Absolute Error (MAE) is the average of the absolute values ​​of the difference between the predicted value and the actual value. MAE measures the average difference between the predicted value and the actual value, thereby evaluating the prediction accuracy of the model. The smaller the value of MAE, the more accurate the model's prediction, and the higher the fitness value. Root Mean Square Error (RMSE) is the square root of the average of the squares of the prediction errors. RMSE measures the degree of difference between the predicted value and the actual observed value. The smaller its value, the more accurate the model's prediction, and the higher the fitness value. RMSE squares the prediction error, so it can significantly reflect the large error in the prediction. Mean Absolute Percentage Error (MAPE%) is the average of the percentage difference between the predicted value and the true value. MAPE expresses the difference between the predicted value and the true value in the form of a percentage. The closer MAPE is to 0%, the better the prediction performance of the model is, and the higher the fitness value is.

[0049] S22: Determine the fitness of the second prediction model when the model parameter of the second prediction model is the second parameter.

[0050] The second parameter is any model parameter of the second prediction model. The second prediction model may have one second parameter or multiple second parameters. The number of second parameters depends on the type of the selected second prediction model.

[0051] The fitness of the second prediction model for wave height prediction can also be judged by statistical indicators of fitness. The statistical indicators of fitness can include determination coefficient, mean absolute error, root mean square error and mean absolute percentage error.

[0052] Exemplarily, if there are multiple second parameters, the following steps can be repeated to respectively determine the fitness corresponding to each second parameter of the second prediction model, and determine the second parameter with the highest fitness as the second target parameter. S22 may specifically include: S221. Determine a second parameter interval of a second parameter according to the wave sample data.

[0053] S222, determining the adaptability of the second prediction model to wave height prediction when any second parameter is selected from the second parameter interval; S223. When the second prediction model has the highest adaptability to wave height prediction, determine the current second parameter as the second target parameter.

[0054] Among them, the second target prediction model is a model whose model parameters are the second target parameters.

[0055] The second parameter interval can also be set according to the number of collected samples and prior knowledge. For example, in an embodiment of the present application, the second parameter of the RF model can be the number of features and the number of decisions of the decision tree, the first parameter interval can be the interval of the number of features and the number of decisions of the decision tree, the interval of the number of features of the decision tree can be specifically {0, 1, 2, 3, 4, 5, 6, 7, 8}, and the preset interval of the number of decisions can be from 0 to 1000.

[0056] In an embodiment of the present application, the method for optimizing the parameters of the RF model can also be performed by using the golden jackal optimization algorithm, continuously optimizing the second parameter of the iterative model in a set second parameter interval, so as to determine the second target parameter of the RF model when the fitness of the model is the highest.

[0057] The fitness of the second prediction model for wave height prediction can also be judged by the statistical indicators of fitness. The statistical indicators of fitness can include the determination coefficient, mean absolute error, root mean square error and mean absolute percentage error. The process of parameter optimization by using the golden jackal optimization algorithm can refer to the content of parameter optimization for the ELM model in the above description, which will not be repeated here.

[0058] In the above implementation process, by adjusting the model parameters, the optimized model parameters can reduce the error between the predicted value and the actual value, make the model better fit the wave height data, make the prediction result closer to the actual situation, and thus improve the accuracy of the prediction. By searching for the first parameter and the second parameter that make the prediction model most adaptable in the first parameter interval of the first parameter and the second parameter interval of the second parameter, the parameter search range can be narrowed, the efficiency of model parameter optimization can be improved, and the calculation cost can be reduced.

[0059] Example 3 This embodiment is a specific implementation plan for obtaining an integrated prediction model by coupling the first target prediction model and the second target prediction model in the above-mentioned embodiment 1, and outputting the wave height prediction result through the integrated prediction model.

[0060] The steps of coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting the wave height prediction result through the integrated prediction model may specifically include: S31. Determine a first weighting parameter of a first target prediction model and a second weighting parameter of a second target prediction model respectively.

[0061] Wherein, respectively determining a first weighted parameter of the first target prediction model and a second weighted parameter of the second target prediction model may include: A first weighting parameter of the first target prediction model and a second weighting parameter of the second target prediction model are determined according to the adaptability to the wave height prediction.

[0062] Among them, the first weighted parameter of the first target prediction model and the second weighted parameter of the second target prediction model can be determined by the fitness indicators such as the determination coefficient, mean absolute error, root mean square error and mean absolute percentage error in the above description. When the fitness index of the first target prediction model is higher than the fitness index of the second target prediction model, it can be determined that the first weighted parameter is greater than the second weighted parameter. When the fitness index of the first target prediction model is lower than the fitness index of the second target prediction model, it can be determined that the first weighted parameter is less than the second weighted parameter.

[0063] S32, determining an integrated prediction model based on a linear weighted method using the output result of the first target prediction model, the output result of the second target prediction model, the first weighting parameter and the second weighting parameter, and outputting the wave height prediction result through the integrated prediction model.

[0064] In the embodiment of the present application, the first target prediction model and the second target prediction model can be coupled by a direct linear weighting method to obtain an integrated prediction model, and the output of the integrated prediction model can be expressed as: in, represents the output of the hybrid ensemble model, is the output result of the first target prediction model, is the output result of the second target prediction model, m is the first weighting parameter, n is the second weighting parameter, and m+n=1.

[0065] In the above implementation process, the integrated prediction model obtained by coupling the first target prediction model with the second target prediction model can further integrate the advantages of the two models, and then the wave height is predicted by the integrated prediction model, thereby improving the accuracy of predicting the effective wave height of the waves. The introduction of weighted parameters can make the integrated prediction model correct the prediction errors of different models when fusing the output results of the two models, thereby reducing the overall prediction error and improving the prediction accuracy.

[0066] Example 4 This embodiment is a practical application example of a wave height prediction method in Embodiment 1.

[0067] The following is a set of input wave sample data shown in an embodiment of the present application: Wind Direction WDIR=140.2917; Wind Speed ​​WSPD=8.6982m / s; Gust Speed ​​GST=10.4994m / s; Dominant Period DPD=9.1970s; Average Period APD=6.1268s; Mean Wind Direction MWD=140.0476; Pressure PRES=1020.0333hPa; Water Temperature WTMP=25.7315℃, output effective wave height WVHT=2.2147m.

[0068] Based on the above wave sample data, the ELM model and RF model were constructed respectively. The number of hidden neurons in the ELM model is 102, and the input weights and hidden layer biases are all {0, 1, 2, 3, 4, 5}. The number of decision trees (n=Ntree) in the RF model ranges from 0 to 1000; the number of features randomly sampled when building each decision tree (m=Mtry) ranges from {0, 1, 2, 3, 4, 5, 6, 7, 8}.

[0069] Please see Figure 2 and Figure 3 , Figure 2 A graph showing iterative optimization of the parameters of the ELM model provided in an embodiment of the present application. Figure 3 A graph for iterative optimization of the parameters of the RF model provided in the embodiment of the present application. In the manner provided in the above description, after using the GJO algorithm to optimize the parameters of the ELM model and the RF model according to the fitness for wave height prediction, the number of decision trees of the RF model (n=Ntree) = 670, the number of features randomly sampled when constructing each decision tree (m=Mtry) = 6, the number of hidden layer neurons of the ELM model is 102, and the input weights and hidden layer biases are matrices of 102*8 and 102*1, respectively.

[0070] Please refer to Table 1 below, which is a matrix diagram of the input weights and hidden layer biases of the ELM model: Table 1 After coupling the first target prediction model and the second target prediction model to obtain the integrated prediction model, the integrated prediction model outputs the wave height prediction result of WVHT=2.2328. Table 2 below is a performance comparison diagram of the first target prediction model, the second target prediction model and the integrated prediction model: Table 2 It can be seen from Table 2 that the integrated prediction model has a higher adaptability in wave height prediction than the first target prediction model and the second target prediction model. Therefore, the wave height prediction results output by the integrated prediction model are more accurate.

[0071] Example 5 Please see Figure 4 , Figure 4 Schematic diagram of a wave height prediction device provided in an embodiment of the present application. The wave height prediction device 40 may include: The model building module 41 is used to build a first prediction model and a second prediction model according to the wave sample data.

[0072] The parameter adjustment module 42 is used to adjust the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtain the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest.

[0073] The coupling module 43 is used to couple the first target prediction model and the second target prediction model to obtain an integrated prediction model, and output the wave height prediction result through the integrated prediction model.

[0074] Optionally, the parameter adjustment module 42 may be specifically used for: Determine the fitness of the first prediction model when the model parameter of the first prediction model is the first parameter; wherein the first parameter is any model parameter of the first prediction model; Determine the fitness of the second prediction model when the model parameter of the second prediction model is the second parameter; wherein the second parameter is any model parameter of the second prediction model.

[0075] Optionally, the parameter adjustment module 42 may be specifically used for: Determine a first parameter interval of the first parameter according to wave sample data; determine the fitness of the first prediction model for wave height prediction when any first parameter is selected from the first parameter interval; when the fitness of the first prediction model for wave height prediction is the highest, determine the current first parameter as the first target parameter; wherein the first target prediction model is a model whose model parameter is the first target parameter.

[0076] And determine the second parameter interval of the second parameter according to the wave sample data; determine the fitness of the second prediction model for wave height prediction when any second parameter is selected from the second parameter interval; when the fitness of the second prediction model for wave height prediction is the highest, determine the current second parameter as the second target parameter; wherein the second target prediction model is a model whose model parameter is the second target parameter.

[0077] Optionally, the coupling module 43 may be specifically used for: Determine the first weighting parameter of the first target prediction model and the second weighting parameter of the second target prediction model respectively; determine the integrated prediction model based on the output result of the first target prediction model, the output result of the second target prediction model, the first weighting parameter and the second weighting parameter based on the linear weighting method, and output the wave height prediction result through the integrated prediction model.

[0078] Optionally, the coupling module 43 may be specifically used for: A first weighting parameter of the first target prediction model and a second weighting parameter of the second target prediction model are determined according to the adaptability to the wave height prediction.

[0079] It should be understood that, when the various modules of the wave height prediction device 40 provided in the above embodiment perform wave height prediction of ocean waves, only the division of the various functional modules in the above description is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0080] The functional modules in the above embodiments may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit, and the above integrated unit may be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the protection scope of the embodiments of the present application.

[0081] Based on the same application concept, an embodiment of the present application also provides a computer device, which may include a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the wave height prediction method as described in the above description is implemented.

[0082] Based on the same application concept, an embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the wave height prediction method described in the above description.

[0083] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A wave height prediction method, characterized in that: include: constructing a first prediction model and a second prediction model based on the wave sample data; Adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest; The first target prediction model and the second target prediction model are coupled to obtain an integrated prediction model, and a wave height prediction result is output through the integrated prediction model.

2. The method according to claim 1, characterized in that: The step of adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest, comprises: Determine the fitness of the first prediction model when the model parameter of the first prediction model is a first parameter; wherein the first parameter is any model parameter of the first prediction model; Determine the fitness of the second prediction model when the model parameter of the second prediction model is a second parameter; wherein the second parameter is any model parameter of the second prediction model.

3. The method according to claim 2, characterized in that In the case where the model parameter of the first prediction model is determined to be the first parameter, the fitness of the first prediction model includes: determining a first parameter interval of the first parameter according to the wave sample data; determining the adaptability of the first prediction model to wave height prediction when any first parameter is selected from the first parameter interval; In the case where the first prediction model has the highest adaptability to wave height prediction, determining the current first parameter as the first target parameter; wherein the first target prediction model is a model whose model parameter is the first target parameter; In the case where the model parameter of the second prediction model is determined to be the second parameter, the fitness of the second prediction model includes: determining a second parameter interval of the second parameter according to the wave sample data; determining the adaptability of the second prediction model to wave height prediction when any second parameter is selected from the second parameter interval; When the second prediction model has the highest adaptability to wave height prediction, the current second parameter is determined as the second target parameter; wherein the second target prediction model is a model whose model parameter is the second target parameter.

4. The method according to claim 1, characterized in that The step of coupling the first target prediction model and the second target prediction model to obtain an integrated prediction model, and outputting a wave height prediction result through the integrated prediction model, comprises: respectively determining a first weighting parameter of the first target prediction model and a second weighting parameter of the second target prediction model; The integrated prediction model is determined based on a linear weighted method by the output result of the first target prediction model, the output result of the second target prediction model, the first weighting parameter and the second weighting parameter, and the wave height prediction result is output through the integrated prediction model.

5. The method according to claim 4, characterized in that The respectively determining a first weighted parameter of the first target prediction model and a second weighted parameter of the second target prediction model comprises: A first weighting parameter of the first target prediction model and a second weighting parameter of the second target prediction model are determined according to the adaptability to the wave height prediction.

6. The method according to claim 5, characterized in that The adaptability to wave height prediction includes one or more of the determination coefficient, mean absolute error, root mean square error and mean absolute percentage error of the first target prediction model and the second target prediction.

7. The method according to any one of claims 1 to 6, characterized in that: The first prediction model is an extreme learning machine model, and the second prediction model is a random forest model.

8. A wave height prediction device, characterized in that: include: A model building module, used for building a first prediction model and a second prediction model according to wave sample data; A parameter adjustment module, used for adjusting the model parameters in the first prediction model and the second prediction model according to the fitness of the wave height prediction, and obtaining the first target prediction model and the second target prediction model when the fitness of the first prediction model and the second prediction model is the highest; A coupling module is used to couple the first target prediction model and the second target prediction model to obtain an integrated prediction model, and output a wave height prediction result through the integrated prediction model.

9. A computer device, characterized in that: The computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images