A wave field reconstruction method, device, electronic equipment and medium

Abstract

The present invention discloses a wave field reconstruction method, device, electronic device and medium. The method comprises: obtaining wave field data of a target area, using a connection function to fill missing values ​​in the wave field data, and obtaining a complete data set; obtaining wave field reanalysis data of the target area, and obtaining ocean model reanalysis data according to the wave field reanalysis data; based on a diffusion model, performing data fusion on the complete data set and the ocean model reanalysis data to obtain a target data set; based on the target data set, using a pulse neural network to classify the wave field, and obtaining a type classification result; based on the type classification result, performing regional reconstruction of each type of wave field in the target area based on a preset reconstruction method; based on the regional reconstruction result, using a maximum cross-correlation method to connect and fuse different regions, and obtaining a full-region wave field reconstruction result. The present invention can efficiently perform wave field reconstruction and can be widely used in the field of data processing technology.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a wave field reconstruction method, device, electronic equipment and medium. Background Art

[0002] At present, the most commonly used methods for wave field reconstruction are statistical methods, numerical simulation methods, methods based on remote sensing data, and methods based on machine learning. Statistical methods analyze historical observation data and establish statistical models to estimate future wave fields;

[0003] Two commonly used numerical simulation methods are based on the finite difference method (FDM) and the finite element method (FEM). Among them, the finite difference method discretizes the continuous wave field into a grid, and obtains the numerical solution of the wave field by approximately calculating the derivatives and differential equations at the grid points. The finite element method divides the wave field into many small units, establishes the relationship and boundary conditions between the units, and uses the variational principle to solve the approximate solution of the wave field. In addition, the boundary element method and the spectral method are also used in wave field reconstruction. The boundary element method discretizes the boundary of the wave field and solves the wave field through integral equations on the boundary; the spectral method uses Fourier series to expand the wave field and approximates the solution of the wave field by expressing the wave field as a linear combination of a set of basis functions;

[0004] The method based on remote sensing data uses satellite, radar and other remote sensing technologies to obtain information on ocean wave fields and reconstruct the wave fields through data processing and analysis;

[0005] Machine learning-based methods use large amounts of observational and simulation data to train models to predict and reconstruct wave fields.

[0006] According to the existing wave field reconstruction method, the existing technology has the following disadvantages:

[0007] 1) Statistical methods are less effective for complex ocean environments and nonlinear wave fields;

[0008] 2) Among the numerical simulation methods, the finite difference method is relatively difficult to handle complex boundary conditions and irregular grids, and its accuracy is limited by the grid resolution; the finite element method has high computational complexity and requires grid division and unit connection, and the calculation time is long for large-scale problems; the boundary integral equations in the boundary element method are difficult to solve, and it is necessary to deal with singular integrals and singularity problems, and it is mainly applicable to two-dimensional wave field problems, but needs to be expanded for three-dimensional problems; the spectral method has high requirements for the grid, and requires evenly distributed nodes to ensure the calculation accuracy. In addition, a large number of function expansions and coefficient calculations are required during the calculation process, which results in a large amount of calculation.

[0009] 3) The method based on remote sensing data is limited by the resolution and accuracy of remote sensing data and has poor effect on small-scale wave fields;

[0010] 4) Machine learning-based methods require a large amount of observational data for model training, which is greatly affected by data quality. In order to better capture the characteristics of the wave field, the model may be more complex and have higher computational and storage costs. Summary of the invention

[0011] The main purpose of the embodiments of the present invention is to provide a wave field reconstruction method, device, electronic device and medium, in order to solve at least one problem of the prior art. The present invention can efficiently perform wave field reconstruction.

[0012] To achieve the above objective, an embodiment of the present invention provides a wave field reconstruction method, the method comprising:

[0013] Obtain the wave field data of the target area, use the connection function to fill the missing values ​​of the wave field data, and obtain a complete data set;

[0014] Obtain wave field reanalysis data in the target area, and obtain ocean model reanalysis data based on the wave field reanalysis data;

[0015] Based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain the target data set;

[0016] According to the target data set, the pulse neural network is used to classify the wave field and obtain the type classification result of the target area;

[0017] According to the type classification results, the wave field of each type in the target area is reconstructed in different regions based on the preset reconstruction method;

[0018] Based on the results of regional reconstruction, the maximum cross-correlation method is used to connect and merge different regions to obtain the reconstruction results of the wave field in the entire region.

[0019] In some embodiments, the missing values ​​of the wave field data are filled using a connection function to obtain a complete data set, including:

[0020] Pre-fill missing values ​​of original wave field data by using difference method;

[0021] According to the correlation between variables in the pre-filled wave field data, the missing values ​​are estimated by the t-Copula function;

[0022] Among them, the parameters of the t-Copula function are obtained by maximum likelihood method;

[0023] According to the results of missing value estimation, the missing values ​​of the original wave field data are filled to obtain a complete data set.

[0024] In some embodiments, based on the diffusion model, data fusion is performed on the complete data set and the ocean model reanalysis data to obtain a target data set, including:

[0025] The full dataset and ocean mode were normalized by standard scores;

[0026] Based on the preset feature selection method, principal component analysis is performed on the normalized data to obtain the principal component features of each data source;

[0027] The diffusion model is established by using partial differential equations as wave propagation equations;

[0028] Use Euclidean distance to calculate the similarity between different data sources, and determine the diffusion rules based on the similarity;

[0029] Based on the diffusion model, the finite difference method is used for discretization to obtain the spatiotemporal distribution of the wave field, and the information propagation parameters are adjusted according to the diffusion rule.

[0030] Based on the relative importance of the propagated information and the original observation data of the complete dataset and ocean model reanalysis data, a weighted average is used for data fusion to obtain the target dataset.

[0031] In some embodiments, based on the target data set, a pulse neural network is used to classify the wave field to obtain a type classification result of the target area, including:

[0032] According to the target data set, the wave characteristics of each area in the target area are obtained by using statistical methods; the wave characteristics include wave height and wave period;

[0033] The wave characteristics are input into the pulse neural network to classify the wave field and obtain the type classification result of the target area;

[0034] The spiking neural network includes an input layer, a hidden layer, and an output layer; the hidden layer is constructed based on the Izhikevich neuron model, and the number of neurons in the output layer is the same as the total number of wave field types;

[0035] The spiking neural network is obtained by training the network and adjusting parameters through stochastic gradient descent and mean square error based on the wave field dataset with labeled wave field types.

[0036] In some embodiments, the type classification results include estuary area, nearshore area, offshore area and deep sea area; according to the type classification results, the wave field of each type in the target area is reconstructed by region based on a preset reconstruction method, including:

[0037] For the estuary area, a pre-built ocean dynamics model is used to reconstruct the first wave field in combination with numerical simulation methods. The ocean dynamics model is obtained by coupling the ocean-river interaction model, the tidal model and the ocean wave model.

[0038] For the nearshore area, wave parameters are extracted from radar images in the target dataset, and then the second wave field is reconstructed by combining convolutional autoencoder and dual attention mechanism;

[0039] For offshore areas, the third wave field is reconstructed using a high-order spectral method;

[0040] For deep sea areas, a cellular automaton model is established based on the remote sensing data in the target data set to reconstruct the fourth wave field.

[0041] In some embodiments, based on the results of regional reconstruction, the maximum cross-correlation method is used to connect and merge different regions to obtain the reconstruction results of the wave field in the entire region, including:

[0042] The cross-correlation function is calculated for the adjacent wave field data in the results of regional reconstruction, and the connection position is determined according to the peak value of the cross-correlation function;

[0043] The wave field data of adjacent areas are connected and fused according to the connection positions, and the connection positions are interpolated and smoothed to obtain the wave field reconstruction results of the entire area.

[0044] In some embodiments, the method further comprises:

[0045] The structural equation model was used to check and optimize the wave field data for the entire region;

[0046] The structural equation model was repeatedly introduced into confirmatory factor analysis during the process of checking and optimizing.

[0047] To achieve the above object, another aspect of an embodiment of the present invention provides a wave field reconstruction device, the device comprising:

[0048] The first module is used to obtain the wave field data of the target area and fill the missing values ​​of the wave field data using the connection function to obtain a complete data set;

[0049] The second module is used to obtain the wave field reanalysis data of the target area, and to obtain the ocean model reanalysis data based on the wave field reanalysis data;

[0050] The third module is used to fuse the complete data set and the ocean model reanalysis data based on the diffusion model to obtain the target data set;

[0051] The fourth module is used to classify the wave field using the pulse neural network according to the target data set to obtain the type classification result of the target area;

[0052] The fifth module is used to reconstruct the wave field of each type in the target area by region based on the type classification result and a preset reconstruction method;

[0053] The sixth module is used to connect and merge different regions based on the results of regional reconstruction using the maximum cross-correlation method to obtain the reconstruction results of the wave field in the entire region.

[0054] In some embodiments, the apparatus further comprises:

[0055] The seventh module is used to check and optimize the wave field data of the whole area using structural equation model;

[0056] The structural equation model was repeatedly introduced into confirmatory factor analysis during the process of checking and optimizing.

[0057] To achieve the above objective, another aspect of an embodiment of the present invention provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above method when executing the computer program.

[0058] To achieve the above objective, another aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0059] The embodiments of the present invention include at least the following beneficial effects: the present invention provides a wave field reconstruction method, device, electronic device and medium, which obtains the wave field data of the target area, fills the missing values ​​of the wave field data with the connection function, and obtains the complete data set; obtains the wave field reanalysis data of the target area, and obtains the ocean model reanalysis data according to the wave field reanalysis data; based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain the target data set; according to the target data set, the wave field is classified by the pulse neural network to obtain the type classification result of the target area; according to the type classification result, the wave field of each type in the target area is reconstructed by the preset reconstruction method; based on the result of the regional reconstruction, the maximum cross-correlation method is used to connect and fuse different regions to obtain the reconstruction result of the wave field in the whole region. The present invention uses multi-source data fusion technology to analyze and process data collected from different fields or data sources, which can improve the shortcomings of inaccurate and incomplete data commonly existing in a single data source, increase the accuracy of data collection, and obtain relatively accurate and complete reconstruction results. In addition, the present invention can achieve a large range of wave field reconstruction by reconstructing the wave field by region while ensuring accuracy. The present invention can efficiently reconstruct the wave field. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of a wave field reconstruction method provided by an embodiment of the present invention;

[0061] Figure 2 is a multi-source data fusion flow chart provided by an embodiment of the present invention;

[0062] Figure 3 is a flow chart of wave field regional reconstruction provided by an embodiment of the present invention;

[0063] Figure 4 This is an example diagram of water depth division areas of a certain sea area provided by an embodiment of the present invention;

[0064] Figure 5 is a flow chart of nearshore wave field reconstruction provided by an embodiment of the present invention;

[0065] Figure 6 is a global data inspection and optimization flow chart provided by an embodiment of the present invention;

[0066] Figure 7 is a schematic structural diagram of a wave field reconstruction device provided by an embodiment of the present invention;

[0067] Figure 8 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present invention. They are only examples of devices and methods consistent with some aspects of the embodiments of the present invention as detailed in the attached claims.

[0069] It is understood that the terms "first", "second", etc. used in the present invention may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0070] The terms "at least one", "multiple", "each", "any", etc. used in the present invention, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0071] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0072] The wave field reconstruction method provided by the embodiment of the present invention relates to the field of data processing technology. The wave field reconstruction method provided by the embodiment of the present invention can be applied to a terminal, can also be applied to a server, and can also be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or it can be configured as a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements the wave field reconstruction method, etc., but is not limited to the above forms.

[0073] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0074] Figure 1 is an optional flow chart of the wave field reconstruction method provided by an embodiment of the present invention. Figure 1 The method may include but is not limited to steps S100 to S600.

[0075] S100, obtaining wave field data of the target area, and using the connection function to fill the missing values ​​of the wave field data to obtain a complete data set;

[0076] It should be noted that, in some embodiments, using the connection function to fill missing values ​​in the wave field data to obtain a complete data set may include: pre-filling the missing values ​​of the original wave field data by the difference method; estimating the missing values ​​by the t-Copula function according to the correlation between the variables in the wave field data after pre-filling; wherein the parameters of the t-Copula function are obtained by processing with the maximum likelihood method; and filling the missing values ​​of the original wave field data according to the result of the missing value estimation to obtain a complete data set.

[0077] For example, in some specific implementations, wave field data are first acquired and preprocessed: wave field reconstruction usually requires ocean wave data, marine meteorological data, seafloor topography data, and satellite image data, etc. The ocean wave data and marine meteorological data can be obtained from marine meteorological departments, research institutions, and ocean observation stations; the seafloor topography data can be obtained from geographic information system (GIS) data providers; and satellite image data are usually provided by the National Space Administration or satellite operators.

[0078] For the missing value problem that may appear in the acquired wave field related data, the present invention uses the Copula function family to fill the missing values. This method estimates the missing values ​​by establishing a multivariate distribution function. The specific process is as follows:

[0079] ① Check whether there are missing values ​​in the data, and determine the location and number of missing values; this can be achieved through relevant detection tools, which is not the technical focus of the present invention and will not be described in detail;

[0080] ② Select a suitable copula function based on the correlation between variables. Considering that wave field data is usually continuous data and is affected by multiple factors such as circulation and wind, it usually does not conform to a strict normal distribution. Therefore, the t-Copula function is selected in this step. This function is similar to the Gaussian Copula, but has better robustness to outliers and can better handle data that does not conform to the normal distribution;

[0081] ③ Handle missing values ​​and use interpolation methods to fill in missing values ​​for variables with missing values;

[0082] ④ Based on the existing data, the maximum likelihood estimation method is used to estimate the parameters of the t-copula function. The parameters that need to be estimated for the T-Copula function include the correlation coefficient ρ and the degrees of freedom v. The correlation coefficient is used to represent the linear correlation between variables, while the degrees of freedom affect the sensitivity of the t-Copula function to extreme values.

[0083] ⑤ According to Copula theory, for any sequence u, if there is a joint distribution F, then there is a unique Copula function C(x) for the sequence u:

[0084] F(u1,u2...u T )=C(F1(u1),F2(u2)...F T (u T ))

[0085] Among them, C(x) is the Copula function, and t-Copula is used in this step.

[0086] ⑥ Log-likelihood function of t-Copula:

[0087]

[0088] Where ρ is the correlation coefficient; v is the degree of freedom of the distribution function F, which represents the number of independent observations in the sample data minus the number of estimated parameters; T represents the length of the sequence, t represents the tth position of the sequence, and u t Represents the value of the tth position in the sequence; Among them, y t ′ is y tThe transpose of ; i represents the i-th variable. This formula takes a double variable as an example, so the value of i is 1, 2; y it represents the y of the i-th variable t ; represents the inverse function of the Student t distribution function with v degrees of freedom; Γ() represents the gamma function.

[0089] ⑦ Missing value estimation: For missing values ​​that still exist after filling, a certain number of random samples are generated using the fitted t-Copula model to simulate the possible distribution of missing values, and the estimated values ​​of the missing values ​​are extracted from them;

[0090] ⑧Fill the generated missing values ​​into the original data to obtain a complete data set.

[0091] S200, obtaining wave field reanalysis data of the target area, and obtaining ocean model reanalysis data according to the wave field reanalysis data;

[0092] For example, in some specific implementations, the acquisition of wave field reanalysis data is achieved as follows: Reanalysis uses data assimilation technology to fuse and optimally integrate observational data from various sources and types with numerical forecast products, which can reconstruct long-term historical data and solve the problem of uneven temporal and spatial distribution of observational data. The ocean reanalysis data commonly used in ocean model driving fields include the International Comprehensive Ocean-Atmosphere Data Set ICOADS, the National Center for Environmental Prediction / National Center for Atmospheric Research, and the European Center for Medium-Range Weather Forecasts ECMWF data. The ocean model reanalysis data calculated by the assimilation method include the SODA ocean reanalysis data set, OFES data, etc.

[0093] S300, based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain the target data set;

[0094] It should be noted that, in some embodiments, step S300 may include: normalizing the complete data set and the ocean model through standard scores; performing principal component analysis on the normalized data based on a preset feature selection method to obtain the principal component characteristics of each data source; establishing a diffusion model by using partial differential equations as wave propagation equations; calculating the similarity between different data sources using Euclidean distance, and determining the diffusion rule based on the similarity; based on the diffusion model, using the finite difference method for discretization processing to solve the spatiotemporal distribution of the wave field, and adjusting the information propagation parameters according to the diffusion rule; based on the relative importance of the propagated information and the original observation data of the complete data set and the ocean model reanalysis data, using weighted average to perform data fusion to obtain the target data set.

[0095] For example, in some specific embodiments, Figure 2 As shown in Figure 2, the specific steps of multi-source data fusion are as follows:

[0096] 1. Normalization processing:

[0097] Before fusing multi-source data, it is necessary to unify the value ranges of different types of data and perform normalization. The present invention adopts a standardized approach to represent data through standard scores, i.e.

[0098]

[0099] Among them, δ is the data mean and ω is the standard deviation.

[0100] In the Z-score standardization with a mean of 0, the variance of the original data is normalized to form new data, and each dimension has the same value dimension, so all belong to the normal distribution with a mean of 0 and a variance of 1, which is conducive to subsequent analysis and processing.

[0101] 2. Feature selection:

[0102] After normalization, the feature dimensions of various data are still different. The present invention uses the improved feature selection method IPCA (Incremental PCA) to find the principal components in the data to reduce the dimension of the data while retaining as much information as possible. For the existing data matrix X, calculate its covariance matrix C, that is,

[0103]

[0104] Where N is the number of samples, is the mean vector of each column of the data matrix X, and T represents transpose.

[0105] Subsequently, the covariance matrix C is eigenvalue decomposed to obtain eigenvalues ​​and eigenvectors. In each iteration, IPCA can update the principal component P through incremental learning:

[0106]

[0107] Among them, P new is the updated principal component, P old is the principal component before updating, X batch is the data of the current batch, is the mean vector of the current batch of data, and α is the learning rate.

[0108] 3. Diffusion model:

[0109] In the process of multi-source data fusion, the present invention establishes a diffusion model, uses partial differential equations to describe the propagation and evolution of ocean waves, uses data from different observation sites and satellites to calibrate and optimize the model, and finally obtains more accurate results by fusing the diffusion model with the observation data.

[0110] The specific method is as follows:

[0111] ① Establish a diffusion model. Select the BBM equation, the full name of which is the Benjamin-Bona-Mahony equation, which is a nonlinear partial differential equation used to simulate small-amplitude long waves in fluid mechanics. As the wave propagation equation, the boundary conditions of the model are reasonably set according to the actual situation, including the ocean boundary, land boundary, etc., and the diffusion term is introduced into the wave propagation equation. Combined with the nonlinear term in the propagation equation, it can describe the propagation and diffusion of waves;

[0112] BBM equation: u t +uu x -u xxt =0;

[0113] ② Calculate similarity. Use Euclidean distance to calculate the similarity between different data sources;

[0114] Euclidean distance:

[0115] Similarity:

[0116] ③ Define diffusion rules. According to the similarity matrix, define the rules for information dissemination, that is, observations with high similarity have a larger diffusion weight, and observations with low similarity have a smaller diffusion weight;

[0117] ④ Information propagation. The established diffusion model is used to simulate the propagation and diffusion process of ocean waves. The finite difference method is used to discretize the equations and solve the spatial and temporal distribution of the wave field. According to the defined diffusion rules, the direction and intensity of information propagation are adjusted to make the information propagation stronger between observations with high similarity.

[0118] ⑤ Data fusion: According to the relative importance of the disseminated information and the original observation data, weighted average is used for data fusion.

[0119] S400, based on the target data set, using a pulse neural network to classify the wave field, and obtain a type classification result of the target area;

[0120] It should be noted that, in some embodiments, step S400 may include: according to the target data set, using a statistical method to obtain the wave characteristics of each area in the target area; the wave characteristics include wave height and wave period; the wave characteristics are input into a pulse neural network for wave field classification to obtain the type classification result of the target area; wherein the pulse neural network includes an input layer, a hidden layer and an output layer; the hidden layer is constructed based on the Izhikevich neuron model, and the number of neurons in the output layer is the same as the total number of wave field types; the pulse neural network is obtained by network training and parameter adjustment through stochastic gradient descent and mean square error based on the wave field data set with labeled wave field types.

[0121] For example, in some specific embodiments, Figure 3 As shown in Figure 2, the implementation process of wave field classification is as follows:

[0122] Considering that wave fields in different regions have different characteristics, the present invention will use pulse neural networks to classify wave fields and output corresponding classification results. The training process of pulse neural networks is as follows:

[0123] ① Data preprocessing has been completed in the previous steps to ensure the quality and consistency of the data;

[0124] ② Divide the data into training set, test set and validation set;

[0125] ③ Based on the water depth data, the research sea area is divided into four areas according to the water depth range: estuary area (0-10m), nearshore area (10-50m), offshore area (50-200m), and deep sea area (above 200m) (such as Figure 4 As shown in the figure, taking a certain sea area as an example, the area is divided according to the water depth), and the wave characteristics of each area are determined by statistical methods, such as the average wave height, average period, etc.;

[0126] ④Neural modeling. Considering the continuity, dynamics and nonlinearity of wave data, the Izhikevich neuron model is selected in this step. The main thing to do during the construction process is to determine the network structure and parameter settings.

[0127] Input layer: wave height and wave period are used as input features;

[0128] Hidden layer: The hidden layer is constructed using the Izhikevich neuron model, which includes two differential equations that describe the changes in the membrane potential and membrane conductance of the neuron. In this step, cross-validation is introduced to determine the number of hidden layer neurons. By trying different numbers of hidden layer neurons, cross-validation is used to evaluate the performance of the model on the test set and select the combination with the best performance;

[0129] Output layer: The number of neurons in the output layer is determined according to the divided areas. Each neuron corresponds to one area. This method divides the wave field into four areas, so the output layer requires four neurons.

[0130] ⑤ Use random initialization method to initialize the network weights and bias coefficients;

[0131] ⑥ Use the training set to train the constructed spiking neural network, and pay attention to using time series coding to convert the input samples into pulse signals. During the training process, in order to improve the accuracy of the model output results, select stochastic gradient descent as the optimization algorithm and mean square error as the loss function;

[0132] ⑦ Use the validation set to evaluate the performance of the trained model by calculating indicators such as classification accuracy, recall rate, and precision, and tune the network structure and hyperparameters based on the evaluation results to improve the generalization ability and performance of the model;

[0133] ⑧The output of the model is a discrete category label and a probability distribution corresponding to each category.

[0134] It should be noted that the data processing principle of the pulse neural network can be clearly understood according to the training process of the pulse neural network, so the logic of using the pulse neural network for wave field classification will not be repeated.

[0135] S500, reconstructing the wave field of each type in the target area by region based on a preset reconstruction method according to the type classification result;

[0136] It should be noted that the type classification results include estuary areas, nearshore areas, offshore areas and deep-sea areas; in some embodiments, step S500 may include: for the estuary area, pre-constructing an ocean dynamics model, and then reconstructing the first wave field in combination with a numerical simulation method; wherein the ocean dynamics model is obtained by combining an ocean-river interaction model, a tidal model and an ocean wave model through a coupling method; for the nearshore area, extracting wave parameters from radar images in the target data set, and then reconstructing the second wave field by combining a convolutional autoencoder and a dual attention mechanism; for the offshore area, using a high-order spectral method to reconstruct the third wave field; for the deep-sea area, establishing a cellular automaton model based on the remote sensing data in the target data set to reconstruct the fourth wave field.

[0137] For example, in some specific embodiments, Figure 3 As shown in the figure, the implementation process of regional reconstruction is as follows:

[0138] Based on the classification results of the pulse neural network in the previous step, the wave field is reconstructed by region. The specific method is as follows:

[0139] 1) Estuary area: Due to the input of rivers and the influence of tides, the wave field in the estuary area is relatively complex. Therefore, for the reconstruction of the wave field in the estuary area, in order to obtain more accurate reconstruction results, the present invention combines the ocean-river interaction model, the tidal model and the wave model, and considers the coupling effect of runoff-tide-wave. The specific method is as follows:

[0140] ① Independently establish the ocean-river interaction model, tidal model and ocean wave model, and combine them together through coupling methods to form a comprehensive ocean dynamics model;

[0141] ② The ocean-river interaction model mainly considers the interaction process between the ocean and the river, and uses numerical simulation methods to simulate the transmission and exchange of parameters such as water flow, salinity, and temperature;

[0142] ③ The tidal model is used to simulate the formation and propagation of tides, and uses the tidal equation or tidal potential equation to describe the changing law of tides;

[0143] ④ The ocean wave model is used to simulate the process of wave generation and propagation. Linear wave theory or nonlinear wave theory can be used to describe the law of wave changes;

[0144] ⑤ The model is coupled through numerical calculation methods (iterative solution or coupled solution) to obtain the reconstruction result of the wave field.

[0145] 2) Nearshore area: In the nearshore area, the convolutional autoencoder and the dual attention mechanism are combined to extract wave parameters from radar images, making full use of the advantages of the two models to improve the accuracy and robustness of wave parameter extraction. Figure 5 As shown, the specific method is as follows:

[0146] ① Data preprocessing: Preprocess the collected radar image data, including normalization, cropping, resizing, etc., to adapt to the input format of the convolutional neural network;

[0147] ②Encoder design. This part uses convolutional neural networks to extract image features. The convolutional neural network architecture is designed, including multiple convolutional layers, pooling layers, and fully connected layers. In addition, the appropriate convolution kernel size, step size, and padding method are selected in combination with the characteristics of the wave field to capture the wave structure information in the image.

[0148] ③Encoder training. Use the training set to train the encoder and optimize the model parameters by minimizing the reconstruction loss;

[0149] ④ Feature extraction. Use the trained encoder to encode each image in the training set, validation set, and test set to obtain the feature representation of the image;

[0150] ⑤Decoder design. The decoder includes the deconvolution layer (core), the depooling layer, and the output layer. A dual attention mechanism is built in the deconvolution layer.

[0151] ⑥ Construction of dual attention mechanism. Design global attention mechanism and local attention mechanism to capture global and local wave features respectively;

[0152] ⑦Use the training set with labeled wave parameters, select the mean square error as the loss function to train the dual attention mechanism model, and use the trained dual attention mechanism to further extract wave parameters from the features extracted by the encoder;

[0153] ⑧ Model evaluation and optimization. Use the test set to evaluate the performance of the entire system in wave parameter extraction, and adjust and optimize the parameters through cross-validation method based on the evaluation results.

[0154] 3) Offshore area: In offshore areas, high-order spectrum methods are used to reconstruct wave fields. The specific methods are as follows:

[0155] ① Perform Fourier transform on the pre-processed wave field data for spectrum analysis. Use Fast Fourier Transform (FFT) to calculate the first-order spectrum and second-order spectrum. The first-order spectrum describes the change of wave height with frequency, and the second-order spectrum represents the change of wave energy with frequency;

[0156] ② Calculate the third-order spectrum. The third-order spectrum is the double Fourier transform of the third-order moment function, which mainly provides the directional information of the wave field. The third-order spectrum can be estimated by using the classical estimation method of the spectrum. The specific method is as follows:

[0157] Segment the observed data sequence;

[0158] Preprocess the segmented data to have a mean of zero;

[0159] Make third-order moment estimates for each segment of data;

[0160] Find the mean of all the third-order moment estimates to form a third-order spectrum estimate;

[0161] ③ Analyze the third-order spectrum to obtain the directional distribution characteristics of the wave field. The propagation direction of the wave field can be determined by observing the main direction of the third-order spectrum. Use peak recognition to further analyze the peak and distribution of the third-order spectrum to determine the main direction, secondary direction and intensity of the directional distribution of the wave.

[0162] ④ Combining the wave height and period information of the first-order spectrum and the second-order spectrum, the wave field is reconstructed according to the wave propagation direction and wave energy distribution using the interpolation method;

[0163] ⑤ Compare the characteristics of the reconstructed wave spectrum with the actual observed spectrum, such as the main frequency, energy distribution, etc., to see if there are obvious differences. If there are differences, reconstruct and compare again by adjusting the parameters of the interpolation method.

[0164] 4) Deep and deep sea area: Considering that the deep and deep sea area is far away from the land, a cellular automation model is established based on remote sensing data to reconstruct the wave field in the deep and deep sea area. The specific method is as follows:

[0165] ① Data preparation. Use satellite remote sensing data to obtain wave data and wind field data in deep sea areas. Seabed topography information can be obtained from topographic survey data or satellite altimetry data;

[0166] ② Cell division and initialization. The study area is divided into cell grids. The grid size is determined by the spatial resolution of the remote sensing data. The wave and wind field information are mapped to the state value of the corresponding cell to complete the initialization of the cell state.

[0167] ③Iterative update rule definition:

[0168] Neighbor relationship determination: Each cell has 5*5 neighboring cells as its domain cells;

[0169] Update function definition: The deep sea area has the characteristics of small wave amplitude and large water depth, which meets the applicable conditions of linear wave theory. Therefore, in this step, the update function is defined based on the linear wave theory. The specific function is as follows:

[0170] Wave height update: H(t+Δt)=H(t)+ΔH wind ;

[0171] Wave Speed ​​Update:

[0172] Wavelength update: L(t+Δt)=c(t+Δt)*T(t);

[0173] Wave propagation direction update: θ(t+Δt)=θ wind +π;

[0174] Among them, H(t) is the wave height of the current time step, H(t+Δt) is the wave height of the next time step, and ΔH wind is the change in wave height caused by wind; c(t+Δt) is the wave speed in the next time step; T(t) is the wave period in the current time step; L(t+Δt) is the wavelength in the next time step; θ(t+Δt) is the wave direction in the next time step, θ wind For wind direction.

[0175] Note: The wave period will be automatically updated with the update of wave speed and wavelength. Generally, the wave speed and wavelength of the next time step are calculated based on the wave period of the current time step.

[0176] Update order determination: that is, determine the order in which the cell states are updated in each time step. In this step, a random order is used for updating;

[0177] ④ End condition setting. You can determine the end condition of the simulation by setting the simulation time period. In addition, you can control the simulation accuracy and calculation cost by setting the simulation time step and the number of updates per time step.

[0178] ⑤Result output and verification. Generate wave field reconstruction data based on simulation results and remote sensing data and perform visualization, and compare with remote sensing data to verify the accuracy of the results;

[0179] ⑥ Model optimization and iteration: According to the verification results and actual application requirements, the model parameters are adjusted to improve the applicability and accuracy of the model.

[0180] S600, based on the results of regional reconstruction, the maximum cross-correlation method is used to connect and merge different regions to obtain the reconstruction result of the wave field in the whole region;

[0181] It should be noted that, in some embodiments, step S600 may include: calculating a cross-correlation function for adjacent wave field data in the results of regional reconstruction, and determining a connection position according to a peak value of the cross-correlation function; connecting and fusing the wave field data of adjacent regions according to the connection position, and interpolating and smoothing the connection position to obtain a wave field reconstruction result for the entire region.

[0182] For example, in some specific embodiments, Figure 3 As shown in the figure, the process of regional connection and integration is as follows:

[0183] After obtaining the reconstruction results of each region, the maximum cross-correlation method is used to connect and fuse the regions. The maximum cross-correlation method is a method that determines the similarity between two signals by calculating the cross-correlation function between them. In the ocean wave field data, the maximum cross-correlation function can determine the best matching position of the wave field between adjacent regions and realize the connection and fusion of data.

[0184] The specific method is as follows:

[0185] ① For adjacent wave field data, treat them as two signals respectively, calculate the cross-correlation function between them, and determine the best connection position by finding the peak of the cross-correlation function;

[0186] ② After determining the best connection position, the wave field data of adjacent areas are connected and merged, and the connection is processed using interpolation, smoothing and other methods to ensure the continuity and consistency of the data;

[0187] ③ Evaluate the wave field data after connection and fusion, adjust parameters and optimize algorithms in the connection and fusion process, and ensure that the fused wave field data meets actual needs.

[0188] In some embodiments, the method may further include: using a structural equation model to check and optimize the wave field data of the entire region; wherein the structural equation model is repeatedly introduced into confirmatory factor analysis during the checking and optimization process.

[0189] For example, Figure 3 As shown, in some specific implementations, Figure 6 As shown, the inspection and optimization of the wave field data in the whole area are implemented as follows:

[0190] After completing the reconstruction of the wave field in the entire region, the structural equation model is used to check and optimize the wave field data in the entire region, and confirmatory factor analysis is introduced to verify the validity of the model. The specific methods are as follows:

[0191] ① Specify the observed variables of each latent variable. For example, if the latent variable is wave height, the observed variable is the measured wave height (i.e., the peak height of the waves measured by buoys or other devices);

[0192] ② Use structural equation model software to develop a confirmatory factor analysis model and estimate it. During the process, it is necessary to check indicators such as factor loading and inter-factor correlation, and evaluate the fitness of the model based on fitness indicators such as fit index, and continuously adjust parameters and re-estimate until the fitness requirements are met;

[0193] ③ Add the relationship between latent variables to the model and specify the path, use structural equation model software to estimate the structural model parameters and evaluate the fit of the structural model;

[0194] ④ Use statistical indicators such as fitting index and stepwise fitting index to evaluate the fitness of the model, and make continuous adjustments and re-estimations;

[0195] ⑤During the inspection and optimization process, confirmatory factor analysis is repeated to verify the validity and credibility of the model.

[0196] In summary, in view of the shortcomings of the prior art, the present invention makes the following improvements:

[0197] 1) The present invention uses multi-source data fusion technology to analyze and process data collected from different fields or data sources, which can improve the shortcomings of inaccurate and incomplete data that are common in a single data source, increase the accuracy of data collection, and obtain relatively accurate and complete reconstruction results;

[0198] 2) The present invention reconstructs the wave field in different regions and uses different reconstruction methods in combination with the characteristics of the wave field in each region. Under the premise of ensuring accuracy, the wave field reconstruction in a larger range can be achieved;

[0199] 3) In the reconstruction of the wave field in the estuary area, the coupling effect of runoff, tide and wave is taken into account, which improves the accuracy of the reconstruction results of the wave field in the estuary area;

[0200] 4) In the reconstruction of the wave field in the nearshore area, the dual attention mechanism and window moving method are introduced to assist in the extraction of wave parameters, which not only improves the calculation speed but also improves the accuracy of the extracted wave parameters;

[0201] 5) In the reconstruction of the wave field in the offshore area, the third-order spectrum is used for calculation, which not only avoids the problem of too high order and large amount of calculation, but also reflects the third-order interaction relationship between different frequency components in the wave field, including resonance between waves, nonlinear interaction, etc., to meet the accuracy requirements of the reconstruction of the wave field in the offshore area;

[0202] 6) In the reconstruction of the wave field in deep sea areas, a cellular automaton model is established based on remote sensing data, which avoids the difficulty in obtaining wave observation data in deep sea areas. At the same time, the linear wave theory is used in combination with the characteristics of the wave field in deep sea areas, which greatly reduces the amount of calculation;

[0203] 7) In the process of checking and optimizing the reconstruction results, confirmatory factor analysis combined with structural equation modeling was introduced for evaluation, which effectively verified the validity and credibility of the model.

[0204] Compared with the prior art, the present invention has at least the following beneficial effects:

[0205] The present invention uses multi-source data fusion technology to analyze and process data collected from different fields or data sources, which can improve the common shortcomings of inaccurate and incomplete data from a single data source, increase the accuracy of data collection, and obtain relatively accurate and complete reconstruction results.

[0206] The present invention reconstructs the wave field in different regions, and uses different reconstruction methods in combination with the characteristics of the wave field in each region, so that a larger range of wave field reconstruction can be achieved under the premise of ensuring accuracy; in the reconstruction of the wave field in the estuary area, the coupling effect of runoff-tide-wave is considered, and the accuracy of the reconstruction result of the wave field in the estuary area is improved; in the reconstruction of the wave field in the nearshore area, a double attention mechanism and a window moving method are introduced to assist in the extraction of wave parameters, which not only improves the calculation speed, but also improves the accuracy of the extracted wave parameters; in the reconstruction of the wave field in the offshore area, the third-order spectrum is used for calculation, which not only avoids the problem of too high order and large amount of calculation, but also can reflect the third-order interaction relationship between different frequency components in the wave field, including resonance between waves, nonlinear interaction, etc., to meet the accuracy requirements of the reconstruction of the wave field in the offshore area; in the reconstruction of the wave field in the deep and deep sea area, a cellular automaton model is established based on remote sensing data, which avoids the problem of difficulty in obtaining wave observation data in the deep and deep sea area, and at the same time, the linear wave theory is used to define the update function in combination with the characteristics of the wave field in the deep and deep sea area, which greatly reduces the amount of calculation. Finally, in the process of checking and optimizing the reconstruction results, confirmatory factor analysis was introduced in combination with the structural equation model for evaluation, which effectively verified the validity and credibility of the model.

[0207] See also Figure 7 The embodiment of the present invention further provides a wave field reconstruction device 900, which can implement the above wave field reconstruction method, and the device includes:

[0208] The first module 910 is used to obtain the wave field data of the target area, and fill the missing values ​​of the wave field data using the connection function to obtain a complete data set;

[0209] The second module 920 is used to obtain wave field reanalysis data of the target area, and to obtain ocean model reanalysis data according to the wave field reanalysis data;

[0210] The third module 930 is used to perform data fusion on the complete data set and the ocean model reanalysis data based on the diffusion model to obtain the target data set;

[0211] The fourth module 940 is used to classify the wave field using a pulse neural network according to the target data set to obtain a type classification result of the target area;

[0212] The fifth module 950 is used to reconstruct the wave field of each type in the target area by region based on the type classification result and a preset reconstruction method;

[0213] The sixth module 960 is used to connect and merge different regions based on the results of regional reconstruction using the maximum cross-correlation method to obtain the reconstruction results of the wave field in the entire region.

[0214] In some embodiments, the apparatus may further include:

[0215] The seventh module is used to check and optimize the wave field data of the whole area using structural equation model;

[0216] The structural equation model was repeatedly introduced into confirmatory factor analysis during the process of checking and optimizing.

[0217] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0218] The embodiment of the present invention further provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above wave field reconstruction method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0219] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0220] See also Figure 8 , Figure 8 The hardware structure of an electronic device 1000 of another embodiment is illustrated. The electronic device includes:

[0221] The processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention;

[0222] The memory 1002 may be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 may store an operating system and other application programs. When the technical solution provided in the embodiments of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1002, and the processor 1001 calls and executes the wave field reconstruction method of the embodiment of the present invention;

[0223] Input / output interface 1003, used to implement information input and output;

[0224] The communication interface 1004 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0225] A bus 1005 , which transmits information between various components of the device (e.g., the processor 1001 , the memory 1002 , the input / output interface 1003 , and the communication interface 1004 );

[0226] The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0227] An embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the wave field reconstruction method described above is implemented.

[0228] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0229] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0230] The wave field reconstruction method, wave field reconstruction device, electronic device and medium provided by the embodiment of the present invention obtain the wave field data of the target area, fill the missing values ​​of the wave field data by using the connection function, and obtain the complete data set; obtain the wave field reanalysis data of the target area, and obtain the ocean model reanalysis data according to the wave field reanalysis data; based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain the target data set; according to the target data set, the wave field is classified by using the pulse neural network to obtain the type classification result of the target area; according to the type classification result, each type of wave field in the target area is reconstructed by region based on the preset reconstruction method; based on the result of the region reconstruction, the maximum cross-correlation method is used to connect and fuse different regions to obtain the reconstruction result of the wave field in the whole region. The present invention uses multi-source data fusion technology to analyze and process data collected from different fields or data sources, which can improve the shortcomings of inaccurate and incomplete data commonly existing in a single data source, increase the accuracy of data acquisition, and obtain relatively accurate and complete reconstruction results. In addition, the present invention can achieve wave field reconstruction in a larger range by reconstructing the wave field by region while ensuring accuracy. The present invention can efficiently reconstruct the wave field.

[0231] The embodiments described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art can appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0232] Those skilled in the art will appreciate that the technical solutions shown in the figures do not limit the embodiments of the present invention and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0233] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0234] Those skilled in the art will appreciate that all or some of the steps, devices, and functional modules / units in the above disclosed methods may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0235] The terms "first", "second", "third", "fourth", etc. (if any) in the description of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0236] It should be understood that in the present invention, "at least one (item)" refers to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can represent: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0237] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0238] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0239] In addition, each functional unit in each embodiment of the present invention 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0240] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including multiple instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store programs.

[0241] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the embodiments of the present invention is not limited thereby. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present invention shall be within the scope of the rights of the embodiments of the present invention.

Claims

1. A wave field reconstruction method, characterized in that: The method comprises: Acquire the wave field data of the target area, and use the connection function to fill the missing values ​​of the wave field data to obtain a complete data set; Acquire wave field reanalysis data of the target area, and obtain ocean model reanalysis data according to the wave field reanalysis data; Based on the diffusion model, data fusion is performed on the complete data set and the ocean model reanalysis data to obtain a target data set; Wherein, based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain a target data set, including: normalizing the complete data set and the ocean model by a standard score; Performing principal component analysis on the normalized data based on a preset feature selection method to obtain principal component features of each data source; The diffusion model is established by using partial differential equations as wave propagation equations; Calculating similarities between different data sources using Euclidean distance, and determining diffusion rules based on the similarities; Based on the diffusion model, the finite difference method is used for discretization processing to obtain the spatiotemporal distribution of the wave field, and the information propagation parameters are adjusted according to the diffusion rule; Based on the relative importance of the propagated information and the original observation data of the complete data set and the ocean model reanalysis data, a weighted average is used to perform data fusion to obtain a target data set; According to the target data set, a pulse neural network is used to classify the wave field to obtain a type classification result of the target area; According to the type classification result, reconstructing each type of wave field in the target area by region based on a preset reconstruction method; Based on the results of the regional reconstruction, the maximum cross-correlation method is used to connect and merge different regions to obtain the reconstruction result of the wave field in the entire region.

2. The method according to claim 1, characterized in that The use of the connection function to fill missing values ​​in the wave field data to obtain a complete data set includes: Pre-filling missing values ​​of the original wave field data by using the difference method; According to the correlation between the variables in the wave field data after the pre-filling, the missing values ​​are estimated by the t-Copula function; Wherein, the parameters of the t-Copula function are obtained by maximum likelihood method; The missing values ​​of the original wave field data are filled according to the result of the missing value estimation to obtain a complete data set.

3. The method according to claim 1, characterized in that The wave field classification is performed using a pulse neural network according to the target data set to obtain a type classification result of the target area, including: According to the target data set, using a statistical method to obtain wave characteristics of each area in the target area; the wave characteristics include wave height and wave period; Inputting the wave characteristics into the pulse neural network to classify the wave field, and obtaining the type classification result of the target area; The pulse neural network includes an input layer, a hidden layer and an output layer; the hidden layer is constructed based on the Izhikevich neuron model, and the number of neurons in the output layer is the same as the total number of types of wave fields; The pulse neural network is obtained by performing network training and adjusting parameters through stochastic gradient descent and mean square error according to a wave field data set with labeled wave field types.

4. The method according to claim 1, characterized in that: The type classification results include estuary area, nearshore area, offshore area and deep sea area; and the wave field of each type in the target area is reconstructed by region based on the preset reconstruction method according to the type classification results, including: For the estuary area, an ocean dynamics model is pre-constructed, and then a first wave field is reconstructed in combination with a numerical simulation method; wherein the ocean dynamics model is obtained by combining an ocean-river interaction model, a tidal model, and an ocean wave model through a coupling method; For the nearshore area, wave parameters are extracted from the radar images in the target dataset, and then a second wave field is reconstructed by combining a convolutional autoencoder and a dual attention mechanism; For offshore areas, the third wave field is reconstructed using a high-order spectral method; For the deep sea area, a cellular automaton model is established based on the remote sensing data in the target data set to reconstruct the fourth wave field.

5. The method according to claim 1, characterized in that Based on the results of the regional reconstruction, the maximum cross-correlation method is used to connect and merge different regions to obtain the reconstruction results of the wave field in the entire region, including: Calculating a cross-correlation function for adjacent wave field data in the result of the sub-region reconstruction, and determining a connection position according to a peak value of the cross-correlation function; The wave field data of adjacent areas are connected and merged according to the connection positions, and the connection positions are interpolated and smoothed to obtain the wave field reconstruction result of the entire area.

6. The method according to claim 1, characterized in that The method further comprises: The structural equation model was used to check and optimize the wave field data for the entire region; The structural equation model was repeatedly tested and optimized by introducing confirmatory factor analysis.

7. A wave field reconstruction device, characterized in that: The device comprises: The first module is used to obtain the wave field data of the target area, and fill the missing values ​​of the wave field data using the connection function to obtain a complete data set; The second module is used to obtain the wave field reanalysis data of the target area, and to obtain the ocean model reanalysis data according to the wave field reanalysis data; The third module is used to perform data fusion on the complete data set and the ocean model reanalysis data based on a diffusion model to obtain a target data set; Wherein, based on the diffusion model, the complete data set and the ocean model reanalysis data are fused to obtain a target data set, including: normalizing the complete data set and the ocean model by a standard score; Performing principal component analysis on the normalized data based on a preset feature selection method to obtain principal component features of each data source; The diffusion model is established by using partial differential equations as wave propagation equations; Calculating similarities between different data sources using Euclidean distance, and determining diffusion rules based on the similarities; Based on the diffusion model, the finite difference method is used for discretization processing to obtain the spatiotemporal distribution of the wave field, and the information propagation parameters are adjusted according to the diffusion rule; Based on the relative importance of the propagated information and the original observation data of the complete data set and the ocean model reanalysis data, a weighted average is used to perform data fusion to obtain a target data set; The fourth module is used to classify the wave field using a pulse neural network according to the target data set to obtain a type classification result of the target area; A fifth module is used to reconstruct the wave field of each type in the target area by region based on the type classification result and a preset reconstruction method; The sixth module is used to connect and merge different regions based on the results of the sub-regional reconstruction using the maximum cross-correlation method to obtain the reconstruction results of the wave field in the entire region.

8. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images