Method, system and equipment for predicting tidal current velocity in ocean and medium

Abstract

The invention discloses an ocean inner tidal current velocity prediction method, system and device and a medium. The method comprises the following steps: acquiring historical inner tidal current velocity data of a target sea area; and inputting the historical internal tidal current velocity data into a preset internal tidal current velocity prediction model to obtain an internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model, so that the problem of large prediction error caused by the fact that a traditional internal tidal current prediction method based on harmonic analysis cannot predict an incoherent internal tidal current part can be solved. And meanwhile, an internal tide flow velocity prediction model can be trained by using relatively few training resources, so that a full-depth high-precision internal tide prediction result of different sea area conditions can be obtained through prediction.

Description

Technical Field

[0001] This application relates to the field of ocean internal tide prediction, and in particular to a method, system, device and medium for predicting ocean internal tide current velocity. Background Art

[0002] Internal tide is an internal wave with a tidal period generated by the interaction between astronomical tides and topography, which has a significant impact on ocean engineering facilities, underwater acoustic communication, submarine underwater navigation, etc. Many risks can be avoided through accurate prediction. During the propagation of internal tide, factors such as ocean stratification, background circulation, and mesoscale eddy turbulence have an impact on it, and there are incoherent internal tides that no longer remain phase-locked with tidal forcing or have amplitude variations.

[0003] Existing internal tide prediction technologies not only have limited prediction capabilities for incoherent internal tides in internal tides, but also can only be trained based on single mooring data, resulting in poor model generalization ability, inability to handle complex ocean conditions, low prediction result accuracy, and thus being not conducive to practical applications. Summary of the Invention

[0004] The following is an overview of the subject matter described in detail in this article. This overview is not intended to limit the scope of protection of the claims.

[0005] The main purpose of the embodiments of the present disclosure is to propose a method, system, device and storage medium for predicting ocean internal tide current velocity, which can reduce prediction errors and improve prediction accuracy and efficiency.

[0006] The first aspect of the embodiments of this application provides a method for predicting ocean internal tide current velocity for a central controller, and the method includes:

[0007] Obtain historical internal tide current velocity data of the target sea area;

[0008] Input the historical internal tide current velocity data into a preset internal tide current velocity prediction model to obtain the internal tide current velocity prediction result of the target sea area output by the internal tide current velocity prediction model;

[0009] Among them, the process of the internal tide current velocity prediction result of the target sea area output by the internal tide current velocity prediction model includes:

[0010] Decompose the historical internal tide current velocity data into trend component data and seasonal component data;

[0011] Input the trend component data and the seasonal component data into an encoding layer respectively to obtain a first encoding feature of the trend component data and a second encoding feature of the seasonal component data output by the encoding layer;

[0012] Input the first coding feature and the second coding feature into a decoding layer respectively to obtain a first decoding feature corresponding to the first coding feature and a second decoding feature corresponding to the second coding feature output by the decoding layer;

[0013] Output an internal tidal current velocity prediction result of the target sea area according to the first decoding feature and the second decoding feature.

[0014] An embodiment of the present application provides a method for predicting internal tidal current velocity. By obtaining historical internal tidal current velocity data of a target sea area; inputting the historical internal tidal current velocity data into a preset internal tidal current velocity prediction model to obtain an internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model, it can solve the problem of large prediction error caused by the traditional internal tide prediction method based on harmonic analysis being unable to predict the incoherent internal tide part. At the same time, it can also use an internal tidal current velocity prediction model trained with fewer training resources to achieve a high-precision internal tide prediction result for the full depth of different sea area conditions.

[0015] In some embodiments of the present application, the training process of the internal tidal current velocity prediction model includes:

[0016] Select the CORAv2.0 reanalysis dataset;

[0017] Extract historical zonal velocity data and / or historical meridional velocity data from the CORAv2.0 reanalysis dataset;

[0018] Extract historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data from the historical zonal velocity data and / or historical meridional velocity data;

[0019] Determine training data according to the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data;

[0020] Train the internal tidal current velocity prediction model according to the training data.

[0021] In some embodiments of the present application, before determining the training data according to the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data, it further includes:

[0022] Perform format processing on the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data to obtain the diurnal internal tidal current velocity of the historical diurnal internal tidal current velocity data and / or the semi-diurnal internal tidal current velocity of the historical semi-diurnal internal tidal current velocity data. The format processing includes at least one of data standardization and linear interpolation;

[0023] The determination of the training data according to the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data includes:

[0024] Use the intra - tidal current speed within the historical full - day tidal current speed data and / or the intra - tidal current speed within the historical half - day tidal current speed data as the training data.

[0025] In some embodiments of the present application, the decomposition of the historical intra - tidal current speed data into trend component data and seasonal component data includes:

[0026] Pad the first data of the historical intra - tidal current speed data repeatedly in the time dimension to obtain the first intra - tidal current speed data;

[0027] Use a one - dimensional average pooling layer to extract the trend component data from the first intra - tidal current speed data;

[0028] Calculate the difference between the first intra - tidal current speed data and the trend component data to obtain the seasonal component data.

[0029] In some embodiments of the present application, the training of the intra - tidal current speed prediction model according to the training data further includes:

[0030] Decompose the training data into trend component training data and seasonal component training data;

[0031] Input the trend component training data and the seasonal component training data into the encoding layer respectively to obtain the third encoding feature of the trend component training data and the fourth encoding feature of the seasonal component data output by the encoding layer;

[0032] Input the third encoding feature and the fourth encoding feature into the decoding layer respectively to obtain the third decoding feature corresponding to the third encoding feature and the fourth decoding feature corresponding to the fourth encoding feature output by the decoding layer;

[0033] Output the intra - tidal current speed prediction result of the CORAv2.0 re - analysis dataset according to the third decoding feature and the fourth decoding feature;

[0034] Train the intra - tidal current speed prediction model according to the intra - tidal current speed prediction result of the CORAv2.0 re - analysis dataset until the trained intra - tidal current speed prediction model is obtained.

[0035] In some embodiments of the present application, the structure of the encoding layer includes long - short - term memory units, and the structure of the decoding layer includes long - short - term memory units with an attention mechanism introduced;

[0036] After outputting the intra - tidal current speed prediction result of the CORAv2.0 re - analysis dataset according to the third decoding feature and the fourth decoding feature, it further includes:

[0037] Inverse standardize the internal tidal current velocity prediction results of the CORA v2.0 reanalysis dataset to obtain the internal tidal predicted velocity of the CORA v2.0 reanalysis dataset;

[0038] The formula for the inverse standardization includes:

[0039] X = z·σ + μ;

[0040] Where X is the internal tidal predicted velocity, z is the internal tidal current velocity prediction result corresponding to the CORA v2.0 reanalysis dataset, σ is the standard deviation of the CORA v2.0 reanalysis dataset, and μ is the mean of the CORA v2.0 reanalysis dataset.

[0041] In some embodiments of the present application, the loss function of the internal tidal current velocity prediction model is:

[0042]

[0043] Where MSE is the mean square error, y i is the true internal tidal velocity of the CORA v2.0 reanalysis dataset, is the predicted internal tidal velocity of the CORA v2.0 reanalysis dataset, and N is the number of data in the CORA v2.0 reanalysis dataset.

[0044] To achieve the above object, a second aspect of the embodiments of the present invention provides an internal tidal current velocity prediction system, the system includes:

[0045] An acquisition module for acquiring historical internal tidal current velocity data of a target sea area;

[0046] A prediction module for inputting the historical internal tidal current velocity data into a preset internal tidal current velocity prediction model to obtain the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model;

[0047] Wherein, the process of the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model includes:

[0048] Decompose the historical internal tidal current velocity data into trend component data and seasonal component data;

[0049] Input the trend component data and the seasonal component data into an encoding layer respectively to obtain a first encoding feature of the trend component data and a second encoding feature of the seasonal component data output by the encoding layer;

[0050] Input the first encoding feature and the second encoding feature into a decoding layer respectively to obtain a first decoding feature corresponding to the first encoding feature and a second decoding feature corresponding to the second encoding feature output by the decoding layer;

[0051] Output the predicted result of the internal tidal current velocity in the target sea area according to the first decoding feature and the second decoding feature.

[0052] To achieve the above object, a third aspect of the embodiments of the present invention provides an electronic device, including: at least one control processor and a memory communicatively connected to the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor to enable the at least one control processor to execute the above-mentioned method for predicting the internal tidal current velocity in the ocean.

[0053] To achieve the above object, a fourth aspect of the embodiments of the present invention provides a computer-readable storage medium, which stores computer-executable instructions for causing a computer to execute the above-mentioned method for predicting the internal tidal current velocity in the ocean.

[0054] It can be understood that the beneficial effects of the above-mentioned second aspect to the fourth aspect compared with the related art are the same as those of the above-mentioned first aspect compared with the related art. For the relevant descriptions, reference can be made to the relevant descriptions in the above-mentioned first aspect, and details will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the accompanying drawings, where:

[0056] Figure 1 is a schematic flowchart of an implementation manner of a method for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0057] Figure 2 is a model structure diagram of an implementation manner of a method for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0058] Figure 3 is a schematic diagram of data processing of an implementation manner of a method for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0059] Figure 4 is a schematic diagram of the diurnal tide error of an embodiment of a method for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0060] Figure 5 is a schematic diagram of the semi-diurnal tide error of another embodiment of a method for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0061] Figure 6 is a schematic structural diagram of an implementation manner of a system for predicting the internal tidal current velocity in the ocean provided by an embodiment of the present application;

[0062] Figure 7 It is a schematic diagram of the hardware structure of an embodiment of the electronic device provided by an embodiment of the present application. Specific embodiments

[0063] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and should not be construed as a limitation of the present application.

[0064] In the description of the present application, if the first, second, etc. are described, it is only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features or implicitly indicating the sequence of the indicated technical features.

[0065] In the description of the present application, it should be understood that with respect to the orientation description, such as up, down, etc., the orientation or positional relationship indicated is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present application.

[0066] In the description of the present application, it should be noted that unless otherwise clearly defined, words such as setting, installing, connecting, etc. should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above words in the present application in combination with the specific content of the technical solution.

[0067] First, several nouns involved in the present application are analyzed:

[0068] In the ocean field, internal tides in the ocean refer to internal waves with a tidal period generated by the interaction between astronomical tides and topography. Internal tide waves are not only carriers of matter and energy transport, but also an important part of the ocean scale cascade and one of the energy sources of ocean turbulence. These waves are generated synchronously with gravity currents and may remain phase-locked (referred to as coherent internal tides). During the propagation of internal tides, factors such as ocean stratification, background circulation, and mesoscale eddy turbulence have an impact on them. There are incoherent internal tides that no longer remain phase-locked with tidal forcing or have amplitude changes. Those parts of the internal tides that no longer remain phase-locked with tidal forcing or have amplitude changes are called incoherent internal tides.

[0069] In the ocean field, predicting ocean internal tides has a greater impact on ocean engineering facilities, underwater acoustic communication, and submarine underwater navigation, etc. Many risks can be avoided through accurate prediction. Currently, the basis for predicting internal tides is mostly harmonic analysis. Traditional harmonic analysis methods can effectively predict coherent internal tides, but have limited prediction ability for incoherent internal tides.

[0070] In addition, the existing internal tide prediction technology not only has limited prediction ability for the incoherent internal tide in the internal tide, but also can only be trained based on single mooring data, resulting in poor generalization ability of the model, being unable to cope with complex ocean conditions, low accuracy of prediction results, and thus being not conducive to practical applications.

[0071] Based on this, the embodiments of the present application provide a method, a system, an electronic device and a medium for predicting the internal tide current velocity, aiming to reduce prediction errors and improve prediction accuracy and efficiency.

[0072] The method, the system, the electronic device and the medium for predicting the internal tide current velocity provided by the embodiments of the present application are specifically described through the following embodiments. First, the method for predicting the internal tide current velocity in the embodiments of the present application is described.

[0073] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results of theory, method, technology and application system.

[0074] The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, robotics technology, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0075] The method for predicting the internal tide current velocity provided by the embodiments of the present application relates to the technical field. The method for predicting the internal tide current velocity provided by the embodiments of the present application can be applied to a terminal, or can be applied to a server side, or can also be software running on a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or can be configured as a server cluster or a distributed system composed of multiple physical servers, or can also be configured as a cloud server providing 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 software can be an application for implementing the method of deducting trading resources, etc., but is not limited to the above forms.

[0076] This application can be used in numerous general-purpose or special-purpose 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 electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. This application 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. This application can also be practiced in a distributed computing environment where 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.

[0077] It should be noted that in each specific embodiment of this application, when it comes to performing relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first. Moreover, the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiments of this application need to obtain the user's sensitive personal information, the user's separate permission or separate consent will be obtained through methods such as pop-up windows or redirecting to a confirmation page. After clearly obtaining the user's separate permission or separate consent, the necessary user-related data for the normal operation of the embodiments of this application will be obtained.

[0078] For this reason, referring to Figure 1 , an embodiment of this application provides a method for predicting the internal tidal current velocity in the ocean. This method is applied to a central controller. The controller can be a server, an electronic device, or a mobile terminal, etc., and no specific limitation is made here. The method includes the following steps S110 to step S120:

[0079] Step S110: Obtain historical internal tidal current velocity data of the target sea area.

[0080] In this step, obtaining the historical internal tidal current velocity data of the target sea area can preferably be achieved by obtaining the tidal reanalysis dataset of the target sea area and preliminarily processing it to obtain the historical internal tidal current velocity data. According to the target sea area, a reanalysis dataset containing tidal signals is obtained. The reanalysis dataset covers a wider time and space range compared to mooring data, and has strong data consistency, thus avoiding the noise errors of different observation payloads. Moreover, the reanalysis data is more conducive to deep learning training than mooring data. Therefore, the full-depth velocity data of the target sea area is extracted according to the scope of the ocean to be predicted, corresponding to the historical velocity data of each point at each depth layer, including the meridional velocity and the zonal velocity. For the historical velocity data, only one item can be selected as the training data during the training process of the internal tidal current velocity prediction model to ensure that the obtained data has high spatio-temporal resolution and good consistency, which helps to improve the prediction accuracy.

[0081] The following explains the specific training steps of the internal tidal current velocity prediction model:

[0082] Select the CORAv2.0 reanalysis dataset;

[0083] Extract the historical zonal velocity data and / or historical meridional velocity data from the CORAv2.0 reanalysis dataset;

[0084] Extract the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data from the historical zonal velocity data and / or historical meridional velocity data;

[0085] Determine the training data according to the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data;

[0086] Train the internal tidal current velocity prediction model according to the training data.

[0087] In this step, according to the scope of the target sea area for training, select and download the CORAv2.0 reanalysis dataset of the target sea area. Specifically, the data can be obtained by accessing the official website or database of CORAv2.0 data. By downloading the historical zonal velocity (v-component of current) and meridional velocity (u-component of current) data, information containing multiple time steps and depth levels is obtained in the NetCDF (network Common Data Form) network general data format.

[0088] Further, perform data preprocessing on the acquired data. Preferably, data cleaning can be carried out, specifically including: removing outliers or missing values to ensure data quality, or using interpolation methods to fill in missing values and other data processing methods. Then, extract the full-day tidal current velocity data and half-day tidal current velocity data from the processed data. Among them, the full-day tidal current velocity refers to the tide with a period of 24 hours, and the half-day tidal current velocity refers to the tide with a period of 12 hours.

[0089] In this embodiment, the data sea area range intercepted is from 105°E to 120°E, 5°N to 25°N, and the time span is the reanalysis dataset from January 1, 2015 to December 15, 2019. According to the predicted area, obtain the CORAv2.0 reanalysis dataset from the National Marine Information Center. The horizontal resolution of the dataset obtained here is 1 / 12°, with 50 vertical layers, the time span is from 1989 to 2020, and the time resolution is 3h. The CORAv2.0 data has velocity data at 14480 sampling moments for each point and each layer depth, including zonal velocity and meridional velocity. Select the zonal velocity from January 1, 2015 to June 1, 2019 as the training set and validation set, corresponding to 12896 sampling moments, and the data from June 1, 2019 to December 15, 2019 can be used as the test set.

[0090] In some implementation manners of this embodiment, extract different internal tide signals and preprocess them as model training data. First, calculate the baroclinic velocity according to the acquired reanalysis dataset, perform full-depth integration on the velocity field of the current target sea area and then average to obtain the barotropic velocity, subtract the barotropic flow field from the real flow field to obtain the baroclinic flow field, and then use the following method to extract the internal tide information from the baroclinic flow field.

[0091] The specific formula for calculating the baroclinic velocity is as follows:

[0092]

[0093] Among them, H is the water depth, z is the depth, t is the time, dz represents the interval depth of each layer. is the barotropic velocity, u(z,t) is the historical velocity, and u′(z,t) is the baroclinic velocity.

[0094] In this embodiment, the filtering frequency range for extracting the full-day internal tide is [0.85, 1.06]ω K1 (referring to the tidal frequency of the K1 partial tide); the filtering frequency range for extracting the half-day internal tide is [0.92, 1.1]ω M2(Referring to the tidal frequency of M2 tide), since the diurnal tide and semi-diurnal tide contain internal tide signals of various frequencies, a fourth-order Butterworth filter is used to select the filtering range to perform band-pass filtering on the obtained baroclinic flow field to obtain diurnal internal tide data and semi-diurnal internal tide data, which can not only retain the flow velocity signals in the corresponding frequency band to the greatest extent, but also avoid the interference brought by other adjacent frequency signals.

[0095] Furthermore, the obtained diurnal tide data and semi-diurnal tide data are respectively standardized, the data sets at different points are merged, and the structures of the input and output data are adjusted according to application needs.

[0096] In an implementation manner of this embodiment, the first 96 inputs and the last 24 outputs are used to slice the diurnal tide data set and the semi-diurnal tide data set respectively and demarcate 85% to construct a training data set, and 15% to construct a validation set. The specific standardization formula is as follows:

[0097]

[0098] Among them, x is the data to be standardized, z is the standardized data, σ is the standard deviation of x, and μ is the mean of x.

[0099] At the same time, from the extracted diurnal internal tide flow velocity data and semi-diurnal internal tide flow velocity data, select appropriate time periods and depth levels as training data, and perform standardization processing on the training data to obtain a model data set including a training set, a validation set and a test set to evaluate the performance of the model. Moreover, by performing standardization processing on the training data, the training efficiency and accuracy of the model are improved.

[0100] Furthermore, a Dseq2seq model is constructed. The design of the Dseq2seq model is based on the Dlinear model. The trend term and seasonal term decomposition module of the Dlinear model shows that the Dlinear model has extremely few parameters but has better effects than general complex models based on Transformer. And the internal tide, as a fluctuation, is also a physical quantity that is the superposition of a long-term trend and short-term fluctuations. And the deep learning model can fit the coherent internal tide with a fixed frequency while mining the internal relationship of the time series to further fit and predict the incoherent internal tide part.

[0101] Furthermore, train the pre-constructed Dseq2seq model with the training set, monitor the change of the loss function, adjust the hyperparameters to optimize the model performance, then evaluate the generalization ability of the model with the validation set to prevent overfitting, and finally use the test set to further verify the prediction performance of the model, so as to analyze the prediction results of the model and evaluate its performance under different conditions, in order to apply the model to actual scenarios, such as marine environment monitoring, maritime navigation safety, etc. In addition, the Dseq2seq model can be replaced with different machine learning or deep learning models, such as LSTM (Long Short-Term Memory).

[0102] Step S120: Input the historical internal tidal current velocity data into a preset internal tidal current velocity prediction model to obtain the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model.

[0103] In this step, input the historical internal tidal current velocity data of the target sea area into the trained internal tidal current velocity prediction model, so as to obtain the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model.

[0104] The following explains the specific steps of the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model:

[0105] Decompose the historical internal tidal current velocity data into trend component data and seasonal component data;

[0106] Input the trend component data and the seasonal component data into the encoding layer respectively to obtain the first encoding feature of the trend component data and the second encoding feature of the seasonal component data output by the encoding layer;

[0107] Input the first encoding feature and the second encoding feature into the decoding layer respectively to obtain the first decoding feature corresponding to the first encoding feature and the second decoding feature corresponding to the second encoding feature output by the decoding layer.

[0108] In this step, the encoding layer structure includes long short-term memory units, and the decoding layer structure includes long short-term memory units introducing an attention mechanism. By decomposing the historical internal tidal current velocity data into trend components and seasonal components, the time series problem can be significantly simplified, the model fitting difficulty can be reduced, and the prediction accuracy can be improved. Then, input the trend component data and the seasonal component data into the encoding layer respectively to obtain the first encoding feature of the trend component data and the second encoding feature of the seasonal component data output by the encoding layer, and then input the first encoding feature and the second encoding feature into the decoding layer respectively to obtain the first decoding feature corresponding to the first encoding feature and the second decoding feature corresponding to the second encoding feature output by the decoding layer, so as to not only improve the accuracy of the model, but also better understand the internal structure of the data.

[0109] In some implementation manners of this embodiment, such as Figure 2As shown, by decomposing the trend term and the seasonal term, the complex time series problem is simplified. Decomposing it into the trend term and the seasonal term can reduce the fitting difficulty of the deep learning model for the model, reduce the interference during the learning process of the model, and improve the accuracy.

[0110] Among them, when initializing and constructing the sequence decomposition module, a kernel size (kernel_size = 25) and a stride (stride = 1) need to be specified to decompose the historical tidal current velocity data into a trend component and a seasonal component.

[0111] Furthermore, two decoding and encoding layers with the same structure are constructed. The inputs are the trend term and the seasonal term decomposed by the sequence decomposition module respectively. The structure of the decoding and encoding layer is as follows:

[0112] Encoding layer: LSTM cell;

[0113] Decoding layer: Attention mechanism + LSTM cell;

[0114] Among them, the LSTM cell has an input gate, a forget gate, and an output gate to control the flow of information, so as to achieve an effective time memory function and prevent the problem of gradient disappearance; the input gate is similar to the attention of the brain and determines which parts of the newly input data should be added to the current state; the forget gate is similar to a sieve and determines which information in the current state should be retained and which information should be forgotten; the output gate determines which information in the current state should be passed to the state of the next moment and the output of the network.

[0115] The following are the specific steps of the LSTM cell:

[0116] Input gate: i t = σ(W i · [h t-1 , X[t]] + b i );

[0117] Forget gate: f t = σ(W f · [h t-1 , X[t]] + b f );

[0118] Output gate: o t = σ(W o · [h t-1 , X[t]] + b o );

[0119]

[0120] h t = o t ⊙ tanh(C t );

[0121] Among them, X is the input sequence, f t , i t , o t and C t are the forget gate, input gate, cell state, and output gate of time t respectively. W i , b i represent the input gate weight matrix and bias respectively. W f , b f represent the forget gate weight matrix and bias respectively. W o , b o represent the output gate weight matrix and bias respectively. W c , b c represent the cell state weight matrix and bias respectively. C t represents the cell state, represents the candidate cell state, h t and o t 's result is used for subsequent decoding. h t-1 is the hidden state at the previous moment t - 1, x t is the input at time t, σ is the activation function (Sigmoid activation function), h t is the hidden vector state at time t.

[0122] Furthermore, the encoding process is:

[0123] (o t , h t ) = LSTM(X[t], (h t-1 , C t-1 ));

[0124] Introducing the attention mechanism into the decoding process is:

[0125] α t = softmax(W a [h t ; y t-1 );

[0126]

[0127] Among them, the subscript t represents the output time step, i represents the element position of the input sequence, α t represents the attention weight, W a represents the weight matrix, y t-1 represents the output of the previous time step, represents the context vector, h i represents the hidden layer state.

[0128] The LSTM unit in the decoding layer:

[0129]

[0130] Among them, represents the hidden state of the decoding layer.

[0131] Adjust the output through a linear layer:

[0132]

[0133] Among them, Y is the output result, represents the weight matrix and bias vector of the linear layer.

[0134] Furthermore, input the seasonal term output by the sequence decomposition module into the decoding and encoding layer to obtain the output result corresponding to the seasonal term, and input the trend term output by the sequence decomposition module into the decoding and encoding layer to obtain the output result corresponding to the trend term, and directly add the two output results to obtain the final output result.

[0135] In this embodiment, set the model parameters according to the final output result, with the learning rate being 0.001, the random dropout probability being 0.05, and the number of LSTM hidden layer units in the decoding layer and the encoding layer being 128.

[0136] In some embodiments, in the step of decomposing the historical internal tidal current velocity data into trend component data and seasonal component data, the following steps are included:

[0137] Pad the first and last data of the historical internal tidal current velocity data repeatedly in the time dimension to obtain the first internal tidal current velocity data;

[0138] Use a one-dimensional average pooling layer to extract the trend component data from the first internal tidal current velocity data;

[0139] Calculate the difference between the first internal tidal current velocity data and the trend component data to obtain the seasonal component data.

[0140] In this embodiment, first pad the input data repeatedly with the first and last data in the time dimension to ensure that the data at the boundary is not lost during the calculation of the moving average, ensuring the continuity and integrity of the data, and then use a one-dimensional average pooling layer to extract the trend component data from the padded data, and further obtain the seasonal component data by calculating the difference between the padded data and the trend component data.

[0141] In this step, the loss function of the internal tidal current velocity prediction model is:

[0142]

[0143] Among them, MSE is the mean square error, y i is the true internal tidal current velocity of the CORAv2.0 reanalysis dataset, For the predicted velocity of internal tides in the CORA v2.0 reanalysis dataset, N is the number of data in the CORA v2.0 reanalysis dataset.

[0144] In some embodiments, the internal tide velocity prediction model is trained using the mean squared error function to measure the difference between the predicted value and the true value of the model, prevent overfitting, and the error of the model can be quantified by calculating the average of the squares of the differences between the predicted value and the true value, thereby optimizing the model and reducing the error.

[0145] In some embodiments, before the step of determining the training data according to the historical diurnal internal tide velocity data and / or historical semi-diurnal internal tide velocity data, the following steps are included:

[0146] Format the historical diurnal internal tide velocity data and / or historical semi-diurnal internal tide velocity data to obtain the diurnal internal tide velocity of the historical diurnal internal tide velocity data and / or the semi-diurnal internal tide velocity of the historical semi-diurnal internal tide velocity data. The format processing includes at least one of data standardization and linear interpolation.

[0147] In this embodiment, formatting the historical diurnal internal tide velocity data and historical semi-diurnal internal tide velocity data, including data standardization and linear interpolation, can improve the quality of the data and the training effect of the model.

[0148] Among them, data standardization is to scale the data to a specific range (usually 0 to 1 or -1 to 1) to eliminate the influence of dimension. Linear interpolation is a simple method for filling missing values, estimating the missing values through the linear relationship between known data points. Data standardization and linear interpolation can also be combined to further improve the quality of the data, make the model easier to converge, and improve the accuracy of prediction.

[0149] In some embodiments, in the step of determining the training data according to the historical diurnal internal tide velocity data and / or historical semi-diurnal internal tide velocity data, the following steps are included:

[0150] Use the diurnal internal tide velocity of the historical diurnal internal tide velocity data and / or the semi-diurnal internal tide velocity of the historical semi-diurnal internal tide velocity data as the training data.

[0151] In this embodiment, the diurnal tide dataset and the semi-diurnal tide dataset are respectively input into the internal tide velocity prediction model for training. The loss function is set to use MSE to prevent overfitting, and the training is stopped when the validation set does not show a significant decrease for three consecutive epochs to obtain the final diurnal tide internal tide velocity prediction model and semi-diurnal tide internal tide velocity prediction model respectively.

[0152] In some embodiments, in the step of training the internal tide velocity prediction model according to the training data, the following steps are further included:

[0153] Decompose the training data into trend component training data and seasonal component training data;

[0154] Input the trend component training data and the seasonal component training data into the encoding layer respectively, and obtain the third encoding feature of the trend component training data and the fourth encoding feature of the seasonal component data output by the encoding layer;

[0155] Input the third encoding feature and the fourth encoding feature into the decoding layer respectively, and obtain the third decoding feature corresponding to the third encoding feature and the fourth decoding feature corresponding to the fourth encoding feature output by the decoding layer;

[0156] Output the prediction result of the internal tidal current velocity of the CORA v2.0 reanalysis dataset according to the third decoding feature and the fourth decoding feature;

[0157] Train the internal tidal current velocity prediction model according to the prediction result of the internal tidal current velocity of the CORA v2.0 reanalysis dataset until the trained internal tidal current velocity prediction model is obtained.

[0158] In this embodiment, the training data is decomposed into trend component training data and seasonal component training data through the sequence decomposition module, and then the trend component training data is input into the encoding layer to obtain the third encoding feature of the trend component training data, and the seasonal component training data is input into the encoding layer to obtain the fourth encoding feature of the seasonal component data, so as to capture the long-term dependencies in the input sequence through the encoding layer.

[0159] Furthermore, input the third encoding feature into the decoding layer to obtain the third decoding feature corresponding to the third encoding feature, and input the fourth encoding feature into the decoding layer to obtain the fourth decoding feature corresponding to the fourth encoding feature, so as to gradually decode the fixed-length vector generated by the encoding layer into an output sequence, ensuring that each element in the output sequence is associated with the relevant information in the input sequence. Moreover, introducing the attention mechanism in the encoding and decoding layer can dynamically focus on different parts generated by the encoder when generating each output, so as to better capture the local dependencies between the input and the output.

[0160] Finally, output the prediction result of the internal tidal current velocity of the CORA v2.0 reanalysis dataset according to the third decoding feature and the fourth decoding feature. In addition, train the internal tidal current velocity prediction model according to the prediction result of the internal tidal current velocity of the CORA v2.0 reanalysis dataset until the trained internal tidal current velocity prediction model is obtained.

[0161] In some embodiments, after the step of outputting the prediction result of the internal tidal current velocity of the CORA v2.0 reanalysis dataset according to the third decoding feature and the fourth decoding feature, the following steps are further included:

[0162] Inverse standardize the prediction results of the internal tidal current velocity in the CORA v2.0 reanalysis dataset to obtain the predicted internal tidal current velocity in the CORA v2.0 reanalysis dataset;

[0163] The formula for inverse standardization includes:

[0164] X = z·σ + μ;

[0165] where X is the predicted internal tidal current velocity, z is the prediction result of the internal tidal current velocity corresponding to the CORA v2.0 reanalysis dataset, σ is the standard deviation of the CORA v2.0 reanalysis dataset, and μ is the mean of the CORA v2.0 reanalysis dataset.

[0166] In this embodiment, perform inverse standardization on the result output by the data after standardization processing to obtain the final prediction result.

[0167] In some embodiments, as Figure 3 shown, taking the CORA v2.0 reanalysis data test set as an example, extract zonal or meridional internal tidal current velocity data, extract the internal tidal current velocity data of the first 96 time steps to be predicted, input them into the model, where a part is used as model training data and another part is used as actual test data. Then, use the Butterworth filter to select the filtering range to perform band-pass filtering on the obtained baroclinic flow field, and perform standardization processing to obtain diurnal internal tide data and semi-diurnal internal tide data. Furthermore, based on the diurnal internal tide data and semi-diurnal internal tide data, train the pre-constructed Dseq2seq model to obtain the diurnal internal tidal current velocity prediction model and the semi-diurnal internal tidal current velocity prediction model. Finally, perform inverse standardization on the result output by the prediction model to obtain the final prediction result.

[0168] Taking the mooring data as an example, use the mooring (110.9°E, 17°N) as the test set, extract the zonal or meridional internal tidal current velocity fluctuation data, and use the method of linear interpolation to interpolate the mooring data into a format with a 3-hour interval to be consistent with the reanalysis dataset. Extract the internal tidal current velocity data of the first 96 time steps to be predicted, input them into the model, and the training process is the same as that of the above CORA v2.0 reanalysis data test set, which will not be elaborated here.

[0169] In some embodiments, the diurnal tide error within the entire South China Sea is as Figure 4 shown, and the semi-diurnal tide error within the entire South China Sea is as Figure 5As shown, the gradient color scale represents the MSE. Among them, in the diurnal tide prediction, the average MSE of the South China Sea in harmonic analysis is about 0.5, about 0.015 for the Informer model, about 0.012 for the Dseq2seq model, and about 0.026 for the Dlinear model. In the semidiurnal tide prediction, the average MSE of the South China Sea in harmonic analysis is about 1.27, about 0.067 for the Informer model, about 0.062 for the Dseq2seq model, and about 0.09 for the Dlinear model.

[0170] In some embodiments, it is also applicable to physical quantities other than the flow velocity for internal tides. For example, it is applied to the prediction method of isopycnal surface fluctuation data:

[0171] Taking the mooring data as an example, the mooring (110.9°E, 17°N) is used as the test set, and the isopycnal surface fluctuation is extracted. First, the seawater density is calculated. In the embodiment, it is preferably to use the sw_pden function that takes salinity, temperature, and pressure as input parameters and outputs the corresponding seawater density value to calculate the potential density ρ(z,t) of seawater. Then, a 5-day moving average is performed on it to remove the low-frequency fluctuations to obtain <ρ(z,t)> t .

[0172] Furthermore, using linear interpolation, according to <ρ(z,t)> t and the data of depth z, calculate the depth z corresponding to ρ(z,t) ′ (z,t),

[0173] ζ(z,t) = z - z ′ (z,t);

[0174] where ζ(z,t) is the undulation of the isopycnal surface.

[0175] Then, use a fourth-order Butterworth filter to filter the data of the undulation of the isopycnal surface, and the undulations of the isopycnal surface of the diurnal tide and the semidiurnal tide can be obtained respectively. Furthermore, the obtained data is normalized, including using the method of linear interpolation to interpolate the obtained diurnal tide and semidiurnal tide data into a format with a 3h interval, keeping it consistent with the reanalysis dataset. And extract the internal tide flow velocity data of the first 96 time steps to be predicted and input it into the model.

[0176] Finally, the output result of the model is inverse-normalized according to the mean and variance data of the target points in the target sea area in previous years to obtain the final prediction result.

[0177] As Figure 6 shown, some embodiments of the present application provide an internal ocean tide flow velocity prediction system. The system includes an acquisition module 610 and a prediction module 620. Specifically:

[0178] An acquisition module 610 for acquiring historical internal tidal current velocity data of a target sea area;

[0179] A prediction module 620 for inputting the historical internal tidal current velocity data into a preset internal tidal current velocity prediction model to obtain an internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model;

[0180] Among them, the process of the internal tidal current velocity prediction result of the target sea area output by the internal tidal current velocity prediction model includes:

[0181] Decompose the historical internal tidal current velocity data into trend component data and seasonal component data;

[0182] Input the trend component data and the seasonal component data into the encoding layer respectively to obtain a first encoding feature of the trend component data and a second encoding feature of the seasonal component data output by the encoding layer;

[0183] Input the first encoding feature and the second encoding feature into the decoding layer respectively to obtain a first decoding feature corresponding to the first encoding feature and a second decoding feature corresponding to the second encoding feature output by the decoding layer;

[0184] Output the internal tidal current velocity prediction result of the target sea area according to the first decoding feature and the second decoding feature.

[0185] In some embodiments, the prediction module 620 may include: selecting the CORA v2.0 reanalysis dataset.

[0186] In some embodiments, the prediction module 620 may include: extracting historical zonal velocity and / or historical meridional velocity from the CORA v2.0 reanalysis dataset.

[0187] In some embodiments, the prediction module 620 may include: extracting historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data from the historical zonal velocity data and / or historical meridional velocity data.

[0188] In some embodiments, the prediction module 620 may include: determining training data according to the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data.

[0189] In some embodiments, the prediction module 620 may include: training the internal tidal current velocity prediction model according to the training data.

[0190] In some embodiments, the prediction module 620 may include: performing format processing on the historical diurnal internal tidal current velocity data and / or historical semi-diurnal internal tidal current velocity data to obtain the diurnal internal tidal current velocity of the historical diurnal internal tidal current velocity data and / or the semi-diurnal internal tidal current velocity of the historical semi-diurnal internal tidal current velocity data, and the format processing includes at least one of data standardization and linear interpolation;

[0191] In some embodiments, the prediction module 620 may include using the intra-day tidal current velocity of the historical intra-day tidal current velocity data and / or the semi-day tidal current velocity of the historical semi-day tidal current velocity data as training data.

[0192] In some embodiments, the prediction module 620 may include padding the historical intra-day tidal current velocity data by repeating the first data in the time dimension to obtain first intra-day tidal current velocity data.

[0193] In some embodiments, the prediction module 620 may include using a one-dimensional average pooling layer to extract trend component data from the intra-day tidal current velocity data.

[0194] In some embodiments, the prediction module 620 may include calculating the difference between the first intra-day tidal current velocity data and the trend component data to obtain seasonal component data.

[0195] In some embodiments, the prediction module 620 may include decomposing the training data into trend component training data and seasonal component training data.

[0196] In some embodiments, the prediction module 620 may include inputting the trend component training data and the seasonal component training data into an encoding layer respectively to obtain a third encoding feature of the trend component training data and a fourth encoding feature of the seasonal component data output by the encoding layer.

[0197] In some embodiments, the prediction module 620 may include inputting the third encoding feature and the fourth encoding feature into a decoding layer respectively to obtain a third decoding feature corresponding to the third encoding feature and a fourth decoding feature corresponding to the fourth encoding feature output by the decoding layer.

[0198] In some embodiments, the prediction module 620 may include outputting the intra-day tidal current velocity prediction result of the CORA v2.0 reanalysis dataset according to the third decoding feature and the fourth decoding feature.

[0199] In some embodiments, the prediction module 620 may include training an intra-day tidal current velocity prediction model according to the intra-day tidal current velocity prediction result of the CORA v2.0 reanalysis dataset until the trained intra-day tidal current velocity prediction model is obtained.

[0200] In some embodiments, the prediction module 620 may include inverse normalizing the intra-day tidal current velocity prediction result of the CORA v2.0 reanalysis dataset to obtain the predicted intra-day tidal current velocity of the CORA v2.0 reanalysis dataset;

[0201] In some embodiments, the prediction module 620 may include:

[0202] X = z·σ + μ;

[0203] Wherein, X is the predicted velocity of the internal tide, z is the predicted result of the internal tide velocity corresponding to the CORAv2.0 reanalysis dataset, σ is the standard deviation of the CORAv2.0 reanalysis dataset, and μ is the mean of the CORAv2.0 reanalysis dataset.

[0204] In some embodiments, the prediction module 620 may include:

[0205]

[0206] Wherein, MSE is the mean square error, y i is the true velocity of the internal tide in the CORAv2.0 reanalysis dataset, is the predicted velocity of the internal tide in the CORAv2.0 reanalysis dataset, and N is the number of data in the CORAv2.0 reanalysis dataset.

[0207] It should be noted that the marine internal tide velocity prediction system provided in this embodiment and the above-mentioned marine internal tide velocity prediction method are based on the same inventive concept. Therefore, the relevant content of the above-mentioned marine internal tide velocity prediction method also applies to the content of the marine internal tide velocity prediction system. Therefore, it will not be elaborated here.

[0208] In order to solve the problems that the existing internal tide prediction technology not only has limited prediction ability for the incoherent internal tide in the internal tide, but also can only be trained according to single mooring data, resulting in poor model generalization ability, inability to handle complex marine conditions, low accuracy of prediction results, and thus being not conducive to practical use, this method obtains historical internal tide velocity data of the target sea area; inputs the historical internal tide velocity data into a preset internal tide velocity prediction model to obtain the internal tide velocity prediction result of the target sea area output by the internal tide velocity prediction model, which can solve the problem of large prediction error caused by the inability of the traditional internal tide prediction method based on harmonic analysis to predict the incoherent internal tide part, and at the same time can also use less training resources to train the internal tide velocity prediction model to achieve high-precision internal tide prediction results for different sea area conditions at all depths.

[0209] The embodiment of the present application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the above-mentioned marine internal tide velocity prediction method is implemented.

[0210] As Figure 7 , Figure 7 is the schematic diagram of the hardware structure of the electronic device provided by the embodiment of the present application. The electronic device includes:

[0211] At least one battery;

[0212] At least one memory;

[0213] At least one processor;

[0214] At least one program;

[0215] The program is stored in the memory, and the processor executes at least one program to implement a method for predicting ocean internal current velocity as described above in the present disclosure.

[0216] The electronic device can be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), an in-vehicle computer, etc.

[0217] The electronic device of the embodiments of the present application will be introduced in detail below.

[0218] The processor 1600 can be implemented by using a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present disclosure;

[0219] The memory 1700 can 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), etc. The memory 1700 can store an operating system and other application programs. When implementing the technical solutions provided by the embodiments of the present specification through software or firmware, the relevant program codes are stored in the memory 1700 and are called by the processor 1600 to execute a method for predicting ocean internal current velocity in the embodiments of the present disclosure.

[0220] The input / output interface 1800 is used to implement information input and output;

[0221] The communication interface 1900 is used to implement communication interaction between this device and other devices, and can implement communication through a wired method (such as USB, network cable, etc.) or through a wireless method (such as mobile network, WIFI, Bluetooth, etc.);

[0222] The bus 2000 transmits information between various components of the device (such as the processor 1600, the memory 1700, the input / output interface 1800, and the communication interface 1900);

[0223] Among them, the processor 1600, the memory 1700, the input / output interface 1800, and the communication interface 1900 are communicatively connected to each other inside the device through the bus 2000.

[0224] An embodiment of the present disclosure also provides a storage medium, which is a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the above-described method for predicting ocean internal tide current velocity.

[0225] As a non-transitory computer-readable storage medium, a memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory may include high-speed random access memory and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0226] The embodiments described in the embodiments of the present disclosure are for more clearly illustrating the technical solutions of the embodiments of the present disclosure, and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. Those skilled in the art will know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present disclosure are equally applicable to similar technical problems.

[0227] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present disclosure, and may include more or fewer steps than shown in the figures, or combine certain steps, or different steps.

[0228] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0229] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices can be implemented as software, firmware, hardware, and appropriate combinations thereof.

[0230] In the description of the present application and the above-mentioned drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0231] It should be understood that in the present application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships may exist. For example, "A and / or B" may mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or a similar expression means any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c can mean: 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 plural.

[0232] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical, or other forms.

[0233] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0234] In addition, in each embodiment of the present application, each functional unit can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0235] 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 application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. The aforementioned storage medium includes: various media that can store programs, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0236] The above has specifically described the preferred embodiments of the embodiments of the present application, but the embodiments of the present application are not limited to the above-mentioned implementation manners. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the embodiments of the present application, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the embodiments of the present application.

[0237] The above has described the embodiments of the present application in detail with reference to the accompanying drawings, but the present application is not limited to the above embodiments. Within the scope of knowledge possessed by those of ordinary skill in the art, various changes can also be made without departing from the purpose of the present application.

Claims

1. A method for predicting ocean tidal current speed, characterized in that: The method comprises: Obtain historical tidal velocity data of the target sea area; Inputting the historical inland tidal velocity data into a preset inland tidal velocity prediction model to obtain an inland tidal velocity prediction result of the target sea area output by the inland tidal velocity prediction model; The process of outputting the inland tidal velocity prediction result of the target sea area by the inland tidal velocity prediction model includes: Decomposing the historical inland current velocity data into trend component data and seasonal component data; Inputting the trend component data and the season component data into a coding layer respectively, and obtaining a first coding feature of the trend component data and a second coding feature of the season component data output by the coding layer; Inputting the first coding feature and the second coding feature into a decoding layer respectively, and obtaining a first decoding feature corresponding to the first coding feature and a second decoding feature corresponding to the second coding feature output by the decoding layer; The inland current velocity prediction result of the target sea area is output according to the first decoding feature and the second decoding feature.

2. The method for predicting ocean tidal current speed according to claim 1, characterized in that: The training process of the inland current velocity prediction model includes: The CORAv2.0 reanalysis dataset was selected; extracting historical zonal velocity data and / or historical meridional velocity data from the CORA v2.0 reanalysis dataset; Extracting historical full-day inland tidal velocity data and / or historical half-day inland tidal velocity data from the historical latitudinal velocity data and / or historical meridional velocity data; Determine training data based on the historical full-day inland tidal velocity data and / or the historical half-day inland tidal velocity data; The inland current velocity prediction model is trained according to the training data.

3. The method for predicting ocean tidal current speed according to claim 2, characterized in that: Before determining the training data according to the historical full-day inland tidal velocity data and / or the historical half-day inland tidal velocity data, the method further includes: Performing format processing on the historical full-day tidal current velocity data and / or the historical half-day tidal current velocity data to obtain the full-day tidal current velocity of the historical full-day tidal current velocity data and / or the half-day tidal current velocity of the historical half-day tidal current velocity data, wherein the format processing includes at least one of data standardization and linear interpolation; The determining of training data according to the historical full-day inland tidal velocity data and / or the historical half-day inland tidal velocity data comprises: The full-day inland tidal current velocity of the historical full-day inland tidal current velocity data and / or the half-day inland tidal current velocity of the historical half-day inland tidal current velocity data are used as the training data.

4. The method for predicting ocean tidal current speed according to claim 1, characterized in that: Decomposing the historical inland current velocity data into trend component data and seasonal component data includes: Repeating filling the first data of the historical inland current velocity data in the time dimension to obtain the first inland current velocity data; Using a one-dimensional average pooling layer, extracting the trend component data from the first inland current velocity data; The difference between the first inland current velocity data and the trend component data is calculated to obtain the seasonal component data.

5. The method for predicting ocean tidal current speed according to claim 2, characterized in that: The training of the inland flow velocity prediction model according to the training data further includes: Decomposing the training data into trend component training data and seasonal component training data; Input the trend component training data and the season component training data into the encoding layer respectively, and obtain the third encoding feature of the trend component training data and the fourth encoding feature of the season component data output by the encoding layer; Inputting the third coding feature and the fourth coding feature into a decoding layer respectively, and obtaining a third decoding feature corresponding to the third coding feature and a fourth decoding feature corresponding to the fourth coding feature output by the decoding layer; Outputting the inland current velocity prediction result of the CORAv2.0 reanalysis data set according to the third decoding feature and the fourth decoding feature; The inland tidal velocity prediction model is trained according to the inland tidal velocity prediction results of the CORA v2.0 reanalysis data set until a trained inland tidal velocity prediction model is obtained.

6. The method for predicting ocean tidal current speed according to claim 5, characterized in that: The encoding layer structure includes a long short-term memory unit, and the decoding layer structure includes a long short-term memory unit that introduces an attention mechanism; After outputting the inland current velocity prediction result of the CORAv2.0 reanalysis data set according to the third decoding feature and the fourth decoding feature, the method further includes: Denormalizing the inland tide velocity prediction result of the CORAv2.0 reanalysis dataset to obtain the inland tide prediction velocity of the CORAv2.0 reanalysis dataset; The formula for the inverse normalization includes: X = z·σ+μ; Where X is the predicted velocity of the internal tide, z is the prediction result of the internal tide velocity corresponding to the CORAv2.0 reanalysis dataset, σ is the standard deviation of the CORAv2.0 reanalysis dataset, and μ is the mean of the CORAv2.0 reanalysis dataset.

7. The method for predicting ocean inland current velocity according to claim 6, characterized in that: The loss function of the inland current velocity prediction model is: Among them, MSE is the mean square error, y i is the true velocity of the internal tide in the CORA v2.0 reanalysis dataset, is the predicted flow velocity of the internal tide of the CORA v2.0 reanalysis dataset, and N is the number of data in the CORA v2.0 reanalysis dataset.

8. An ocean tidal current velocity prediction system, characterized in that: The system comprises: An acquisition module is used to obtain historical inland current velocity data of the target sea area; A prediction module, used for inputting the historical inland tidal velocity data into a preset inland tidal velocity prediction model to obtain an inland tidal velocity prediction result of the target sea area output by the inland tidal velocity prediction model; The process of outputting the inland tidal velocity prediction result of the target sea area by the inland tidal velocity prediction model includes: Decomposing the historical inland current velocity data into trend component data and seasonal component data; Inputting the trend component data and the season component data into a coding layer respectively, and obtaining a first coding feature of the trend component data and a second coding feature of the season component data output by the coding layer; Inputting the first coding feature and the second coding feature into a decoding layer respectively, and obtaining a first decoding feature corresponding to the first coding feature and a second decoding feature corresponding to the second coding feature output by the decoding layer; The inland current velocity prediction result of the target sea area is output according to the first decoding feature and the second decoding feature.

9. An electronic device, characterized in that: It includes at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute a method for predicting ocean inland current speed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the method for predicting ocean inland current speed as described in any one of claims 1 to 7.

Images