Multi-source marine environment data correction method

Abstract

The invention discloses a multi-source marine environment data correction method, and the method comprises the steps: carrying out the preprocessing of original marine environment data, obtaining standard marine environment data, obtaining a trained marine environment data prediction model based on the standard marine environment data, and carrying out the correction of the marine environment data. Through the feature extraction capability of the marine environment data prediction model, the time sequence modeling capability of the bidirectional long-short-term memory network and the automatic focusing characteristic of the attention mechanism, the precision and reliability of data processing are improved, to-be-corrected marine environment data are input into the trained marine environment data prediction model, and the correction precision of the marine environment data is improved. According to the method, the to-be-corrected marine environment data is predicted to obtain the predicted marine environment data, the to-be-corrected marine environment data is corrected through the predicted marine environment data to obtain the corrected marine environment data, and the correction precision and reliability of the marine environment data are effectively improved.

Description

Technical Field

[0001] The present application relates to the field of multi-source marine environment data correction, and in particular to a multi-source marine environment data correction method. Background Art

[0003] With the advancement of science and technology, the ways of acquiring ocean data are becoming increasingly diversified, including traditional field observations such as ships and buoys, as well as the application of satellite remote sensing and radar technology. However, all of the above methods have limitations: field observations are limited by geographical scope and time intervals; although satellite remote sensing can provide continuous monitoring over a large area, it is subject to weather conditions; although radar technology can effectively monitor sea surface characteristics, signal attenuation and reflection interference reduce data accuracy. In practical applications, in order to more comprehensively and accurately understand the marine environmental conditions, it is often necessary to use these multi-source data in combination; however, due to various observation errors, instrument deviations, environmental interference, and uncertainties in the data processing process, ocean data generally have quality problems, which directly affect the availability of data and the accuracy of scientific research. In order to ensure the reliability and effectiveness of the data, these multi-source data need to be corrected.

[0004] The existing methods for correcting marine environmental data mainly include system bias correction, numerical model assimilation, statistical regression analysis, etc. These methods have limitations in dealing with real-time, nonlinear relationships and multi-source data fusion, resulting in low accuracy of the corrected marine environmental data.

[0005] Therefore, the existing marine environment data correction technology still has the technical problem of low correction accuracy. Summary of the invention

[0006] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0007] The main purpose of the embodiment of the present disclosure is to propose a multi-source ocean environment data correction method, which can improve the accuracy of the corrected ocean environment data.

[0008] A first aspect of an embodiment of the present application provides a multi-source marine environment data correction method, the method comprising:

[0009] Obtain original marine environment data;

[0010] Preprocessing the original marine environment data to obtain standard marine environment data;

[0011] Inputting the standard marine environment data into the marine environment data prediction model to be trained to obtain a trained marine environment data prediction model;

[0012] Inputting the marine environment data to be corrected into the trained marine environment data prediction model to obtain predicted marine environment data;

[0013] The ocean environment data to be corrected is corrected according to the predicted ocean environment data to obtain the corrected ocean environment data.

[0014] The embodiment of the present application provides a multi-source marine environment data correction method, which obtains standard marine environment data by preprocessing original marine environment data, obtains a trained marine environment data prediction model based on the standard marine environment data, and improves the accuracy and reliability of data processing through the feature extraction capability of the marine environment data prediction model, the time series modeling capability of the bidirectional long short-term memory network, and the automatic focusing characteristics of the attention mechanism. The marine environment data to be corrected is input into the trained marine environment data prediction model to obtain predicted marine environment data, and the marine environment data to be corrected is corrected by the predicted marine environment data to obtain the corrected marine environment data. In this way, the accuracy and reliability of the marine environment data correction are effectively improved through the precise error calculation and adjustment of the model output.

[0015] In some embodiments of the present application, the marine environment data prediction model to be trained includes a feature extraction module, a time series modeling module, and an attention module;

[0016] The step of inputting the standard marine environment data into the marine environment data prediction model to be trained to obtain a trained marine environment data prediction model comprises:

[0017] Inputting the standard ocean environment data into the feature extraction module, so that the feature extraction module performs spatial feature extraction on the standard ocean environment data to obtain the spatial features of the standard ocean environment data;

[0018] Inputting the spatial features into the time series modeling module so that the time series modeling module performs past and future time modeling on the spatial features to obtain a time series representation;

[0019] Inputting the time series representation into an attention module so that the attention module calculates the correlation of the time series to obtain a predicted value of the time series representation;

[0020] Determining the loss function value of the marine environment data prediction model to be trained according to the predicted value represented by the time series and the standard marine environment data;

[0021] The marine environment data prediction model to be trained is iteratively updated according to the loss function value until an iteration termination condition is met, thereby obtaining the trained marine environment data prediction model.

[0022] In some embodiments of the present application, the feature extraction module includes a first convolution layer and a second convolution layer connected in sequence, the first convolution layer includes 64 filters, a convolution kernel of size 1, and a first activation function, and the second convolution layer includes 64 filters, a convolution kernel of size 3, and the first activation function.

[0023] In some embodiments of the present application, the attention module includes a similarity calculation module, an attention score module, and a weighted summation module;

[0024] The step of inputting the time series representation into an attention module so that the attention module calculates the correlation of the time series to obtain a predicted value of the time series representation includes:

[0025] Inputting the time series representation into the similarity calculation module to calculate the similarity of each key in the time series to obtain a similarity result of the time series;

[0026] Inputting the similarity result of the time series into the attention score module to obtain the attention score represented by the time series;

[0027] The attention score is input into the weighted sum module to obtain the predicted value represented by the time series.

[0028] In some embodiments of the present application, before inputting the attention score into the weighted sum module to obtain the predicted value represented by the time series, the method further includes:

[0029] Normalizing the attention score to obtain the processed attention score;

[0030] The processed attention score is converted by the following formula to obtain the attention weight:

[0031]

[0032] Among them, t is the time step t, α t is the attention weight of the current time step t, e t is the attention score after processing at the current time step t, and T is the length of the time series;

[0033] The inputting the attention score into the weighted sum module to obtain the predicted value represented by the time series includes:

[0034] The attention weights are weighted and summed by the following formula to obtain the predicted value represented by the time series:

[0035]

[0036] in, is the predicted value represented by the time series, h t is the hidden state at time step t.

[0037] In some embodiments of the present application, the correcting the to-be-corrected marine environment data according to the predicted marine environment data to obtain the corrected marine environment data includes:

[0038] Calculating the difference between the predicted ocean environment data and the ocean environment data to be corrected;

[0039] The marine environment data to be corrected is corrected according to the difference value by the following formula to obtain the corrected marine environment data:

[0040] y′=y+r

[0041]

[0042] Among them, t' is the corrected marine environment data, t is the marine environment data to be corrected, is the predicted marine environment data, and r is the difference.

[0043] In some embodiments of the present application, the loss function of the marine environment data prediction model is:

[0044]

[0045] Among them, L bce is the mean square error function, y i is the true value, is the predicted value, and n is the number of samples.

[0046] A third aspect of an embodiment of the present invention provides an electronic device, comprising: at least one control processor and a memory for communicating with 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 so that the at least one control processor can execute the above-mentioned multi-source ocean environment data correction method.

[0047] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned multi-source marine environment data correction method.

[0048] It can be understood that the beneficial effects of the second to fourth aspects compared with the related art are the same as the beneficial effects of the first aspect compared with the related art. Please refer to the relevant description in the first aspect, and no further details will be given here. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0050] Figure 1 It is a flowchart of a multi-source marine environment data correction method provided in an embodiment of the present application;

[0051] FIG. 2( a ) is a temperature diagram before correction of a multi-source ocean environment data correction method provided in an embodiment of the present application;

[0052] FIG2( b ) is a corrected temperature diagram of a multi-source ocean environment data correction method provided in an embodiment of the present application;

[0053] Figure 3 It is a structural schematic diagram of a multi-source marine environment data correction device provided in an embodiment of the present application;

[0054] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.

[0056] In the description of this application, if there is a description of first, second, etc., it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0057] In the description of the present application, it should be understood that the descriptions involving orientation, such as the orientation or positional relationship indicated as up, down, etc., are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0058] In the description of this application, it should be noted that, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meaning of the above terms in this application based on the specific content of the technical solution.

[0059] Accurate marine environmental data is crucial in marine scientific research and marine environmental monitoring. It has important applications in marine forecasting, climate research, environmental protection, and marine resource management. With the continuous advancement of science and technology, the means of obtaining marine environmental data are becoming increasingly diverse.

[0060] Conventional marine environmental data collection methods are generally through field observation equipment such as ships and buoys. This method can provide relatively direct and accurate local data, but is limited by the observation range and frequency. The emergence of satellite remote sensing technology has greatly expanded the coverage of marine observations and can achieve large-area, long-term series monitoring; however, satellite data is easily affected by factors such as cloud cover and atmospheric interference during the acquisition process, resulting in uncertainty in data quality. Radar technology, as another important monitoring method, has unique advantages in monitoring ocean surface characteristics and dynamic changes, but it also faces problems such as signal attenuation and reflection interference, which affect the accuracy of data. In practical applications, in order to understand the marine environmental conditions more comprehensively and accurately, it is often necessary to use these multi-source data in combination. However, due to various observation errors, instrument deviations, environmental interference, and uncertainties in the data processing process, marine data generally have quality problems, which directly affect the availability of data and the accuracy of scientific research. In order to ensure the reliability and effectiveness of the data, these multi-source data need to be corrected.

[0061] At present, the correction technologies for marine environmental data mainly include systematic bias correction, numerical model assimilation, statistical regression analysis, etc. Systematic bias correction identifies and eliminates systematic errors by comparing historical data with reference standards; numerical model assimilation uses numerical models to combine observational data with forecast models to optimize forecast results; statistical regression analysis corrects data by establishing statistical relationships between observational data and environmental variables. However, these methods have limitations in dealing with real-time, nonlinear relationships, and multi-source data fusion.

[0062] For this purpose, refer to Figure 1 The embodiment of the present application provides a multi-source marine environment data correction method. The method is applied to a controller. The controller may be a server, an electronic device, or a mobile terminal, etc., which is not specifically limited here. The method includes the following steps S101 to S105:

[0063] Step S101, obtaining original ocean environment data.

[0064] In this step, raw ocean environment data is collected through multi-source ocean environment sensors (such as buoys, satellite remote sensing, ship observations, etc.), where the ocean environment data includes but is not limited to various ocean environment parameters such as seawater temperature, salinity, flow rate, wave height, wind speed, and wind direction.

[0065] Step S102: pre-process the original ocean environment data to obtain standard ocean environment data.

[0066] In this step, the time format, spatial coordinate system and measurement unit of the original marine environmental data are first adjusted for consistency, and then the data is cleaned by outlier detection and elimination, missing value processing and other methods. Among them, outlier detection and elimination mainly use statistical methods (box plots or standard deviations) and physical models to detect and eliminate obviously abnormal data points. Missing value processing mainly uses interpolation (linear interpolation) or filling methods based on historical data to process missing values. Finally, the data is standardized so that the data of different parameters have the same dimension and normalized by the following formula to complete the preprocessing of the original marine environmental data:

[0067]

[0068] Among them, x is the original data, μ is the mean of the data, and σ is the standard deviation of the data.

[0069] Step S103: input the standard ocean environment data into the ocean environment data prediction model to be trained to obtain a trained ocean environment data prediction model.

[0070] In this step, the marine environment data prediction model to be trained is a multi-source marine environment data correction model based on convolutional neural networks, bidirectional long short-term memory networks, and attention mechanisms, which integrates multi-source data from different sensors and platforms, thereby enhancing the effectiveness and robustness of data analysis.

[0071] Step S104: input the ocean environment data to be corrected into the trained ocean environment data prediction model to obtain predicted ocean environment data.

[0072] In this step, after obtaining the marine environmental data to be corrected, first, a convolutional neural network (CNN) is used to extract multi-level and local features in the input data to capture spatial features; then, the time series data is processed through a bidirectional long short-term memory network (BiLSTM) to achieve comprehensive modeling of the dynamic changes of the marine environment; finally, the introduction of the attention mechanism enables the model to automatically identify and focus on key time steps and features, thereby improving the accuracy and interpretability of the prediction.

[0073] Step S105: Correct the ocean environment data to be corrected according to the predicted ocean environment data to obtain corrected ocean environment data.

[0074] In this step, the correction value is calculated using the prediction results of the trained ocean environment data model and the ocean environment data to be corrected. The ocean environment data to be corrected is corrected using the correction value to quantify the difference between the model's prediction and reality, thereby improving the accuracy of the data.

[0075] Different from the prior art, the embodiment of the present application obtains standard marine environment data by preprocessing the original marine environment data, obtains a trained marine environment data prediction model based on the standard marine environment data, and improves the accuracy and reliability of data processing through the feature extraction capability of the marine environment data prediction model, the time series modeling capability of the bidirectional long short-term memory network, and the automatic focusing characteristics of the attention mechanism. The marine environment data to be corrected is input into the trained marine environment data prediction model to obtain predicted marine environment data, and the marine environment data to be corrected is corrected by the predicted marine environment data to obtain the corrected marine environment data. In this way, the accuracy and reliability of the marine environment data correction are effectively improved through the precise error calculation and adjustment of the model output.

[0076] The specific implementation methods of the above steps are introduced below.

[0077] Before inputting the standard ocean environment data into the ocean environment data prediction model to be trained in step S103 to obtain the trained ocean environment data prediction model, the method further includes the following steps S201 to S205.

[0078] Step S201: input the standard ocean environment data into a feature extraction module, so that the feature extraction module extracts spatial features of the standard ocean environment data to obtain spatial features of the standard ocean environment data.

[0079] Step S202: input the spatial features into the time series modeling module, so that the time series modeling module performs past and future time modeling on the spatial features to obtain a time series representation.

[0080] Step S203: input the time series representation into the attention module so that the attention module calculates the correlation of the time series to obtain the predicted value of the time series representation.

[0081] Step S204: determining the loss function value of the marine environment data prediction model to be trained according to the predicted value represented by the time series and the standard marine environment data.

[0082] Step S205, iteratively update the marine environment data prediction model to be trained according to the loss function value until the iteration termination condition is met to obtain a trained marine environment data prediction model.

[0083] In this embodiment, the marine environment data prediction model to be trained includes a feature extraction module, a time series modeling module and an attention module.

[0084] In this embodiment, a convolutional neural network is used as a feature extraction module. The local features of the input data are extracted through the convolution layer, and then the spatial dimension of the features is reduced through the pooling layer. The convolutional neural network in this embodiment is composed of two one-dimensional convolutional layers (i.e., the first convolutional layer and the second convolutional layer), and each convolutional layer is followed by a one-dimensional maximum pooling layer and a Dropout layer (discarding layer).

[0085] Specifically, the first convolution layer includes 64 filters, a convolution kernel of size 1, and a first activation function, the second convolution layer includes 64 filters, a convolution kernel of size 3, and the first activation function, and a one-dimensional pooling layer is connected after the first convolution layer, and the one-dimensional pooling layer performs pooling with a window size of 2, wherein the first activation function is a ReLU activation function.

[0086] In this embodiment, different convolution kernel sizes in the convolution layer are used to capture temporal information of different scales in the marine environment data. The first convolution layer has a smaller convolution kernel, which can effectively capture finer features, while the second convolution layer has a larger convolution kernel, which can capture a wider range of temporal patterns. The one-dimensional maximum pooling layer is used to reduce the number of parameters and reduce the risk of overfitting while maintaining the representativeness of the captured features. The Dropout layer is used to randomly discard the output of neurons to enhance the generalization of the model.

[0087] In this embodiment, the above-mentioned convolutional neural network is used as a feature extraction module to extract spatial features of the standard marine environment data to obtain the spatial features of the standard marine environment data.

[0088] In this embodiment, a bidirectional long short-term memory network (Bi-LSTM) is used as a time series modeling module. The traditional recurrent neural network (RNN) is designed to process sequential data and capture time dependencies by passing hidden states across time steps. The long short-term memory network (LSTM) is an enhanced RNN architecture that alleviates the gradient problem in sequence processing by using forget gates, input and output gates. The forget gate discards unnecessary old information, the input gate records relevant new information, and the output gate determines how the current memory state is reflected in the output. These gates enable LSTM to effectively retain long-term information and can effectively solve the problem of predicting sequences.

[0089] Specifically, the LSTM network structure consists of the following four steps: updating the forget gate, input gate, cell state, and output gate. The following are the specific steps of the LSTM network structure:

[0090] Update the forget gate:

[0091] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0092] Update the input gate:

[0093] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0094] Update unit status:

[0095] o t =σ(W o ·[h t-1 ,h t ]+b o )

[0096] Update the output gate:

[0097] C t =f t ·C t-1 +i t tanh(W c ·[h t-1 ,x t ]+b c )

[0098] h t =o t tanh(C t )

[0099] Among them, f t ,i t , o t and C t are the forget gate, input gate, unit state and output gate at time t, W and b are the corresponding weight coefficients and biases, h t-1 is the hidden state at the previous moment t-1, x t is the input at time t, σ is the second activation function (Sigmoid activation function), h t The hidden state at time t.

[0100] In this embodiment, the limitation of the LSTM unit is that it transmits information only in a single direction, while the Bi-LSTM contains two independent LSTM layers, one processing the forward sequence (from the beginning to the end of the sequence) and the other processing the reverse sequence (from the end to the beginning of the sequence), while considering both forward and backward information, enabling it to capture richer contextual information. In this embodiment, the Bi-LSTM is used as a time series modeling module to perform past and future time modeling on the spatial features extracted by the convolutional neural network to obtain the corresponding time series representation.

[0101] In this embodiment, a Bi-LSTM network is used to make full use of past and future information and capture the dynamic changes of time series to improve the modeling ability of time dependencies.

[0102] In this embodiment, a self-attention mechanism is introduced as an attention module to enhance the generalization ability of the marine environment data prediction model. The attention module enables the model to dynamically focus on important parts while processing the sequence. Specifically, the correlation of the time series is calculated through the attention module to obtain the prediction value represented by the time series; and the loss function value of the marine environment data prediction model to be trained is determined according to the prediction value represented by the time series and the standard marine environment data, and the marine environment data prediction model to be trained is iteratively updated according to the loss function value until the iteration termination condition is met to obtain the trained marine environment data prediction model.

[0103] In this embodiment, firstly, the spatial pattern and dynamic changes of marine environmental data are effectively captured through the multi-level features extracted by CNN and the time series modeling capability of BiLSTM, which improves the processing accuracy. Secondly, the introduced attention mechanism enables the model to automatically identify key features and time steps, enhances the reliability and interpretability of the results, and helps researchers understand the main factors affecting the data. In addition, the model can fuse data from different sources, improves the adaptability to heterogeneous data, and makes the results more comprehensive and accurate. Finally, the self-learning and online adjustment capabilities of the model enable it to optimize over time and adapt to new data patterns, thereby showing stronger stability and effectiveness in long-term applications, meeting the actual needs of marine environmental monitoring and research.

[0104] In some embodiments, the loss function of the marine environment data prediction model is a mean square error function (MSE), which can be expressed by the following formula:

[0105]

[0106] Among them, L bce is the mean square error function, y i is the true value, is the predicted value, and n is the number of samples.

[0107] Inputting the time series representation into the attention module in step S203 so that the attention module calculates the correlation of the time series to obtain the predicted value of the time series representation includes the following steps S301 to S303.

[0108] Step S301: input the time series representation into a similarity calculation module to calculate the similarity of each key in the time series to obtain a similarity result of the time series.

[0109] Step S302: input the similarity result of the time series into the attention score module to obtain the attention score represented by the time series.

[0110] Step S303: Input the attention score into a weighted sum module to obtain a predicted value represented by a time series.

[0111] In this embodiment, the attention module includes a similarity calculation module, an attention score module and a weighted summation module.

[0112] In this embodiment, in the similarity calculation module, the output of the Bi-LSTM network is connected through a fully connected layer, that is, the time series representation of the output of the Bi-LSTM network is input into the fully connected layer to obtain the similarity result of the time series. And the attention score is calculated in the attention score module by the following formula:

[0113] e t =tanh(W h h t +b h )

[0114] Among them, e t represents the attention score, W h , b t represents the weight and bias of the attention mechanism, h t Represents the input of the attention mechanism.

[0115] In this embodiment, the calculated attention scores are input into a weighted summation module, and the attention scores are weighted summed to obtain a prediction value represented by a time series, that is, a prediction value output by the marine environment data prediction model.

[0116] Before inputting the attention score into the weighted sum module in step S303 to obtain the predicted value represented by the time series, the method further includes the following steps S401 to S402.

[0117] Step S401: normalize the attention score to obtain the processed attention score.

[0118] Step S402: Convert the processed attention score to obtain the attention weight using the following formula:

[0119]

[0120] Among them, t is the time step t, α t is the attention weight of the current time step t, e t is the attention score after processing at the current time step t, and T is the length of the time series.

[0121] Inputting the attention score into the weighted sum module in step S303 to obtain the predicted value represented by the time series also includes the following step S403.

[0122] Step S403: perform weighted summation on the attention weights using the following formula to obtain the predicted value represented by the time series:

[0123]

[0124] in, is the predicted value represented by the time series, h t is the hidden state at time step t.

[0125] In this embodiment, the attention module enables the model to automatically identify and focus on key features and time steps, thereby enhancing the interpretability and accuracy of the prediction results.

[0126] The step S105 of correcting the to-be-corrected ocean environment data according to the predicted ocean environment data to obtain the corrected ocean environment data includes steps S501 to S502.

[0127] Step S501: Calculate the difference between the predicted ocean environment data and the ocean environment data to be corrected.

[0128] Step S502: Correct the ocean environment data to be corrected according to the difference using the following formula to obtain the corrected ocean environment data:

[0129] y′=y+r

[0130]

[0131] Among them, y′ is the corrected marine environmental data, y is the marine environmental data to be corrected, is the predicted marine environment data, and r is the difference.

[0132] In this embodiment, the main purpose of data correction is to adjust the original measurement data using the results of model prediction to improve the accuracy of the data. Before data correction, the predicted values ​​and actual measurement values ​​generated by the model are collected; the actual measurement values ​​refer to the original marine environment data (such as temperature, salinity, etc.) obtained by sensors or other measuring equipment, that is, the marine environment data to be corrected; the predicted values ​​are the predicted values ​​of the data at the same time point generated by the marine environment data prediction model; the correction values ​​are calculated based on the predicted values ​​output by the marine environment data prediction model to quantify the difference between the model's prediction and the actual.

[0133] In this embodiment, the error between the model output and the actual data is calculated by the following formula:

[0134]

[0135] The error e represents the difference between the predicted value and the actual measured value. A positive value indicates that the model underestimates the actual value, while a negative value indicates that the model overestimates the actual value. The actual measured value is corrected according to the calculated predicted error e. In this embodiment, the correction value r is equal to the error e.

[0136]

[0137] The calculated correction value is used to correct the original measurement data. The corrected data y′ is obtained by the following formula:

[0138] y′=y+r

[0139] Directly expressed as predicted value:

[0140]

[0141] In this embodiment, by appropriately adjusting the measured values, the corrected ocean environment data can more accurately reflect the state of the ocean environment.

[0142] In some embodiments, after the corrected data is generated, the accuracy of the correction is quantitatively evaluated by the mean square error function (MSE).

[0143]

[0144] Among them, y i is the i-th true value, y′ i is the ith correction value, and n is the total number of samples.

[0145] As shown in Figure 2(a) and Figure 2(b), the temperature data in the marine environment data is corrected using the technical solution provided by this embodiment, and the comparison diagrams before and after the temperature correction are obtained. Figure 2(a) is the temperature data before correction, and Figure 2(b) is the temperature data after correction.

[0146] In this embodiment, the original marine environment data is preprocessed to obtain standard marine environment data, and a trained marine environment data prediction model is obtained based on the standard marine environment data. The feature extraction capability of the marine environment data prediction model, the time series modeling capability of the bidirectional long short-term memory network, and the automatic focusing characteristics of the attention mechanism are used to improve the accuracy and reliability of data processing. The marine environment data to be corrected is input into the trained marine environment data prediction model to obtain predicted marine environment data, and the marine environment data to be corrected is corrected by the predicted marine environment data to obtain the corrected marine environment data. In this way, the accuracy and reliability of the marine environment data correction are effectively improved through the precise error calculation and adjustment of the model output.

[0147] like Figure 3 As shown, some embodiments of the present application provide a multi-source marine environment data correction device, the device includes an acquisition module 1100, a processing module 1200, a training module 1300, a prediction module 1400, and a correction module 1500. Specifically:

[0148] An acquisition module 1100 is used to acquire original ocean environment data;

[0149] The processing module 1200 is used to pre-process the original ocean environment data to obtain standard ocean environment data;

[0150] The training module 1300 is used to input the standard ocean environment data into the ocean environment data prediction model to be trained to obtain a trained ocean environment data prediction model;

[0151] The prediction module 1400 is used to input the ocean environment data to be corrected into the trained ocean environment data prediction model to obtain predicted ocean environment data;

[0152] The correction module 1500 is used to correct the marine environment data to be corrected according to the predicted marine environment data to obtain the corrected marine environment data.

[0153] It should be noted that the multi-source ocean environment data correction device provided in this embodiment and the above-mentioned multi-source ocean environment data correction method are based on the same inventive concept, so the relevant content of the above-mentioned multi-source ocean environment data correction method is also applicable to the content of the multi-source ocean environment data correction device, and therefore, it will not be repeated here.

[0154] like Figure 4 The present application also provides an electronic device, which includes:

[0155] At least one hydrogen fuel cell;

[0156] at least one memory;

[0157] at least one processor;

[0158] at least one program;

[0159] The program is stored in the memory, and the processor executes at least one program to implement the multi-source ocean environment data correction method mentioned above in the present disclosure.

[0160] The electronic device may be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), a vehicle-mounted computer, etc.

[0161] The electronic device according to the embodiment of the present application is described in detail below.

[0162] The processor 1600 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an 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 disclosure;

[0163] 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). The memory 1700 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1700, and the processor 1600 calls and executes a multi-source marine environment data correction method of the embodiment of the present disclosure.

[0164] Input / output interface 1800, used to implement information input and output;

[0165] The communication interface 1900 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.);

[0166] Bus 2000 , which transmits information between various components of the device (e.g., processor 1600 , memory 1700 , input / output interface 1800 , and communication interface 1900 );

[0167] The processor 1600 , the memory 1700 , the input / output interface 1800 , and the communication interface 1900 are connected to each other in communication within the device via the bus 2000 .

[0168] An embodiment of the present disclosure also provides a storage medium, which is a computer-readable storage medium. The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned multi-source marine environment data correction method.

[0169] 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.

[0170] The embodiments described in the embodiments of the present disclosure are intended to more clearly illustrate 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 appreciate 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 also applicable to similar technical problems.

[0171] Those skilled in the art will appreciate that the technical solutions shown in the figures do not limit the embodiments of the present disclosure and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0172] 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.

[0173] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0174] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application 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 interchangeable where appropriate, 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 "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0175] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means 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 mean: 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 objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least 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 multiple.

[0176] In the 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 only schematic. For example, the division of 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 system, 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.

[0177] The units described 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.

[0178] In addition, each functional unit in each embodiment of the present application 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.

[0179] 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 is essentially 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, and the computer software product is stored in a storage medium, including multiple instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store programs.

[0180] The above is a specific description of the preferred implementation of the embodiments of the present application, but the embodiments of the present application are not limited to the above-mentioned implementation methods. Technical personnel familiar with the field can also make various equivalent modifications or substitutions without violating the spirit of the embodiments of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the embodiments of the present application.

[0181] The embodiments of the present application are described in detail above in conjunction with the accompanying drawings, but the present application is not limited to the above embodiments. Various changes can be made within the knowledge scope of ordinary technicians in the relevant technical field without departing from the purpose of the present application.

Claims

1. A multi-source marine environment data correction method, characterized in that: The method comprises: Obtain original marine environment data; Preprocessing the original marine environment data to obtain standard marine environment data; Inputting the standard marine environment data into the marine environment data prediction model to be trained to obtain a trained marine environment data prediction model; Inputting the marine environment data to be corrected into the trained marine environment data prediction model to obtain predicted marine environment data; The ocean environment data to be corrected is corrected according to the predicted ocean environment data to obtain the corrected ocean environment data.

2. The multi-source ocean environment data correction method according to claim 1, characterized in that: The marine environment data prediction model to be trained includes a feature extraction module, a time series modeling module and an attention module; The step of inputting the standard marine environment data into the marine environment data prediction model to be trained to obtain a trained marine environment data prediction model comprises: Inputting the standard ocean environment data into the feature extraction module, so that the feature extraction module performs spatial feature extraction on the standard ocean environment data to obtain the spatial features of the standard ocean environment data; Inputting the spatial features into the time series modeling module so that the time series modeling module performs past and future time modeling on the spatial features to obtain a time series representation; Inputting the time series representation into an attention module so that the attention module calculates the correlation of the time series to obtain a predicted value of the time series representation; Determining the loss function value of the marine environment data prediction model to be trained according to the predicted value represented by the time series and the standard marine environment data; The marine environment data prediction model to be trained is iteratively updated according to the loss function value until an iteration termination condition is met, thereby obtaining the trained marine environment data prediction model.

3. The multi-source ocean environment data correction method according to claim 2, characterized in that: The feature extraction module includes a first convolution layer and a second convolution layer connected in sequence, the first convolution layer includes 64 filters, a convolution kernel of size 1 and a first activation function, and the second convolution layer includes 64 filters, a convolution kernel of size 3 and the first activation function.

4. The multi-source ocean environment data correction method according to claim 2, characterized in that: The attention module includes a similarity calculation module, an attention score module and a weighted summation module; The step of inputting the time series representation into an attention module so that the attention module calculates the correlation of the time series to obtain a predicted value of the time series representation includes: Inputting the time series representation into the similarity calculation module to calculate the similarity of each key in the time series to obtain a similarity result of the time series; Inputting the similarity result of the time series into the attention score module to obtain the attention score represented by the time series; The attention score is input into the weighted sum module to obtain the predicted value represented by the time series.

5. The multi-source ocean environment data correction method according to claim 4, characterized in that: Before inputting the attention score into the weighted sum module to obtain the predicted value represented by the time series, the method further includes: Normalizing the attention score to obtain the processed attention score; The processed attention score is converted by the following formula to obtain the attention weight: Among them, t is the time step t, α t is the attention weight of the current time step t, e t is the attention score after processing at the current time step t, and T is the length of the time series; The inputting the attention score into the weighted sum module to obtain the predicted value represented by the time series includes: The attention weights are weighted and summed by the following formula to obtain the predicted value represented by the time series: in, is the predicted value represented by the time series, h t is the hidden state at time step t.

6. The multi-source ocean environment data correction method according to claim 1, characterized in that: The step of correcting the marine environment data to be corrected according to the predicted marine environment data to obtain the corrected marine environment data comprises: Calculating the difference between the predicted ocean environment data and the ocean environment data to be corrected; The marine environment data to be corrected is corrected according to the difference value by the following formula to obtain the corrected marine environment data: y′=y+r Among them, y′ is the corrected marine environmental data, y is the marine environmental data to be corrected, is the predicted marine environment data, and r is the difference.

7. The multi-source ocean environment data correction method according to claim 2, characterized in that: The loss function of the marine environment data prediction model is: Among them, L bce is the mean square error function, y i is the true value, is the predicted value, and n is the number of samples.

8. A multi-source ocean environment data correction device, characterized in that: The device comprises: An acquisition module is used to obtain raw marine environment data; A processing module, used for preprocessing the raw ocean environment data to obtain standard ocean environment data; A training module, used for inputting the standard marine environment data into the marine environment data prediction model to be trained to obtain a trained marine environment data prediction model; A prediction module, used for inputting the marine environment data to be corrected into the trained marine environment data prediction model to obtain predicted marine environment data; The correction module is used to correct the marine environment data to be corrected according to the predicted marine environment data to obtain the corrected marine environment data.

9. An electronic device, characterized in that: It comprises at least one control processor and a memory for communicating with 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 so that the at least one control processor can execute a multi-source marine environment data correction method 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 a multi-source marine environment data correction method as described in any one of claims 1 to 7.

Images