Method, System and Electronic Device for Processing Maritime Meteorological Forecast Data

Abstract

The present invention provides a method, a system and an electronic device for processing marine weather forecast data, which relates to the technical field of weather forecasting. The method constructs a new data set by using the difference data between the forecast data and the analysis data contained in the limited meteorological historical data, and fuses the data set with the meteorological historical data for use in the generation and update process of the prediction model, realizing the processing of the weather forecast data by making full use of the time characteristics and spatial characteristics of the data under limited data conditions, thereby fully capturing the dynamic changes of meteorological variables and improving the accuracy of marine weather forecasting.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological forecasting, and in particular, to a method, system and electronic device for processing marine meteorological forecasting data. Background Art

[0002] In the process of marine meteorological forecasting, limited by marine environmental factors, it is impossible to set up many fixed observation stations. Therefore, non-fixed observation points such as ships are required to obtain and predict marine meteorological data. Since the amount of data of such ship observation data is small and the real-time performance of ship observation data is poor, it is difficult to obtain marine meteorological data within a relatively wide time period, and there is a problem of poor quality of meteorological data, so that the dynamic changes of meteorological variables cannot be fully captured, resulting in low accuracy of marine meteorological forecasting. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a method, system and electronic device for processing marine meteorological forecasting data. This method constructs a new data set by using the difference data between the forecast data and the analysis data contained in the limited meteorological historical data, and fuses this data set with the meteorological historical data for the generation and update process of the prediction model, realizing the full utilization of the time characteristics and space characteristics of the data to process the meteorological forecasting data under limited data conditions, so as to fully capture the dynamic changes of meteorological variables and improve the accuracy of marine meteorological forecasting.

[0004] In the first aspect, an embodiment of the present invention provides a method for processing marine meteorological forecasting data, the method comprising:

[0005] Obtain the ship observation data corresponding to the observation point based on the position data of the observation point, and use the ship observation data to obtain the meteorological forecast historical data and meteorological analysis historical data corresponding to the observation point;

[0006] Obtain a random point corresponding to the observation point, use the random point to construct an observation area corresponding to the observation point, and use the meteorological forecast historical data and meteorological analysis historical data contained in the observation area to construct a first meteorological data set;

[0007] Determine the difference data between the meteorological forecast historical data and the meteorological analysis historical data in the observation area at the same historical moment, and use the difference data to construct a second meteorological data set corresponding to the observation area;

[0008] Construct a prediction model corresponding to the observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set, and update the prediction model with the difference data in the second meteorological data set;

[0009] Obtain the meteorological forecast data corresponding to the observation point at the current moment, and use the updated prediction model to obtain the meteorological analysis data corresponding to the meteorological forecast data.

[0010] Optionally, obtain the ship observation data corresponding to the observation point based on the position data of the observation point, and use the ship observation data to obtain the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point, including:

[0011] Determine the ship observation data collected by the meteorological ship in the observation point according to the position data determined by the observation point;

[0012] Determine the corresponding ocean area between the meteorological ship and the observation point, and obtain the corresponding meteorological historical data in the ocean area;

[0013] Determine the historical meteorological forecast data and historical meteorological analysis data based on the meteorological historical data.

[0014] Optionally, obtain the random points corresponding to the observation point, use the random points to construct the observation area corresponding to the observation point, and use the historical meteorological forecast data and historical meteorological analysis data included in the observation area to construct the first meteorological data set, including:

[0015] Obtain multiple random points corresponding to the observation point in the ocean area between the meteorological ship and the observation point;

[0016] Use the position data of the multiple random points to determine the boundary of the area corresponding to the observation point, and use the area boundary to construct the observation area corresponding to the observation point;

[0017] Obtain the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point and random points in the observation area;

[0018] Use the historical meteorological forecast data and historical meteorological analysis data to construct the first meteorological data set.

[0019] Optionally, determine the difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, and use the difference data to construct the second meteorological data set corresponding to the observation area, including:

[0020] Obtain the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation area at the same historical moment, and determine the meteorological forecast value and meteorological observation value corresponding to the observation area based on the historical meteorological forecast data and the historical meteorological analysis data;

[0021] Use the meteorological forecast value and the meteorological observation value to determine the difference data, and after transferring the difference data to the grid field corresponding to the observation area by the inverse distance weighted interpolation method, use the grid field to generate the error field corresponding to the observation area;

[0022] After updating the first meteorological dataset based on the error field, a second meteorological dataset is constructed according to the updated first meteorological dataset.

[0023] Optionally, constructing a prediction model corresponding to the observation area according to the temporal feature data and scale feature data corresponding to the observation area in the first meteorological dataset includes:

[0024] Obtaining the temporal feature data corresponding to the observation area in the first meteorological dataset, and constructing a temporal convolutional module using the temporal feature data;

[0025] Obtaining the scale feature data corresponding to the observation area in the first meteorological dataset, and constructing a squeeze-and-excitation network module using the scale feature data;

[0026] Constructing a prediction model corresponding to the observation area using the temporal convolutional module and the squeeze-and-excitation network module.

[0027] Optionally, constructing a prediction model corresponding to the observation area using the temporal convolutional module and the squeeze-and-excitation network module includes:

[0028] Constructing a temporal convolutional network corresponding to the temporal feature data using the temporal convolutional module, and constructing a squeeze-and-excitation network corresponding to the scale feature data using the squeeze-and-excitation network module;

[0029] Constructing a prediction model corresponding to the observation area according to the temporal convolutional network and the squeeze-and-excitation network; wherein, the prediction model is trained using historical meteorological forecast data and historical meteorological analysis data.

[0030] Optionally, updating the prediction model using the difference data in the second meteorological dataset includes:

[0031] Constructing revised forecast data corresponding to the observation area using the difference data in the second meteorological dataset;

[0032] Updating the historical meteorological analysis data using the revised forecast data, and updating the prediction model using the updated historical meteorological analysis data and historical meteorological forecast data.

[0033] Optionally, obtaining the meteorological forecast data corresponding to the observation point at the current moment, and obtaining the meteorological analysis data corresponding to the meteorological forecast data using the updated prediction model includes:

[0034] Obtaining the meteorological forecast data corresponding to the observation point at the current moment, and updating the first meteorological dataset and the second meteorological dataset using the meteorological forecast data;

[0035] After updating the prediction model using the updated first meteorological dataset and the second meteorological dataset, obtaining the temporal feature data and scale feature data corresponding to the meteorological forecast data according to the updated prediction model;

[0036] Determine the meteorological analysis data corresponding to the meteorological forecast data according to the temporal feature data and the scale feature data.

[0037] In a second aspect, the present invention provides a marine meteorological forecast data processing system, which includes:

[0038] A data acquisition unit, configured to obtain ship observation data corresponding to an observation point based on the position data of the observation point, and use the ship observation data to obtain historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point;

[0039] A first meteorological data set generation unit, configured to obtain random points corresponding to the observation point, construct an observation area corresponding to the observation point by using the random points, and construct a first meteorological data set by using the historical meteorological forecast data and the historical meteorological analysis data included in the observation area;

[0040] A second meteorological data set generation unit, configured to determine the difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, and construct a second meteorological data set corresponding to the observation area by using the difference data;

[0041] A prediction model construction unit, configured to construct a prediction model corresponding to the observation area according to the temporal feature data and the scale feature data corresponding to the observation area in the first meteorological data set, and update the prediction model by using the difference data in the second meteorological data set;

[0042] A meteorological forecast data processing unit, configured to obtain the meteorological forecast data corresponding to the observation point at the current moment, and obtain the meteorological analysis data corresponding to the meteorological forecast data by using the updated prediction model.

[0043] In a third aspect, an embodiment of the present invention further provides an electronic device, including a processor and a memory, where the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the steps of the marine meteorological forecast data processing method provided in the first aspect.

[0044] In a fourth aspect, an embodiment of the present invention further provides a storage medium, which stores computer executable instructions. When the computer executable instructions are called and executed by a processor, the computer executable instructions cause the processor to implement the steps of the marine meteorological forecast data processing method provided in the first aspect.

[0045] A method, system and electronic device for processing marine meteorological forecast data provided by an embodiment of the present invention. In the process of processing marine meteorological forecast data, the method first obtains ship observation data corresponding to an observation point based on the position data of the observation point, and uses the ship observation data to obtain meteorological forecast historical data and meteorological analysis historical data corresponding to the observation point; then obtains random points corresponding to the observation point, uses the random points to construct an observation area corresponding to the observation point, and uses the meteorological forecast historical data and meteorological analysis historical data included in the observation area to construct a first meteorological data set; subsequently, determines the difference data between the meteorological forecast historical data and the meteorological analysis historical data in the observation area at the same historical moment, and uses the difference data to construct a second meteorological data set corresponding to the observation area; then constructs a prediction model corresponding to the observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set, and updates the prediction model using the difference data in the second meteorological data set; finally, obtains the meteorological forecast data corresponding to the observation point at the current moment, and uses the updated prediction model to obtain the meteorological analysis data corresponding to the meteorological forecast data. This method constructs a new data set using the difference data between the forecast data and the analysis data included in the limited meteorological historical data, and fuses this data set with the meteorological historical data for the generation and update process of the prediction model, realizing the full utilization of the time characteristics and spatial characteristics of the data to process the meteorological forecast data under limited data conditions, thereby fully capturing the dynamic changes of meteorological variables and improving the accuracy of marine meteorological forecasts.

[0046] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention are achieved and obtained by the structures specifically pointed out in the specification, claims and drawings.

[0047] To make the above objectives, features and advantages of the present invention more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. Description of the Drawings

[0048] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0049] Figure 1 It is a flowchart of a method for processing marine meteorological forecast data provided by an embodiment of the present invention;

[0050] Figure 2Flow chart of step S101 in a method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0051] Figure 3 Flow chart of step S102 in a method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0052] Figure 4 Flow chart of step S103 in a method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0053] Figure 5 In step S103 of a method for processing offshore meteorological forecast data provided by an embodiment of the present invention, flow chart for constructing a prediction model corresponding to an observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set;

[0054] Figure 6 Flow chart of step S503 in a method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0055] Figure 7 In step S103 of a method for processing offshore meteorological forecast data provided by an embodiment of the present invention, flow chart for updating the prediction model by using the difference data in the second meteorological data set;

[0056] Figure 8 Flow chart of step S105 in a method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0057] Figure 9 Flow chart of another method for processing offshore meteorological forecast data provided by an embodiment of the present invention;

[0058] Figure 10 Structure diagram of an offshore meteorological forecast data processing system provided by an embodiment of the present invention;

[0059] Figure 11 Structure diagram of an electronic device provided by an embodiment of the present invention.

[0060] Icon:

[0061] 1010 - Data acquisition unit; 1020 - First meteorological data set generation unit; 1030 - Second meteorological data set generation unit; 1040 - Prediction model construction unit; 1050 - Meteorological forecast data processing unit;

[0062] 101 - Processor; 102 - Memory; 103 - Bus; 104 - Communication interface. Detailed implementation manners

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0064] In the process of marine weather forecasting, due to the limitation of marine environmental factors, it is impossible to set up many fixed observation stations. Therefore, non-fixed observation points such as ships are needed to obtain and predict marine weather data. Since the amount of data of such ship observation data is small and the real-time performance of ship observation data is poor, it is difficult to obtain marine weather data within a relatively wide time period, and there is a problem of poor quality of meteorological data, thus it is impossible to fully capture the dynamic changes of meteorological variables, resulting in a low accuracy of marine weather forecasting. Based on this, the embodiments of the present invention provide a marine weather forecasting data processing method, system, and electronic device. This method constructs a new data set using the difference data between the forecast data and the analysis data contained in the limited meteorological historical data, and fuses this data set with the meteorological historical data for the generation and update process of the prediction model, realizing the full utilization of the time characteristics and spatial characteristics of the data to process meteorological forecast data under limited data conditions, thus fully capturing the dynamic changes of meteorological variables and improving the accuracy of marine weather forecasting.

[0065] To facilitate the understanding of this embodiment, first, a marine weather forecasting data processing method disclosed in the embodiments of the present invention will be introduced in detail. This method is as Figure 1 shown and includes:

[0066] Step S101, obtaining the ship observation data corresponding to the observation point based on the position data of the observation point, and using the ship observation data to obtain the meteorological forecast historical data and meteorological analysis historical data corresponding to the observation point.

[0067] In the process of conducting marine weather forecasting, the position data of the observation point is the starting key information for the entire process. Generally speaking, the observation point can be a non-fixed observation point such as a ship, or a fixed observation point such as an island in the ocean or an oil extraction facility. The position data can be information such as longitude and latitude coordinates obtained through a satellite positioning system (such as GPS, Beidou, etc.). Based on these position data, precise matching and screening can be carried out in the relevant database of ship observation data, so as to obtain the ship observation data corresponding to the observation point at different time points. The ship observation data contains various meteorological historical data. Using the position data of the observation point as an index, precise retrieval can be carried out in the relevant database of ship observation data, so as to obtain meteorological forecast historical data related to the observation point (such as weather prediction information for the observation point issued at different historical times, including temperature, precipitation probability, etc.) and meteorological analysis historical data (such as analysis reports on the past meteorological conditions of the observation point, including climate trend analysis, etc.).

[0068] Step S102: Obtain the random points corresponding to the observation point, construct the observation area corresponding to the observation point by using the random points, and construct the first meteorological data set by using the meteorological forecast historical data and meteorological analysis historical data included in the observation area.

[0069] When obtaining the random points corresponding to the observation point, a specific random algorithm will be used to obtain several random points within a certain range around the observation point. The distribution range and quantity of these random points can be set according to actual needs and accuracy requirements. By constructing through the ranges included in these random points, an observation area containing the observation point and all random points is obtained. The observation area can be an irregular polygon area or a specific geometric shape according to requirements. After the construction of the observation area is completed, all the meteorological forecast historical data and meteorological analysis historical data within the observation area will be collected and sorted, duplicate and invalid data records will be removed, and the valid data will be organized according to a certain format and standard, so as to construct the first meteorological data set. This data set contains various meteorological-related information within the observation area in the past period of time, providing basic data support for subsequent analysis and modeling.

[0070] Step S103: Determine the difference data between the meteorological forecast historical data and the meteorological analysis historical data in the observation area at the same historical moment, and construct the second meteorological data set corresponding to the observation area by using the difference data.

[0071] Since there is less ship observation data in the observation points, it is necessary to expand it. By determining the difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, the corresponding second meteorological data set of the observation area is constructed. Specifically, after the first meteorological data set is constructed, by comparing and analyzing the historical meteorological forecast data and the historical meteorological analysis data in the first meteorological data set moment by moment, the corresponding difference data between the two can be obtained. For example, compare the difference between the predicted wind speed in the meteorological forecast at a certain moment and the recorded wind speed in the actual meteorological analysis, and summarize, collect and sort it to construct the corresponding second meteorological data set of the observation area.

[0072] Step S104: Construct a prediction model corresponding to the observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set, and update the prediction model with the difference data in the second meteorological data set.

[0073] During the implementation of this step, the data in the first meteorological data set will be deeply analyzed to extract the time series feature data corresponding to the observation area (such as the change trend of meteorological elements over time, periodic laws, etc.) and scale feature data (such as the distribution characteristics of meteorological elements at different spatial scales, etc.). In the actual scenario, based on these feature data, the corresponding modeling algorithm (such as neural network algorithm, time series analysis algorithm, etc. in machine learning) is selected to construct the prediction model corresponding to the observation area. This prediction model can predict the future meteorological conditions according to the input meteorological-related information. Then, the difference data in the second meteorological data set is used to update and optimize the constructed prediction model. By inputting the difference data as feedback information into the prediction model, the parameters and structure of the model are adjusted, so that the prediction model can better adapt to the actual meteorological situation and improve the accuracy and reliability of the prediction.

[0074] Step S105: Obtain the meteorological forecast data corresponding to the observation point at the current moment, and use the updated prediction model to obtain the meteorological analysis data corresponding to the meteorological forecast data.

[0075] It can be seen that this method constructs a new data set by using the difference data between the forecast data and the analysis data contained in the limited meteorological historical data, and fuses this data set with the meteorological historical data for the generation and update process of the prediction model, realizing the full utilization of the time and space characteristics of the data to process the meteorological forecast data under limited data conditions, so as to fully capture the dynamic changes of meteorological variables and improve the accuracy of marine meteorological forecasts.

[0076] Optionally, step S101 of obtaining the ship observation data corresponding to the observation point based on the position data of the observation point and using the ship observation data to obtain the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point is as follows Figure 2 shown, and includes:

[0077] Step S201: Determine the ship observation data collected by the meteorological ship in the observation point according to the position data determined by the observation point;

[0078] Step S202: Determine the corresponding ocean area between the meteorological ship and the observation point, and obtain the corresponding historical meteorological data in the ocean area;

[0079] Step S203: Determine the historical meteorological forecast data and historical meteorological analysis data based on the historical meteorological data.

[0080] In the specific implementation process, the ship observation data collected by the meteorological ship in the observation point is determined through the position data corresponding to the observation point. Furthermore, the historical meteorological data in this area is collected based on the ocean area between the meteorological ship and the observation point, and the historical meteorological forecast data and historical meteorological analysis data included in the historical meteorological data are used for the subsequent construction of the first meteorological data set. Optionally, step S102 of obtaining the random points corresponding to the observation point, using the random points to construct the observation area corresponding to the observation point, and using the historical meteorological forecast data and historical meteorological analysis data included in the observation area to construct the first meteorological data set is as follows Figure 3 shown, and includes:

[0081] Step S301: Obtain multiple random points corresponding to the observation point in the ocean area between the meteorological ship and the observation point;

[0082] Step S302: Use the position data of the multiple random points to determine the boundary of the area corresponding to the observation point, and use the area boundary to construct the observation area corresponding to the observation point;

[0083] Step S303: Obtain the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point and the random points in the observation area;

[0084] Step S304: Use the historical meteorological forecast data and historical meteorological analysis data to construct the first meteorological data set.

[0085] Specifically, first, multiple random points are obtained in the ocean area between the meteorological ship and the observation point, and the boundary of the area corresponding to the observation point is determined using the position data of the random points, so as to construct the observation area corresponding to the observation point using these area boundaries. Subsequently, the historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point and the random points in the observation area are intercepted for constructing the first meteorological data set.

[0086] Optionally, determine the difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, and use the difference data to construct the second meteorological data set S103 corresponding to the observation area, as Figure 4 shown, including:

[0087] Step S401, obtain the historical meteorological forecast data and the historical meteorological analysis data corresponding to the observation area at the same historical moment, and determine the meteorological forecast value and the meteorological observation value corresponding to the observation area based on the historical meteorological forecast data and the historical meteorological analysis data;

[0088] Step S402, use the meteorological forecast value and the meteorological observation value to determine the difference data, and after using the inverse distance weighted interpolation method to transfer the difference data to the grid field corresponding to the observation area, use the grid field to generate the error field corresponding to the observation area;

[0089] Step S403, update the first meteorological data set based on the error field, and construct the second meteorological data set according to the updated first meteorological data set.

[0090] The acquisition process of the second meteorological data set is determined based on the difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same moment. First, determine the meteorological forecast value and the meteorological observation value corresponding to the observation area based on the historical meteorological forecast data and the historical meteorological analysis data. After determining the difference data using the meteorological forecast value and the meteorological observation value, use the inverse distance weighted interpolation method to transfer the difference data to the grid field corresponding to the observation area, and use the grid field to generate the error field corresponding to the observation area. Finally, update the first meteorological data set based on the error field, and construct the second meteorological data set according to the updated first meteorological data set.

[0091] Optionally, construct a prediction model corresponding to the observation area according to the time series feature data and the scale feature data corresponding to the observation area in the first meteorological data set, as Figure 5 shown, including:

[0092] Step S501, obtain the time series feature data corresponding to the observation area in the first meteorological data set, and use the time series feature data to construct a time series convolutional module;

[0093] Step S502, obtain the scale feature data corresponding to the observation area in the first meteorological data set, and use the scale feature data to construct a squeeze-and-excitation network module;

[0094] Step S503, use the time series convolutional module and the squeeze-and-excitation network module to construct a prediction model corresponding to the observation area.

[0095] In the existing correction algorithms based on fixed stations and machine learning, the models used are mainly based on models such as LightGBM or random forests. Using various observation data and static environmental feature data of a station over a period of time as input, a corresponding prediction model is obtained after training the relevant machine learning models. Since the marine environment is more complex than that on land, it is difficult to obtain a sufficient number of fixed observation points, and there is a lack of observation data that is continuous in time and fixed in position for training the model. In this embodiment, a temporal convolutional module is constructed through the temporal feature data corresponding to the observation area in the first meteorological dataset, and then a squeeze-and-excitation network module is constructed using the scale feature data corresponding to the observation area in the first meteorological dataset.

[0096] Optionally, the step S503 of constructing a prediction model corresponding to the observation area using the temporal convolutional module and the squeeze-and-excitation network module is as Figure 6 shown and includes:

[0097] Step S601, constructing a temporal convolutional network corresponding to the temporal feature data using the temporal convolutional module, and constructing a squeeze-and-excitation network corresponding to the scale feature data using the squeeze-and-excitation network module;

[0098] Step S602, constructing a prediction model corresponding to the observation area according to the temporal convolutional network and the squeeze-and-excitation network; wherein, the prediction model is trained using historical meteorological forecast data and historical meteorological analysis data.

[0099] The temporal convolutional network can utilize the 3DTCN (Temporal Convolutional Network) part: mainly responsible for capturing the temporal dependencies in the input data. To enhance the model's temporal modeling ability, dilated convolution is introduced in the TCN. Dilated convolution can capture a larger range of temporal features without increasing the model parameters by expanding the receptive field of the convolutional kernel, thereby effectively modeling long-term dependencies. In 3DTCN, by stacking multiple layers of dilated convolution, each layer using a different dilation rate (such as 1, 2, 4, 8, etc.), the model can capture dependencies with a longer time span with fewer layers. In addition, to ensure that the prediction results do not depend on future information, 3DTCN adopts causal convolution, that is, the convolutional operation at each time step only considers the current and past time steps, ensuring that the model output is consistent with the actual meteorological forecast time series. To improve the training effect of the model, residual connections can also be introduced between each layer of dilated convolution. This design can effectively transmit the gradients of the deep convolutional network, avoiding the common gradient vanishing problem in deep networks, thereby stabilizing the model training process.

[0100] The compression excitation network can utilize the SEU-Net network structure and is responsible for extracting multi-scale spatial information from meteorological data. SEU-Net is based on the U-Net structure and combines the Squeeze-and-Excitation (SE) module, which further enhances the dynamic adjustment ability of the importance of feature channels. The encoder part of SEU-Net consists of a series of convolutional layers and pooling layers. After each convolutional operation, the resolution of the feature map is reduced by max pooling to gradually extract spatial features at different scales. During this process, the SE module is inserted after each encoder layer to compress the feature map output by the convolutional layer and dynamically adjust the importance of different channels through an activation function.

[0101] Optionally, the prediction model is updated using the difference data in the second meteorological dataset, such as Figure 7 shown, including:

[0102] Step S701, constructing the revised forecast data corresponding to the observation area using the difference data in the second meteorological dataset;

[0103] Step S702, updating the historical meteorological analysis data using the revised forecast data, and updating the prediction model using the updated historical meteorological analysis data and historical meteorological forecast data.

[0104] The update process of the prediction model can be implemented in combination with the training process of the model. Specifically, the first meteorological dataset is used as the pre-training dataset to train the TCN + SEU-Net revised model parameters. The inputs of the model are the meteorological forecast data at the previous moment, the meteorological analysis data at the previous moment, and the forecast data at the current moment, and the output of the model is the meteorological analysis data at the current moment; subsequently, the forecast data revised by the error field generated by the observations and forecasts of the mobile observation points is used to replace the meteorological analysis data to obtain the second meteorological dataset as the fine-tuning training dataset, so as to update the historical meteorological analysis data using the revised forecast data and update the prediction model using the updated historical meteorological analysis data and historical meteorological forecast data.

[0105] Optionally, step S105 of obtaining the meteorological analysis data corresponding to the meteorological forecast data at the current moment for the observation point using the updated prediction model, such as Figure 8 shown, including:

[0106] Step S801, obtaining the meteorological forecast data corresponding to the observation point at the current moment and updating the first meteorological dataset and the second meteorological dataset using the meteorological forecast data;

[0107] In step S802, after updating the prediction model using the updated first meteorological dataset and second meteorological dataset, obtain the time series feature data and scale feature data corresponding to the meteorological forecast data according to the updated prediction model;

[0108] In step S803, determine the meteorological analysis data corresponding to the meteorological forecast data according to the time series feature data and scale feature data.

[0109] During the actual execution process, in the process of processing the marine meteorological forecast data, obtain the meteorological forecast data corresponding to the observation point at the current moment, and update the first meteorological dataset and the second meteorological dataset using the meteorological forecast data. Then use the updated first meteorological dataset and second meteorological dataset to update the prediction model. Subsequently, obtain the time series feature data and scale feature data corresponding to the meteorological forecast data according to the updated prediction model, and then determine the meteorological analysis data corresponding to the meteorological forecast data according to the time series feature data and scale feature data.

[0110] As Figure 9 shown in the flowchart of another method for processing marine meteorological forecast data, which mainly involves six steps, specifically as follows:

[0111] In step S1, collect the reanalysis data of various meteorological elements globally, the historical meteorological forecast data, and collect the ship observation data for a certain range of sea areas.

[0112] In step S2, based on the observation point and the random points around the observation point generated based on the observation point, calculate the boundary of the local area centered on the observation point, intercept the corresponding reanalysis data and historical forecast data, and construct a pre-training dataset.

[0113] In step S3, based on the position of the observation point, calculate the boundary of the local area centered on the observation point. By matching the corresponding meteorological forecast data, use the inverse distance weighted interpolation method to transfer the error between the observed value and the forecast value to this local area to form an error field, correct the forecast at the corresponding moment, and construct a fine-tuning training dataset.

[0114] Step S4: Construct a meteorological forecast correction model based on 3DTCN + SEU-Net. The 3DTCN part combines the Temporal Convolutional Network (TCN) and the U-Net structure with a Squeeze-and-Excitation (SE) module to effectively model and correct the spatio-temporal features of meteorological data. The 3DTCN (Temporal Convolutional Network) part is mainly responsible for capturing the temporal dependencies in the input data. To enhance the model's temporal modeling ability, dilated convolution is introduced in the TCN. Dilated convolution can capture a larger range of temporal features without increasing the model parameters by expanding the receptive field of the convolutional kernel, thus effectively modeling long-term dependencies. In 3DTCN, by stacking multiple layers of dilated convolution with different dilation rates (such as 1, 2, 4, 8, etc.) in each layer, the model can capture dependencies over a longer time span with fewer layers. In addition, to ensure that the prediction results do not depend on future information, causal convolution is adopted in 3DTCN, that is, the convolution operation at each time step only considers the current and past time steps to ensure that the model output is consistent with the time series of the actual meteorological forecast. To improve the training effect of the model, residual connections are also introduced between each layer of dilated convolution. This design can effectively transmit the gradients of the deep convolutional network and avoid the common problem of gradient vanishing in deep networks, thus stabilizing the training process of the model. The SEU-Net part is responsible for extracting multi-scale spatial information from meteorological data. SEU-Net is based on the classic U-Net structure and combines the Squeeze-and-Excitation (SE) module to further enhance the dynamic adjustment ability of the importance of feature channels. The encoder part of SEU-Net consists of a series of convolutional layers and pooling layers. After each convolution operation, the resolution of the feature map is reduced by max pooling to gradually extract spatial features at different scales. During this process, the SE module is inserted after each encoder layer to compress the feature map output by the convolutional layer and dynamically adjust the importance of different channels through an activation function.

[0115] Step S5: During the model training process, the TCN + SEU-Net correction model parameters are trained using a pre-training dataset composed of analysis data and historical forecast data. The inputs of the model are the meteorological forecast at the previous moment, the reanalysis data at the previous moment, and the forecast data at the current moment, and the output of the model is the reanalysis data at the current moment. After the pre-training network, a fine-tuning training dataset composed of the forecast data corrected by the error field generated by the observed values and forecasts of the mobile observation points instead of the reanalysis data is used, and finally a pre-trained correction model is obtained.

[0116] Step S6: Obtain the ship observation data at the previous moment and the forecast data at the previous moment, generate the forecast correction data at the previous moment using the method in Step S3, and input the forecast data and correction data at the previous moment and the forecast data at the current moment into the 3D TCN+SEU-Net correction model trained in Step S5 to output the corrected forecast at the current moment in the local area around the observation point.

[0117] The above method transfers the observed values of the ship observation points to the grid field through inverse distance weighting, distributes the correction difference to a local space to obtain a difference field, and superimposes it on the forecast field in the original space to replace the actual data at the adjacent moment to correct the forecast data for the next moment. In addition, by constructing a 3D TCN + SE-U-Net structure and adding an SE module to the original U-Net encoder and decoder, while strengthening the extraction of spatial features, a 3D TCN model is added in front of the U-Net to learn the temporal features of continuous forecasts, enabling the entire model to have strong temporal and spatial feature learning capabilities. Thus, it is not necessary to obtain the reanalysis data at the current adjacent moment to guide the correction of the forecast data, realizing the correction of meteorological forecasts relying only on mobile observation points; this method does not require obtaining the reanalysis data at the current adjacent moment to guide the correction of the forecast data, realizing the correction of meteorological forecasts relying only on mobile observation points.

[0118] As can be seen from the marine meteorological forecast data processing method mentioned in the above embodiments, this method constructs a new dataset using the difference data between the forecast data and the analysis data contained in the limited meteorological historical data, and fuses this dataset with the meteorological historical data for the generation and update process of the prediction model, realizing the processing of meteorological forecast data by making full use of the temporal and spatial features of the data under limited data conditions, thereby fully capturing the dynamic changes of meteorological variables, and thus correcting the local sea area around the observation point without relying on a large number of fixed observation points, which is more suitable for the marine meteorological forecast correction scenario and improves the accuracy of marine meteorological forecasts.

[0119] Corresponding to the marine meteorological forecast data processing method provided in the foregoing embodiments, an embodiment of the present invention provides a marine meteorological forecast data processing system, as Figure 10As shown in the figure, the system includes:

[0120] A data acquisition unit 1010, configured to obtain ship observation data corresponding to an observation point based on the position data of the observation point, and obtain historical meteorological forecast data and historical meteorological analysis data corresponding to the observation point by using the ship observation data;

[0121] A first meteorological data set generation unit 1020, configured to obtain random points corresponding to the observation point, construct an observation area corresponding to the observation point by using the random points, and construct a first meteorological data set by using the historical meteorological forecast data and historical meteorological analysis data included in the observation area;

[0122] A second meteorological data set generation unit 1030, configured to determine difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, and construct a second meteorological data set corresponding to the observation area by using the difference data;

[0123] A prediction model construction unit 1040, configured to construct a prediction model corresponding to the observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set, and update the prediction model by using the difference data in the second meteorological data set;

[0124] A meteorological forecast data processing unit 1050, configured to obtain meteorological forecast data corresponding to the observation point at the current moment, and obtain meteorological analysis data corresponding to the meteorological forecast data by using the updated prediction model.

[0125] It can be seen from the marine meteorological forecast data processing system mentioned in the above embodiments that the system constructs a new data set by using the difference data between the forecast data and the analysis data included in the limited meteorological historical data, and fuses the data set with the meteorological historical data for the generation and update process of the prediction model, realizing the full utilization of the time characteristics and space characteristics of the data to process the meteorological forecast data under limited data conditions, thereby fully capturing the dynamic changes of meteorological variables and improving the accuracy of marine meteorological forecasts.

[0126] The marine meteorological forecast data processing system provided by the embodiments of the present invention has the same implementation principle and the same technical effects as those of the foregoing embodiments of the marine meteorological forecast data processing method. For the sake of brief description, for the parts not mentioned in the system embodiments, reference may be made to the corresponding content in the foregoing embodiments of the marine meteorological forecast data processing method.

[0127] This embodiment also provides an electronic device, and the structural schematic diagram of the electronic device is as Figure 11As shown, the device includes a processor 101 and a memory 102; among them, the memory 102 is used to store one or more computer instructions, and the one or more computer instructions are executed by the processor to implement the steps of the above-mentioned marine weather forecast data processing method.

[0128] Figure 11 The electronic device shown further includes a bus 103 and a communication interface 104, and the processor 101, the communication interface 104 and the memory 102 are connected through the bus 103.

[0129] Among them, the memory 102 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory, such as at least one disk memory. The bus 103 may be an ISA bus, a PCI bus or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For the sake of simplicity of representation, Figure 11 only a two-way arrow is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0130] The communication interface 104 is used to connect to at least one user terminal and other network units through a network interface, and send the encapsulated IPv4 packet or IPv4 packet to the user terminal through the network interface.

[0131] The processor 101 may be an integrated circuit chip with the ability to process signals. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor 101 or the instructions in the form of software. The above-mentioned processor 101 may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (DSP for short), an application specific integrated circuit (ASIC for short), a field-programmable gate array (FPGA for short), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present disclosure can be directly embodied as being executed and completed by the hardware decoding processor, or can be executed and completed by the combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 102, and the processor 101 reads the information in the memory 102 and combines its hardware to complete the steps of the method in the foregoing embodiments.

[0132] An embodiment of the present invention also provides a storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the method for processing marine weather forecast data in the foregoing embodiments.

[0133] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, equipment, and methods can be implemented in other ways. The system embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other may be through some communication interfaces, and the indirect coupling or communication connection of the devices or units may be in an electrical, mechanical, or other form.

[0134] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may 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.

[0135] In addition, each functional unit in various embodiments of the present invention may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit.

[0136] If the above function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a 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 several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0137] Finally, it should be noted that the above-described embodiments are only specific implementation manners of the present invention, used to illustrate the technical solutions of the present invention, and are not intended to limit it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the technical field of the present invention can still modify the technical solutions described in the foregoing embodiments or easily think of changes, or make equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for processing marine meteorological forecast data, characterized in that The method includes: Obtaining the ship observation data corresponding to the observation point based on the position data of the observation point, and obtaining the historical weather forecast data and historical weather analysis data corresponding to the observation point by using the ship observation data; Obtaining the random points corresponding to the observation point, constructing the observation area corresponding to the observation point by using the random points, and constructing the first meteorological data set by using the historical weather forecast data and the historical weather analysis data included in the observation area; Determining the difference data between the historical weather forecast data and the historical weather analysis data in the observation area at the same historical moment, and constructing the second meteorological data set corresponding to the observation area by using the difference data; Constructing a prediction model corresponding to the observation area according to the time series feature data and scale feature data corresponding to the observation area in the first meteorological data set, and updating the prediction model by using the difference data in the second meteorological data set; Obtaining the weather forecast data corresponding to the observation point at the current moment, and obtaining the weather analysis data corresponding to the weather forecast data by using the updated prediction model.

2. The marine weather forecast data processing method according to claim 1, wherein Obtaining the ship observation data corresponding to the observation point based on the position data of the observation point, and obtaining the historical weather forecast data and historical weather analysis data corresponding to the observation point by using the ship observation data, including: Determining the ship observation data collected by the meteorological ship in the observation point according to the position data determined by the observation point; Determining the corresponding ocean area between the meteorological ship and the observation point, and obtaining the corresponding meteorological historical data in the ocean area; Determining the historical weather forecast data and the historical weather analysis data based on the meteorological historical data.

3. The method for processing marine weather forecast data according to claim 2, wherein, Obtaining the random points corresponding to the observation point, constructing the observation area corresponding to the observation point by using the random points, and constructing the first meteorological data set by using the historical weather forecast data and the historical weather analysis data included in the observation area, including: Obtaining a plurality of random points corresponding to the observation point in the ocean area between the meteorological ship and the observation point; Determining the regional boundary corresponding to the observation point by using the position data of the plurality of random points, and constructing the observation area corresponding to the observation point by using the regional boundary; Obtaining the historical weather forecast data and the historical weather analysis data corresponding to the observation point and the random points in the observation area; Constructing the first meteorological data set by using the historical weather forecast data and the historical weather analysis data.

4. The method for processing marine weather forecast data according to claim 1, wherein Determining the difference data between the historical weather forecast data and the historical weather analysis data in the observation area at the same historical moment, and constructing the second meteorological data set corresponding to the observation area by using the difference data, including: Obtaining the historical weather forecast data and the historical weather analysis data corresponding to the observation area at the same historical moment, and determining the weather forecast value and weather observation value corresponding to the observation area based on the historical weather forecast data and the historical weather analysis data; Determine the difference data using the meteorological forecast values and the meteorological observation values, and after transferring the difference data to the grid field corresponding to the observation area using the inverse distance weighted interpolation method, generate the error field corresponding to the observation area using the grid field; After updating the first meteorological data set based on the error field, construct the second meteorological data set according to the updated first meteorological data set.

5. The method for processing marine meteorological forecast data according to claim 1, wherein Construct a prediction model corresponding to the observation area according to the time series feature data and the scale feature data corresponding to the observation area in the first meteorological data set, including: Obtain the time series feature data corresponding to the observation area in the first meteorological data set, and construct a time series convolutional module using the time series feature data; Obtain the scale feature data corresponding to the observation area in the first meteorological data set, and construct a squeeze-and-excitation network module using the scale feature data; Construct the prediction model corresponding to the observation area using the time series convolutional module and the squeeze-and-excitation network module.

6. The method for processing marine weather forecast data according to claim 5, wherein Construct the prediction model corresponding to the observation area using the time series convolutional module and the squeeze-and-excitation network module, including: Construct a time series convolutional network corresponding to the time series feature data using the time series convolutional module, and construct a squeeze-and-excitation network corresponding to the scale feature data using the squeeze-and-excitation network module; Construct the prediction model corresponding to the observation area according to the time series convolutional network and the squeeze-and-excitation network; wherein, the prediction model is trained using the meteorological forecast historical data and the meteorological analysis historical data.

7. The marine weather forecast data processing method according to claim 1, characterized in that Update the prediction model using the difference data in the second meteorological data set, including: Construct the corrected forecast data corresponding to the observation area using the difference data in the second meteorological data set; Update the meteorological analysis historical data using the corrected forecast data, and update the prediction model using the updated meteorological analysis historical data and the meteorological forecast historical data.

8. The marine weather forecast data processing method according to claim 1, characterized in that Obtain the meteorological forecast data corresponding to the observation point at the current moment, and obtain the meteorological analysis data corresponding to the meteorological forecast data using the updated prediction model, including: Obtain the meteorological forecast data corresponding to the observation point at the current moment, and update the first meteorological data set and the second meteorological data set using the meteorological forecast data; After updating the prediction model using the updated first meteorological data set and the second meteorological data set, obtain the time series feature data and the scale feature data corresponding to the meteorological forecast data according to the updated prediction model; Determine the meteorological analysis data corresponding to the meteorological forecast data according to the time series feature data and the scale feature data.

9. A marine weather forecast data processing system, characterized in that, The system includes: A data acquisition unit, configured to obtain the ship observation data corresponding to the observation point based on the position data of the observation point, and obtain the meteorological forecast historical data and the meteorological analysis historical data corresponding to the observation point using the ship observation data; The first meteorological dataset generation unit is configured to obtain random points corresponding to the observation points, construct an observation area corresponding to the observation points by using the random points, and construct a first meteorological dataset by using the historical meteorological forecast data and the historical meteorological analysis data included in the observation area; The second meteorological dataset generation unit is configured to determine difference data between the historical meteorological forecast data and the historical meteorological analysis data in the observation area at the same historical moment, and construct a second meteorological dataset corresponding to the observation area by using the difference data; The prediction model construction unit is configured to construct a prediction model corresponding to the observation area according to the time series feature data and the scale feature data corresponding to the observation area in the first meteorological dataset, and update the prediction model by using the difference data in the second meteorological dataset; The meteorological forecast data processing unit is configured to obtain meteorological forecast data corresponding to the observation points at the current moment, and obtain meteorological analysis data corresponding to the meteorological forecast data by using the updated prediction model.

10. An electronic device, characterized in that, It includes a processor and a memory. The memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the steps of the marine meteorological forecast data processing method according to any one of claims 1 to 8.

Images