A Method and Device for Correcting Wind Element Data in Meteorological Forecasts Based on Data Decomposition

Abstract

This application provides a method and apparatus for correcting wind element data in weather forecasts based on data decomposition. Using the data decomposition-based wind element data correction model trained in this application, the wind element sub-data related to wind speed in the wind element data of the weather forecast for the future time period of the point to be corrected is processed to obtain the corrected wind speed. For example, the data decomposition-based wind element data correction model includes multiple wind element data correction sub-models. Each sub-model processes the wind element sub-data related to wind speed in the wind element data of the weather forecast for the future time period of the point to be corrected, obtaining its own output wind speed correction component. Then, based on the respective output wind speed correction components, the corrected wind speed for the future time period of the point to be corrected is obtained. Thus, this application can improve the accuracy of wind speed correction.

Description

Technical Field

[0001] This application relates to the field of wind energy technology, and in particular to a method and apparatus for correcting wind element data in meteorological forecasts based on data decomposition. Background Art

[0002] As wind power technology matures, the single-unit capacity of wind turbines and the scale of grid-connected wind farms continue to expand, and the proportion of wind power in the total power generation of the power system is increasing year by year. The penetrating power of wind farms is also increasing, leading to a series of increasingly prominent problems for the power system and seriously threatening its normal operation. Therefore, the need for timely and accurate prediction of wind power generation has been established, which can significantly enhance the safety, stability, economy, and controllability of the power system.

[0003] Wind data is a key input for predicting wind power generation, and it is obtained from weather forecast data. Summary of the Invention

[0004] This application discloses a method and apparatus for correcting wind element data in meteorological forecasts based on data decomposition.

[0005] In a first aspect, this application discloses a method for training a weather forecast wind element data correction model based on data decomposition, the method comprising:

[0006] Obtain the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0007] Calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period.

[0008] The calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data for the weather forecast of the first historical time period of the sample points.

[0009] Based on the random forest algorithm, wind element sub-data related to wind speed are determined in the second wind element data;

[0010] Obtain the historical measured wind speed of the sample points for the first historical time period;

[0011] Using the fully adaptive noise ensemble empirical mode decomposition (CEEMDAN) algorithm, the historical measured wind speed of the sample points in the first historical time period is decomposed to obtain multiple intrinsic mode functions (IMF) wind speed components and a residual component.

[0012] For any one of the multiple IMF wind speed components and a residual component, a wind element data correction sub-model corresponding to the component is trained using wind element sub-data related to wind speed as sample data and the component as labeled data.

[0013] At least based on the wind element data corresponding to each component, a wind element data correction model for meteorological forecasts can be obtained.

[0014] Secondly, this application discloses a method for correcting meteorological forecast wind element data based on data decomposition, the method comprising:

[0015] Obtain the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure, and sunshine.

[0016] Calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of the meteorological forecast for the point to be corrected and the historical measured wind speed of each second time period for the point to be corrected; the second time period is earlier than the first time period.

[0017] The calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data for the first time period of the weather forecast of the point to be corrected.

[0018] Based on the random forest algorithm, wind element sub-data related to wind speed are determined in the second wind element data;

[0019] Wind element sub-data related to wind speed is input into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on the output wind speed correction component.

[0020] The meteorological forecast wind element data correction model based on data decomposition is trained using any of the methods described in the first aspect.

[0021] Thirdly, this application discloses an apparatus for training a weather forecast wind element data correction model based on data decomposition, the apparatus comprising:

[0022] The first acquisition module is used to acquire the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0023] The first calculation module is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error. The preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period.

[0024] The first replacement module is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the meteorological forecast for the first historical time period of the sample point.

[0025] The first determination module is used to determine wind element sub-data related to wind speed from the second wind element data based on the random forest algorithm;

[0026] The second acquisition module is used to acquire the historical measured wind speed of the sample points in the first historical time period.

[0027] The decomposition module is used to decompose the historical measured wind speed of the sample points in the first historical time period using the fully adaptive noise set empirical mode decomposition (CEEMDAN) algorithm, and obtain multiple intrinsic mode functions (IMF) wind speed components and a residual component.

[0028] The training module is used to train a wind element data correction sub-model corresponding to any one of multiple IMF wind speed components and a residual component, using wind element sub-data related to wind speed as sample data and the component as labeled data.

[0029] The third acquisition module is used to obtain a weather forecast wind element data correction model based on data decomposition, at least according to the wind element data corresponding to each component.

[0030] Fourthly, this application discloses a weather forecast wind element data correction device based on data decomposition, the device comprising:

[0031] The fifth acquisition module is used to acquire the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed, and the wind element sub-data also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and light intensity.

[0032] The second calculation module is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of the weather forecast of the point to be corrected and the historical measured wind speed of each second time period of the point to be corrected; the second time period is earlier than the first time period.

[0033] The second replacement module is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the weather forecast for the first time period of the point to be corrected.

[0034] The second determination module is used to determine wind element sub-data related to wind speed from the second wind element data based on the random forest algorithm;

[0035] The processing module is used to input wind element sub-data related to wind speed into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on the output wind speed correction component.

[0036] The meteorological forecast wind element data correction model based on data decomposition is trained using any of the methods described in the first aspect.

[0037] Alternatively, the weather forecast wind element data correction model based on data decomposition can be trained using any of the devices described in the third aspect.

[0038] Fifthly, this application discloses an electronic device comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to perform the method as described in any of the preceding aspects.

[0039] Sixthly, this application discloses a non-transitory computer-readable storage medium that, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform the methods described in any of the preceding aspects.

[0040] In a seventh aspect, this application discloses a computer program product in which, when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to perform the method described in any of the preceding aspects.

[0041] The technical solution provided in this application may include the following beneficial effects:

[0042] When training the weather forecast wind element data correction model based on data decomposition, the wind element sub-data related to wind speed determined in the second wind element data is used. The second wind element data is obtained by replacing the "wind speed in the first wind element data" with the "wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed". In the second wind element data, the wind speed difference can be regarded as the wind speed. This wind speed is closer to the historical measured wind speed, which reduces the gap between the wind speed in the sample data and the historical measured wind speed. This is beneficial to improving the accuracy of the wind speed correction of the trained weather forecast wind element data correction model based on data decomposition.

[0043] Therefore, when correcting the wind speed for a future time period at a location to be corrected, the weather forecast wind element data correction model trained in this application can be used to process the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtain the corrected wind speed. For example, the weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtains its own output wind speed correction component. Then, based on its own output wind speed correction component, the corrected wind speed for the future time period at the location to be corrected can be obtained. In this way, the accuracy of wind speed correction can be improved through this application. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the steps of a method for training a weather forecast wind element data correction model based on data decomposition, as described in this application.

[0045] Figure 2 This is a flowchart illustrating the processing procedure of a random forest algorithm according to this application.

[0046] Figure 3 This is a schematic diagram of the network structure of one model in this application.

[0047] Figure 4 This is a schematic diagram of the network structure of one model in this application.

[0048] Figure 5This is a flowchart of the steps for testing a method for correcting meteorological forecast wind element data based on data decomposition, as described in this application.

[0049] Figure 6 This is a flowchart illustrating the steps of a meteorological forecast wind element data correction method based on data decomposition, as described in this application.

[0050] Figure 7 This is a structural block diagram of an apparatus for training a weather forecast wind element data correction model based on data decomposition, as described in this application.

[0051] Figure 8 This is a structural block diagram of a weather forecast wind element data correction device based on data decomposition according to this application.

[0052] Figure 9 This is a block diagram of an electronic device according to this application.

[0053] Figure 10 This is a block diagram of an electronic device according to this application. Detailed Implementation

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0055] Current weather forecast data is usually simulated based on the WRF (Weather Research and Forecasting Model). However, there is usually a certain deviation between the wind data in the simulated weather forecast data and the actual wind data. In other words, the accuracy of the wind data is poor, which may lead to low accuracy in wind power generation prediction.

[0056] Therefore, it is necessary to correct the wind data simulated by WRF so that the corrected wind data is as close as possible to the actual wind data. This can improve the accuracy of wind power generation prediction, which is of great significance for enhancing the safety, stability and economy of high-proportion renewable energy power systems.

[0057] The wind data that needs correction involves wind speed. Wind speed correction is usually based on machine learning models. However, according to statistics, the accuracy of wind speed correction based on machine learning models is still very low.

[0058] Therefore, this application is filed to improve the accuracy of wind speed correction.

[0059] Reference Figure 1 The diagram illustrates a flowchart of a method for training a weather forecast wind element data correction model based on data decomposition, which can be applied to electronic devices, including terminals or servers. Specifically, the method includes the following steps:

[0060] In step S101, the first wind element data of the meteorological forecast for the first historical time period of the sample point is obtained. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0061] Wind element sub-data can be wind element sub-data for various height levels of the sample points. Height levels can be 100 meters, 90 meters, 80 meters, 70 meters, 60 meters, and other height levels, etc.

[0062] Sample points can be grids, for example, weather forecast data is given with a grid as the minimum resolution. Grids can be rectangles, squares, triangles, pentagons, or hexagons, etc.

[0063] Wind element data in meteorological forecasts can be wind element data from weather forecasting models (WRF models). Weather forecasting models (WRF models) represent a new generation of non-hydrostatic equilibrium, high-resolution mesoscale meteorological research and forecasting numerical models, which mainly consider the forecasting of important weather events from cloud scale to weather scale.

[0064] Wind element data in weather forecasts can be understood as wind element data given by the weather forecasting model (WRF model) over a historical period.

[0065] In step S102, the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error is calculated. The preset benchmark wind speed error includes the first average value of the wind speed differences between the wind speed in the wind element data of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point. The second historical time periods are earlier than the first historical time periods.

[0066] Among them, multiple second historical time periods are all time periods earlier than the first historical time period, and no two second historical time periods overlap. Multiple second historical time periods can be time periods that are sequentially adjacent.

[0067] For any given second historical time period, the wind speed difference between the wind speed in the meteorological forecast wind element data of the sample point during that second historical time period and the historical measured wind speed of the sample point during that second historical time period can be calculated. The same applies to each other second historical time period, thereby obtaining the wind speed difference corresponding to multiple second historical time periods. Then, the average value between the wind speed differences corresponding to multiple second historical time periods can be calculated and used as the preset benchmark wind speed error.

[0068] The wind speed difference between two wind speeds can be the difference between the two wind speeds or the absolute value of the difference. For example, when calculating the wind speed difference between the wind speed in the first wind element data and the preset reference wind speed error, the difference between the wind speed in the first wind element data and the preset reference wind speed error can be calculated.

[0069] In step S103, the calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data of the meteorological forecast for the first historical time period of the sample point.

[0070] In the second wind element data, the wind speed difference can be regarded as the wind speed. This wind speed is the wind speed that is closer to the historical measured wind speed. It reduces the gap between the wind speed in the sample data and the historical measured wind speed, which is conducive to improving the accuracy of the wind speed correction of the meteorological forecast wind element data correction model based on data decomposition.

[0071] In step S104, based on the random forest algorithm, wind element sub-data related to wind speed is determined from the second wind element data.

[0072] Wind-related sub-data includes at least wind speed.

[0073] In one embodiment of this application, the correlation between each wind element sub-data in the second wind element data and wind speed can be determined based on the random forest algorithm, and the top-N wind element sub-data with the highest correlation to wind speed can be selected. N is a positive integer.

[0074] In this step, the random forest algorithm can be used to extract wind element sub-data that has a high correlation with wind speed from each wind element sub-data in the second wind element data.

[0075] The Random Forest algorithm is an algorithm that integrates multiple trees using the idea of ​​ensemble learning.

[0076] Among them, see Figure 2 The processing steps of the random forest algorithm include:

[0077] First, generate a random forest, for i = 1 to i = n tree (ntree (where the number of decision trees is).

[0078] (1) Assume that the sample feature dimension is M, and a constant m << M. Randomly select m feature subsets from the M features as the attribute set for the current decision tree split.

[0079] (2) During each split of the i-th tree, select the best variable and split point from the m features. Each tree grows to its maximum extent (without the pruning process) to generate the decision tree T i .

[0080] (3) Combine the generated n tree trees to form a random forest, and its output is determined by the voting mechanism of the tree classifiers.

[0081] Secondly, calculate the OOB data error. Assume that the training set size is N total .

[0082] for i = 1 to i = n tree .

[0083] (1) For the decision tree T i , use bootstrapping sampling to obtain N i samples from the training set as the training set for this tree. The data not drawn forms the OOB data set N b_i . N i < N total .

[0084] Bootstrapping sampling is a statistical method used to handle small sample data sets. It reconstructs the original distribution by sampling with replacement (there may be repeated observations in the sample).

[0085] (2) Obtain the calculation result of the decision tree T i for the OOB data set N b_i , and compare the calculation result with the actual result to obtain the calculation error E i .

[0086] For feature k (k = 1, 2,..., M), add noise interference to feature k in the OOB data set N b_i , obtain the calculation result of the decision tree T i for the perturbed data set N b_i_k , denoted as E i_k .

[0087] Then, calculate the importance degree D k of feature k, as shown in the following relational expression:

[0088]

[0089] Finally, the importance of each feature is ranked, and features with lower importance are removed, while those with higher importance are retained to form a feature variable dataset.

[0090] Furthermore, based on the processing procedure of the random forest algorithm described above, the correlation between each wind element sub-data in the second wind element data and wind speed can be calculated. That is, the importance score of each wind element sub-data in the second wind element data to wind speed can be obtained. Then, the top-N wind element sub-data with the highest correlation to wind speed can be selected from the various wind element sub-data in the second wind element data. N is a positive integer.

[0091] N is greater than or equal to 1, and N is less than or equal to the number of wind element sub-data in the wind element data of the weather forecast.

[0092] In this embodiment, the second wind element data is processed using the random forest algorithm. Wind element sub-data that is not suitable for training the data decomposition-based meteorological forecast wind element data correction model is removed, while wind element sub-data suitable for training the data decomposition-based meteorological forecast wind element data correction model is retained. This can improve the efficiency and correction accuracy of training the data decomposition-based meteorological forecast wind element data correction model.

[0093] In step S105, the historical measured wind speed of the sample point for the first historical time period is obtained.

[0094] The historical measured wind speed of the sample point during the first historical time period can include the historical measured wind speed at various height levels during the first historical time period of the sample point.

[0095] The height levels for wind speed can be 100 meters, 90 meters, 80 meters, 70 meters, 60 meters, and other height levels. The wind speed range is usually 0-50 m / s.

[0096] The first historical time period of the sample point can be obtained through wind measurement towers or lidar wind measurement equipment. It is real data in the historical process. For example, the historical measured wind speed of the sample point is the real wind speed in the historical process of the sample point.

[0097] The duration of the first historical time period can be the same as the duration of the second historical time period. For example, it can be 15 minutes.

[0098] In step S106, the CEEMDAN algorithm is used to decompose the historical measured wind speed of the sample point in the first historical time period to obtain multiple IMF wind speed components and a residual component.

[0099] In one example, assuming the historical measured wind speed of a sample point in the first historical time period is V, the CEEMDAN (Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Analysis) algorithm is used to decompose the historical measured wind speed V of the sample point in the first historical time period into several IMF (Intrinsic Mode Function) subsequences and residuals R, as shown below:

[0100]

[0101] That is, V = IMF1 + ... + IMF k +R.

[0102] In the aforementioned steps, steps S101 and S105 can be executed in parallel, or they can be executed sequentially. If they are executed sequentially, this application does not limit the execution order of steps S101 and S105.

[0103] Secondly, after step S101 is completed, step S102 can be executed. After step S102 is completed, step S103 can be executed. After step S103 is completed, step S104 can be executed.

[0104] In addition, step S106 can be executed after step S105 is completed.

[0105] After steps S104 and S106 are completed, step S107 can be executed.

[0106] The decomposition process of the CEEMDAN algorithm is illustrated with an example.

[0107] Let E i (·) represents the i-th intrinsic mode component obtained after EMD decomposition, and the i-th intrinsic mode component obtained after CEEMDAN decomposition is... v j For a Gaussian white noise signal that satisfies a standard normal distribution, j=1,2,…,N, where j is the number of times white noise is added, ε is the standard table of white noise, and y(t) is the signal to be decomposed.

[0108] The CEEMDAN decomposition steps are as follows:

[0109] Gaussian white noise is added to the signal V to be decomposed to obtain a new signal. , q=1,2.

[0110] EMD decomposition of the new signal yields the first-order intrinsic mode component C1:

[0111]

[0112] r j This represents the deviation obtained by subtracting each decomposed IMF from the signal to be decomposed after each IMF is decomposed.

[0113] The first intrinsic mode component of the decomposition is obtained by averaging the N generated modal components:

[0114]

[0115] Calculate the residual after removing the first modal component:

[0116]

[0117] Adding paired positive and negative Gaussian white noise to r1(t) yields a new signal. EMD decomposition of this new signal yields the first-order modal component D1, from which the second intrinsic modal component can be obtained.

[0118]

[0119] Calculate the residual after removing the second modal component:

[0120]

[0121] Repeat the above steps until the obtained residual signal is a monotonic function and cannot be further decomposed, at which point the algorithm ends. At this point, the number of intrinsic mode components obtained is K, and the original signal y(t) is decomposed into:

[0122]

[0123] C k (t) represents the IMF k r k (t) is represented by R.

[0124] In step S107, for any one of the multiple IMF wind speed components and a residual component, a wind element data correction sub-model corresponding to that component is trained using wind element sub-data related to wind speed as sample data and the component as labeled data.

[0125] The same applies to each of the multiple IMF wind speed components and each other component in a residual component, thus obtaining the wind element data correction sub-model corresponding to each component.

[0126] It should be noted that, in one embodiment, there can be multiple first historical time periods. No two first historical time periods overlap, and the multiple first historical time periods can be sequentially adjacent time periods. Each first historical time period is later than any second historical time period.

[0127] Thus, the first wind element data of the meteorological forecast for the first historical time period of the sample point can include: the first wind element data of the meteorological forecast for each of the multiple first historical time periods of the sample point.

[0128] Accordingly, the second wind element data of the meteorological forecast for the first historical time period of the sample point may include: the second wind element data of the meteorological forecast for each of the multiple first historical time periods of the sample point.

[0129] Accordingly, the wind element sub-data related to wind speed in the first historical time period of the sample point can include: wind element sub-data related to wind speed in each of the multiple first historical time periods of the sample point.

[0130] The historical measured wind speed of the first historical time period of the sample point can include the historical measured wind speed of each of the multiple first historical time periods of the sample point.

[0131] Thus, for any first historical time period, we can have wind element sub-data related to wind speed for that first historical time period at the sample point, as well as the historical measured wind speed for that first historical time period at the sample point.

[0132] Thus, training data corresponds to this first historical time period.

[0133] The sample data in the training data corresponding to the first historical time period is: wind element sub-data related to wind speed at the sample points during the first historical time period.

[0134] The labeled data in the training data corresponding to the first historical time period is: the historical measured wind speed of the sample points during the first historical time period.

[0135] The same applies to every other first historical period.

[0136] In this way, there can be training data corresponding to multiple first historical time periods. One part of the training data corresponding to each first historical time period is used to train the weather forecast wind element data correction model based on data decomposition, and another part of the training data corresponding to each first historical time period is used to test the trained weather forecast wind element data correction model based on data decomposition.

[0137] In the training data corresponding to the multiple first historical time periods, the amount of training data used for testing the weather forecast wind element data correction model based on data decomposition accounts for at least 10% of the amount of training data corresponding to the multiple first historical time periods.

[0138] In one example, following the example of step S106, using wind element sub-data related to wind speed as sample data and IMF1 as labeled data, train wind element data correction sub-model 1 corresponding to IMF1; using wind element sub-data related to wind speed as sample data and IMF2 as labeled data, train wind element data correction sub-model 2 corresponding to IMF2... using wind element sub-data related to wind speed as sample data and IMF1 as labeled data... k To train the IMF using labeled data k The corresponding wind element data correction sub-model k, and the wind element data correction sub-model R corresponding to R, which is trained using wind element data related to wind speed as sample data and R as labeled data.

[0139] In step S108, at least the wind element data correction sub-model corresponding to each component is used to obtain the meteorological forecast wind element data correction model based on data decomposition.

[0140] In one embodiment, a fully connected layer can be obtained, which is used to overlay and reconstruct the data output from the wind element data correction sub-model corresponding to each component. Overlay reconstruction includes addition, or calculation of average values, etc.

[0141] The output of the wind element data correction sub-model corresponding to each component is connected to the input of the fully connected layer.

[0142] The output of the weather forecast wind element data correction model based on data decomposition includes the output of a fully connected layer.

[0143] For example, the network structure of a weather forecast wind element data correction model based on data decomposition can be as follows: Figure 3 As shown.

[0144] Furthermore, the meteorological forecast wind element data correction model based on data decomposition also includes: unsupervised learning networks.

[0145] The input of the meteorological forecast wind element data correction model based on data decomposition includes the input of an unsupervised learning network.

[0146] The output of the unsupervised learning network is connected to the input of the wind element data correction sub-model corresponding to each component.

[0147] Unsupervised learning networks are used to extract features from wind element sub-data related to wind speed.

[0148] Each wind element data correction sub-model is used to process the wind element sub-data related to wind speed, and obtain the corrected wind speed at the sample point in the first historical time period.

[0149] Among them, the wind element data correction sub-model can include: GBDT (Gradient Boosting Decision Tree), ADABOOST (Adaptive Boosting, an iterative algorithm), XGBOOST (eXtremeGradient Boosting), LIGHTGBM (Light Gradient Boosting Machine, a machine learning algorithm), etc.

[0150] GBDT is an ensemble algorithm based on decision trees. Its core idea is to fit the residual of the previous weak learner by the negative gradient of the loss function. Through the accumulation of multiple rounds of learning, the output gradually approaches the true value, accelerating the convergence to the global or local optimum.

[0151] exist Figure 3 Based on what is shown, see Figure 4 An unsupervised learning network consists of multiple unsupervised learning sub-models connected in sequence. Figure 4 The diagram shows j unsupervised learning sub-models connected in series. This application does not limit the number of unsupervised learning sub-models connected in series. The number of j is not limited in this application and can be determined according to the actual situation; for example, it can be 2, 3, 4, or 5.

[0152] Unsupervised learning sub-models can include: RBM (Restricted Boltzmann Machines) or AE (Auto Encoders), etc.

[0153] Here, the j sequentially connected unsupervised learning sub-models can be understood as follows: the output of the adjacent preceding unsupervised learning sub-model is connected to the input of the adjacent following unsupervised learning sub-model. The j sequentially connected unsupervised learning sub-models are used to perform unsupervised learning on the input data to transform the features of the input data and extract more abstract and beneficial features from the input data.

[0154] The inventors found that wind speed in a sea or land area can vary over time, sometimes irregularly and sometimes abruptly. If the actual wind speed is used directly as labeled data to train the wind speed correction model, the sudden changes in wind speed will prevent the extraction of the actual wind speed pattern, resulting in large gradient changes. Consequently, the wind speed correction model may fail to converge or fail to converge to the optimal solution, leading to low accuracy in wind speed correction.

[0155] In view of this, the inventors devised a method to search for the pattern of wind speed changes over time, thereby reducing gradient changes so that the wind speed correction model can converge, or converge to the optimal solution, thereby improving the accuracy of the wind speed correction model.

[0156] For example, the inventors conceived of using the CEEMDAN algorithm to decompose the historical measured wind speed of a sample point in the first historical time period, obtaining multiple IMF wind speed components and a residual component. Each component possesses different frequency domain characteristics of the historical measured wind speed of the sample point in the first historical time period, and these different components can reflect certain patterns of the historical measured wind speed in the first historical time period of the sample point in different frequency domains. Thus, for each component, a corresponding wind element data correction sub-model can be trained. Because different components can reflect certain patterns of the historical measured wind speed in the first historical time period of the sample point in different frequency domains, the different components... The corresponding wind element data correction sub-model can learn some patterns of historical measured wind speed in different frequency domains for the first historical time period of the sample points. This allows for better, more comprehensive, and more accurate extraction of the characteristics of historical measured wind speed in different frequency domains for the first historical time period of the sample points when training wind element data correction sub-models for different components. This results in smaller gradient changes when training wind element data correction sub-models for different components, enabling the trained wind element data correction sub-models for different components to converge, or converge to the optimal solution. This can improve the accuracy of wind speed correction for the trained weather forecast wind element data correction model based on data decomposition.

[0157] Furthermore, since the CEEMDAN algorithm has the ability to eliminate mode mixing and residual noise, it can adaptively decompose the nonlinear data in the historical measured wind speed of the first historical time period of the sample points, thereby eliminating noise within the data and improving data availability.

[0158] Secondly, the CEEMDAN algorithm can diversify the data, meeting the prerequisite of more complete and diverse data required for using interpolation algorithms, and improving the accuracy of wind speed correction for the trained data decomposition-based meteorological forecast wind element data correction model.

[0159] In addition, the CEEMDAN algorithm can eliminate modal aliasing, achieve zero reconstruction error in the decomposed data, greatly reduce the consumption of computing resources, and improve the efficiency of data decomposition.

[0160] In addition, when training the weather forecast wind element data correction model based on data decomposition, the wind element sub-data related to wind speed determined in the second wind element data is used. The second wind element data is obtained by replacing the "wind speed in the first wind element data" with the "wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error". In the second wind element data, the wind speed difference can be regarded as the wind speed. This wind speed is closer to the historical measured wind speed, which reduces the gap between the wind speed in the sample data and the historical measured wind speed. This is beneficial to improving the accuracy of the wind speed correction of the trained weather forecast wind element data correction model based on data decomposition.

[0161] It should be noted that in the scenario of training a correction model, if the labeled data does not change abruptly, the labeled data can be used directly to train the correction model. This allows the patterns in the unchanging labeled data to be extracted, and the correction model trained in this way often has a high correction accuracy. Therefore, it is neither necessary nor necessary to use the CEEMDAN algorithm to decompose the labeled data, as this would consume a lot of computing resources, increase the training time of the correction model, and would not improve the correction accuracy. Technology will not develop in a direction that does not bring benefits but increases costs.

[0162] The scenario described in this application involves training a wind speed correction model, using historical real wind speeds as the labeled data. Historical real wind speeds can change abruptly, and this application precisely addresses the situation where the labeled data changes drastically. In this case, if the labeled data is directly used to train the wind speed correction model, the correction accuracy of the trained model is often very low because the patterns in the rapidly changing labeled data cannot be extracted. Therefore, the inventors conceived of using the CEEMDAN algorithm to decompose the labeled data and extract the patterns in different frequency domains, which significantly improves the correction accuracy of the wind speed correction model.

[0163] It is evident that in the scenario of training a corrected model, if the labeled data does not change drastically, there is no need or necessity to use the CEEMDAN algorithm.

[0164] Furthermore, after obtaining the weather forecast wind element data correction model based on data decomposition, the second average value of the wind speed difference between the wind speed in the weather forecast wind element data of the third historical time period of the sample points and the historical measured wind speed of the sample points in the third historical time period can be obtained. The third historical time period is later than the second historical time period.

[0165] If the error between the second average and the first average is greater than the preset benchmark wind speed error, it may be due to changes in the organization or technology providing the meteorological forecast wind element data. This could result in the given meteorological forecast wind element data being either closer to or further away from the actual situation than before, rendering the old preset benchmark wind speed error inapplicable. Therefore, to improve the accuracy of correcting the meteorological forecast wind element data, a data decomposition-based meteorological forecast wind element data correction model can be retrained to be suitable for the new preset benchmark wind speed error. For example, the preset benchmark wind speed error can be updated to the second average, and the step of obtaining the first wind element data of the meteorological forecast for the first historical time period of the sample points can be returned. That is, the second average is used as the preset benchmark wind speed error, and the data is retrained... Figure 1 The example shown trains a weather forecast wind element data correction model based on data decomposition.

[0166] Furthermore, after training the multi-channel meteorological forecast wind element data correction model, the multi-channel meteorological forecast wind element data correction model can be tested.

[0167] For example, see Figure 5 The specific steps include:

[0168] In step S201, the first wind element data of the weather forecast for the first historical time period of the test point is obtained. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0169] For details of this step, please refer to the description of step S101.

[0170] In step S202, the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error is calculated. The preset benchmark wind speed error includes the average of the wind speed differences between the wind speed in the meteorological forecast wind element data of multiple second historical time periods at the test point and the historical measured wind speeds of each second historical time period at the respective test point. The second historical time periods are earlier than the first historical time periods.

[0171] The first historical time period in this embodiment and Figure 1 The first historical time period shown can be different historical time periods. The second historical time period in this embodiment is the same as... Figure 1 The second historical time period shown can be a different historical time period.

[0172] For details of this step, please refer to the description of step S102.

[0173] In step S203, the calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data of the weather forecast for the first historical time period of the test point.

[0174] For details of this step, please refer to the description of step S103.

[0175] The first wind element data in this embodiment and Figure 1 The first wind element data shown can be different. The second wind element data in this embodiment is... Figure 1 The second wind element data shown can be different.

[0176] In step S204, based on the random forest algorithm, wind element sub-data related to wind speed is determined from the second wind element data.

[0177] For details of this step, please refer to the description of step S104.

[0178] In step S205, wind element sub-data related to wind speed is input into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the test point in the first historical time period is obtained based on the output wind speed correction component.

[0179] In step S206, the historical measured wind speed of the test point for the first historical time period is obtained.

[0180] For details of this step, please refer to the description of step S105.

[0181] In step S207, the wind speed error index is calculated based on the corrected wind speed of the first historical time period of the test point and the historical measured wind speed of the test point during the first historical time period.

[0182] For example, error indicators can be calculated using mean absolute error (MAE), root mean square error (RMSE), or coefficient of determination (R²).

[0183] Among them, the wind speed error index can be used to evaluate the correction effect of the weather forecast wind element data correction model based on data decomposition. That is, it can evaluate the accuracy of the weather forecast wind element data correction model based on data decomposition on wind speed correction. After the accuracy of wind speed correction meets the requirements, the weather forecast wind element data correction model based on data decomposition can be put into online application.

[0184] For example, a data correction model based on data decomposition can be used to correct the wind speed in a weather forecast.

[0185] For example, refer to Figure 6 The diagram illustrates a flowchart of a method for correcting wind element data in meteorological forecasts based on data decomposition, as described in this application. This method can be applied to electronic devices, which may include terminals or servers, etc. Specifically, the method may include the following steps:

[0186] In step S301, the first wind element data of the meteorological forecast for the first time period of the point to be corrected is obtained. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0187] For details of this step, please refer to the description of step S101.

[0188] In step S302, the wind speed difference between the wind speed in the first wind element data and the preset reference wind speed error is calculated. The preset reference wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of meteorological forecasts for the point to be corrected and the historical measured wind speed of each second time period for the point to be corrected. The second time period is earlier than the first time period.

[0189] The first time period can be a future time period after the current moment, and the second time period can be a historical time period before the current moment.

[0190] For details of this step, please refer to the description of step S102.

[0191] In step S303, the calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data of the weather forecast for the first time period of the point to be corrected.

[0192] For details of this step, please refer to the description of step S103.

[0193] In step S304, based on the random forest algorithm, wind element sub-data related to wind speed is determined from the second wind element data.

[0194] For details of this step, please refer to the description of step S104.

[0195] In step S305, wind element sub-data related to wind speed is input into the meteorological forecast wind element data correction model based on data decomposition. The meteorological forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on its own output wind speed correction component.

[0196] When correcting the wind speed for a future time period at a location to be corrected, the weather forecast wind element data correction model trained in this application can be used to process the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtain the corrected wind speed. For example, the weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtains its own output wind speed correction component. Then, based on the respective output wind speed correction component, the corrected wind speed for the future time period at the location to be corrected can be obtained. In this way, the accuracy of wind speed correction can be improved through this application.

[0197] Furthermore, after obtaining the measured wind speed for the third time period at the location to be corrected, the second average value of the wind speed difference between the wind speed in the meteorological forecast wind element data for the third time period at the location to be corrected and the measured wind speed for the third time period at the location to be corrected can be obtained. The third time period is later than the second time period.

[0198] If the error between the second average value and the first average value is greater than the preset benchmark wind speed error, a prompt message is output. The prompt message is used to prompt the data decomposition-based meteorological forecast wind element data correction model to be retrained with the second average value as the preset benchmark wind speed error.

[0199] If the error between the second average value and the first average value is greater than the preset benchmark wind speed error, it may be due to changes in the organization or technology providing the meteorological forecast wind element data. This could result in the given meteorological forecast wind element data being either closer to or further away from the actual situation than before, and the old preset benchmark wind speed error is no longer applicable. Therefore, in order to improve the accuracy of correcting the meteorological forecast wind element data, a prompt can be made to retrain the meteorological forecast wind element data correction model based on data decomposition that is suitable for the new preset benchmark wind speed error. For example, a prompt message can be output to prompt the retraining of the meteorological forecast wind element data correction model based on data decomposition with the second average value as the preset benchmark wind speed error.

[0200] Alternatively, if a data decomposition-based weather forecast wind element data correction model already exists, which is retrained using the second average value as the preset benchmark wind speed error, then a suggestion can be made to replace the data decomposition-based weather forecast wind element data correction model with the second average value as the preset benchmark wind speed error.

[0201] Another embodiment of this application provides a method for training a weather forecast wind element data correction model based on data decomposition, comprising: acquiring first wind element data of weather forecasts for a first historical time period of sample points, the first wind element data including multiple wind element sub-data, the wind element sub-data including at least wind direction, and the wind element sub-data including at least one of the following: wind speed, precipitation, temperature, humidity, air pressure, and irradiance; calculating the wind direction difference between the wind direction in the first wind element data and a preset benchmark wind direction error; the preset benchmark wind direction error including: a first average value of the wind direction differences between the wind direction in the wind element data of weather forecasts for multiple second historical time periods of sample points and the historical measured wind direction of each second historical time period of sample points; the second historical time periods are earlier than the first historical time periods; and replacing the wind direction in the first wind element data with the calculated wind direction difference. The process involves obtaining second wind element data for the first historical time period of the sample points; determining wind element sub-data related to wind direction from the second wind element data based on the random forest algorithm; acquiring the historical measured wind direction for the first historical time period of the sample points; decomposing the historical measured wind direction for the first historical time period of the sample points using the fully adaptive noise ensemble empirical mode decomposition (CEEMDAN) algorithm to obtain multiple intrinsic mode functions (IMF) wind direction components and one residual component; training a wind element data correction sub-model corresponding to the component using the wind element sub-data related to wind direction as sample data and the component as labeled data for any one of the multiple IMF wind direction components and the residual component; and obtaining a weather forecast wind element data correction model based on data decomposition, based at least on the wind element data correction sub-models corresponding to each component.

[0202] For a detailed explanation of this embodiment, please refer to the description of the foregoing embodiments, which will not be elaborated here.

[0203] The correction model in this embodiment is used for wind direction correction. For example, in another embodiment, a weather forecast wind element data correction method based on data decomposition is also provided, including: acquiring first wind element data of the weather forecast for a first time period of the point to be corrected, the first wind element data including multiple wind element sub-data, the wind element sub-data including at least wind direction, and the wind element sub-data including at least one of the following: wind speed, precipitation, temperature, humidity, air pressure, and irradiance; calculating the wind direction difference between the wind direction in the first wind element data and a preset benchmark wind direction error; the preset benchmark wind direction error includes: a first average value of the wind direction difference between the wind direction in the wind element data of multiple second time periods of the weather forecast for the point to be corrected and the historical measured wind direction of each second time period of the point to be corrected; the second time period is earlier than the first time period; using The calculated wind direction difference replaces the wind direction in the first wind element data to obtain the second wind element data for the first time period of the weather forecast of the point to be corrected. Based on the random forest algorithm, wind element sub-data related to wind direction is determined in the second wind element data. The wind element sub-data related to wind direction is input into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind direction to obtain its own output wind direction correction component. Then, the corrected wind direction of the point to be corrected in the first time period is obtained based on its own output wind direction correction component. The weather forecast wind element data correction model based on data decomposition is trained by the method mentioned above.

[0204] For a detailed explanation of this embodiment, please refer to the description of the foregoing embodiments, which will not be elaborated here.

[0205] The inventors found that when the wind direction changes in a region of the sea or land, the changes in wind direction over time are sometimes irregular and sometimes sudden. If the actual wind direction is used directly as labeled data to train the model for correcting the wind direction, the sudden changes in wind direction will make it impossible to extract the actual wind direction pattern, resulting in large gradient changes. Consequently, the model for correcting the wind direction may not converge, or may not converge to the optimal solution. This leads to low accuracy in correcting wind direction.

[0206] In view of this, the inventors devised a method to search for the pattern of wind direction changes over time, thereby reducing gradient changes so that the wind direction correction model can converge, or converge to the optimal solution, thereby improving the accuracy of the wind direction correction model in correcting wind direction.

[0207] For example, the inventors conceived of using the CEEMDAN algorithm to decompose the historical measured wind direction of a sample point in the first historical time period, obtaining multiple IMF wind direction components and a residual component. Each component possesses different frequency domain characteristics of the historical measured wind direction of the sample point in the first historical time period. These different components can reflect certain patterns of the historical measured wind direction in the first historical time period of the sample point across different frequency domains. Thus, for each component, a corresponding wind element data correction sub-model can be trained. Because different components can reflect certain patterns of the historical measured wind direction in the first historical time period of the sample point across different frequency domains, these different components... The corresponding wind element data correction sub-model can learn some patterns of the historical measured wind direction of the sample points in different frequency domains during the first historical time period. This allows for better, more comprehensive, and more accurate extraction of the characteristics of the historical measured wind direction of the sample points in different frequency domains during the first historical time period when training wind element data correction sub-models corresponding to different components. This results in smaller gradient changes during the training of wind element data correction sub-models corresponding to different components, enabling the trained wind element data correction sub-models corresponding to different components to converge, or to converge to the optimal solution. This can improve the accuracy of wind direction correction by the trained data decomposition-based meteorological forecast wind element data correction model.

[0208] Furthermore, since the CEEMDAN algorithm has the ability to eliminate mode mixing and residual noise, it can adaptively decompose the nonlinear data in the historical measured wind direction decomposition of the first historical time period of the sample points, thereby eliminating noise within the data and improving data availability.

[0209] Secondly, the CEEMDAN algorithm can diversify the data, meeting the prerequisite of more complete and diverse data required for using interpolation algorithms, and improving the accuracy of wind direction correction for the trained weather forecast wind element data correction model based on data decomposition.

[0210] In addition, the CEEMDAN algorithm can eliminate modal aliasing, achieve zero reconstruction error in the decomposed data, greatly reduce the consumption of computing resources, and improve the efficiency of data decomposition.

[0211] In addition, when training the weather forecast wind element data correction model based on data decomposition, the wind element sub-data related to wind direction determined in the second wind element data is used. The second wind element data is obtained by replacing the "wind direction in the first wind element data" with the "wind direction difference between the wind direction in the first wind element data and the preset benchmark wind direction error". In the second wind element data, the wind direction error can be regarded as the wind direction. This wind direction is closer to the historical measured wind direction, which reduces the gap between the wind direction in the sample data and the historical measured wind direction. This is beneficial to improving the accuracy of the wind direction correction of the weather forecast wind element data correction model based on data decomposition.

[0212] Therefore, when correcting the wind direction for a future time period at a location to be corrected, the weather forecast wind element data correction model trained in this application can be used to process the wind element sub-data related to wind direction in the weather forecast wind element data for the future time period at the location to be corrected, and obtain the corrected wind direction. For example, the weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind direction in the weather forecast wind element data for the future time period at the location to be corrected, and obtains its own output wind direction correction component. Then, based on the respective output wind direction correction component, the corrected wind direction for the future time period at the location to be corrected can be obtained. In this way, the accuracy of wind direction correction can be improved through this application.

[0213] It should be noted that in the scenario of training a correction model, if the labeled data does not change abruptly, the labeled data can be used directly to train the correction model. This allows the patterns in the unchanging labeled data to be extracted, and the correction accuracy of the trained correction model is often high. Therefore, it is neither necessary nor necessary to use the CEEMDAN algorithm to decompose the labeled data. This would not only consume a lot of computing resources and increase the time of training the correction model, but would also not improve the correction accuracy. Technology will not develop in a direction that does not bring benefits but increases costs.

[0214] The scenario described in this application involves training a wind direction correction model, using historical real wind directions as the labeled data. Historical real wind directions can change abruptly, and this application precisely addresses the situation where the labeled data changes abruptly. In this case, if the labeled data is directly used to train the wind direction correction model, the correction accuracy of the trained model is often very low because the patterns in the abruptly changing labeled data cannot be extracted. Therefore, the inventors came up with the idea of ​​using the CEEMDAN algorithm to decompose the labeled data in order to extract the patterns in the labeled data in different frequency domains. This will significantly improve the correction accuracy of the wind direction correction model.

[0215] It is evident that in the scenario of training a corrected model, if the labeled data does not change drastically, there is no need or necessity to use the CEEMDAN algorithm.

[0216] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions involved are not necessarily required by this application.

[0217] Reference Figure 7 The diagram shows a structural block diagram of an apparatus for training a weather forecast wind element data correction model based on data decomposition, the apparatus comprising:

[0218] The first acquisition module 11 is used to acquire the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine.

[0219] The first calculation module 12 is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period.

[0220] The first replacement module 13 is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the meteorological forecast for the first historical time period of the sample point.

[0221] The first determining module 14 is used to determine wind element sub-data related to wind speed in the second wind element data based on the random forest algorithm;

[0222] The second acquisition module 15 is used to acquire the historical measured wind speed of the sample point in the first historical time period.

[0223] Decomposition module 16 is used to decompose the historical measured wind speed of the sample point in the first historical time period using the fully adaptive noise set empirical mode decomposition CEEMDAN algorithm, to obtain multiple intrinsic mode function (IMF) wind speed components and a residual component.

[0224] Training module 17 is used to train a wind element data correction sub-model corresponding to any one of multiple IMF wind speed components and a residual component, using wind element sub-data related to wind speed as sample data and the component as labeled data.

[0225] The third acquisition module 18 is used to obtain a weather forecast wind element data correction model based on data decomposition, at least according to the wind element data correction sub-model corresponding to each component.

[0226] In one optional implementation, obtaining the weather forecast wind element data correction model based on data decomposition, according to the wind element data correction sub-model corresponding to each component, includes:

[0227] Obtain the fully connected layer, which is used to overlay and reconstruct the data output from the wind element data correction sub-model corresponding to each component;

[0228] The outputs of the wind element data correction sub-models corresponding to each component are connected to the inputs of the fully connected layer, and the outputs of the weather forecast wind element data correction model based on data decomposition include the outputs of the fully connected layer.

[0229] In an alternative implementation, the apparatus further includes:

[0230] The fourth acquisition module is used to obtain, after obtaining the meteorological forecast wind element data correction model based on data decomposition, the second average value of the wind speed difference between the wind speed in the meteorological forecast wind element data of the third historical time period of the sample point and the historical measured wind speed of the sample point in the third historical time period; the third historical time period is later than the second historical time period.

[0231] The update module is used to update the preset reference wind speed error to the second average value when the error between the second average value and the first average value is greater than the preset reference wind speed error, and then return to the step of obtaining the first wind element data of the meteorological forecast for the first historical time period of the sample points.

[0232] The inventors found that wind speed in a sea or land area can vary over time, sometimes irregularly and sometimes abruptly. If the actual wind speed is used directly as labeled data to train the wind speed correction model, the sudden changes in wind speed will prevent the extraction of the actual wind speed pattern, resulting in large gradient changes. Consequently, the wind speed correction model may fail to converge or fail to converge to the optimal solution, leading to low accuracy in wind speed correction.

[0233] In view of this, the inventors devised a method to search for the pattern of wind speed changes over time, thereby reducing gradient changes so that the wind speed correction model can converge, or converge to the optimal solution, thereby improving the accuracy of the wind speed correction model.

[0234] For example, the inventors conceived of using the CEEMDAN algorithm to decompose the historical measured wind speed of a sample point in the first historical time period, obtaining multiple IMF wind speed components and a residual component. Each component possesses different frequency domain characteristics of the historical measured wind speed of the sample point in the first historical time period, and these different components can reflect certain patterns of the historical measured wind speed in the first historical time period of the sample point in different frequency domains. Thus, for each component, a corresponding wind element data correction sub-model can be trained. Because different components can reflect certain patterns of the historical measured wind speed in the first historical time period of the sample point in different frequency domains, the different components... The corresponding wind element data correction sub-model can learn some patterns of historical measured wind speed in different frequency domains for the first historical time period of the sample points. This allows for better, more comprehensive, and more accurate extraction of the characteristics of historical measured wind speed in different frequency domains for the first historical time period of the sample points when training wind element data correction sub-models for different components. This results in smaller gradient changes when training wind element data correction sub-models for different components, enabling the trained wind element data correction sub-models for different components to converge, or converge to the optimal solution. This can improve the accuracy of wind speed correction for the trained weather forecast wind element data correction model based on data decomposition.

[0235] Furthermore, since the CEEMDAN algorithm has the ability to eliminate mode mixing and residual noise, it can adaptively decompose the nonlinear data in the historical measured wind speed of the first historical time period of the sample points, thereby eliminating noise within the data and improving data availability.

[0236] Secondly, the CEEMDAN algorithm can diversify the data, meeting the prerequisite of more complete and diverse data required for using interpolation algorithms, and improving the accuracy of wind speed correction for the trained data decomposition-based meteorological forecast wind element data correction model.

[0237] In addition, the CEEMDAN algorithm can eliminate modal aliasing, achieve zero reconstruction error in the decomposed data, greatly reduce the consumption of computing resources, and improve the efficiency of data decomposition.

[0238] In addition, when training the weather forecast wind element data correction model based on data decomposition, the wind element sub-data related to wind speed determined in the second wind element data is used. The second wind element data is obtained by replacing the "wind speed in the first wind element data" with the "wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error". In the second wind element data, the wind speed difference can be regarded as the wind speed. This wind speed is closer to the historical measured wind speed, which reduces the gap between the wind speed in the sample data and the historical measured wind speed. This is beneficial to improving the accuracy of the wind speed correction of the trained weather forecast wind element data correction model based on data decomposition.

[0239] Reference Figure 8 The diagram shows a structural block diagram of a meteorological forecast wind element data correction device based on data decomposition according to this application. The device includes:

[0240] The fifth acquisition module 21 is used to acquire the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed, and the wind element sub-data also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and light intensity.

[0241] The second calculation module 22 is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of the weather forecast of the point to be corrected and the historical measured wind speed of each second time period of the point to be corrected; the second time period is earlier than the first time period.

[0242] The second replacement module 23 is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the weather forecast for the first time period of the point to be corrected.

[0243] The second determining module 24 is used to determine wind element sub-data related to wind speed in the second wind element data based on the random forest algorithm;

[0244] Processing module 25 is used to input wind element sub-data related to wind speed into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on the output wind speed correction component.

[0245] In an alternative implementation, the apparatus further includes:

[0246] The sixth acquisition module is used to acquire the second average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast for the third time period of the point to be corrected and the measured wind speed of the point to be corrected for the third time period; the third time period is later than the second time period.

[0247] The output module is used to output a prompt message when the error between the second average value and the first average value is greater than the preset benchmark wind speed error. The prompt message is used to prompt the data decomposition-based meteorological forecast wind element data correction model to be retrained with the second average value as the preset benchmark wind speed error.

[0248] Among them, the meteorological forecast wind element data correction model based on data decomposition is achieved through, for example, Figure 1 It is trained using any of the methods described above.

[0249] Alternatively, the weather forecast wind element data correction model based on data decomposition is achieved through methods such as... Figure 7 The training was obtained using any of the aforementioned devices.

[0250] When correcting the wind speed for a future time period at a location to be corrected, the weather forecast wind element data correction model trained in this application can be used to process the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtain the corrected wind speed. For example, the weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed in the weather forecast wind element data for the future time period at the location to be corrected, and obtains its own output wind speed correction component. Then, based on the respective output wind speed correction component, the corrected wind speed for the future time period at the location to be corrected can be obtained. In this way, the accuracy of wind speed correction can be improved through this application.

[0251] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0252] Optionally, this application also provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the various processes of the above method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0253] This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0254] The figure is a block diagram of an electronic device 800 as shown in the nine applications. For example, the electronic device 800 can be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0255] Reference Figure 9 The electronic device 800 may include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input / output (I / O) interface 812, a sensor component 814, and a communication component 816.

[0256] Processing component 802 typically controls the overall operation of electronic device 800, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 802 may include one or more modules to facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802.

[0257] Memory 804 is configured to store various types of data to support the operation of device 800. Examples of this data include instructions for any application or method operating on electronic device 800, contact data, phonebook data, messages, images, videos, etc. Memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0258] Power supply component 806 provides power to various components of electronic device 800. Power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 800.

[0259] Multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 808 includes a front-facing camera and / or a rear-facing camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0260] Audio component 810 is configured to output and / or input audio signals. For example, audio component 810 includes a microphone (MIC) configured to receive external audio signals when electronic device 800 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 804 or transmitted via communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

[0261] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0262] Sensor assembly 814 includes one or more sensors for providing state assessments of various aspects of electronic device 800. For example, sensor assembly 814 may detect the on / off state of device 800, the relative positioning of components such as the display and keypad of electronic device 800, changes in position of electronic device 800 or a component of electronic device 800, the presence or absence of user contact with electronic device 800, orientation or acceleration / deceleration of electronic device 800, and temperature changes of electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 814 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0263] Communication component 816 is configured to facilitate wired or wireless communication between electronic device 800 and other devices. Electronic device 800 can access wireless networks based on communication standards, such as WiFi, carrier networks (such as 2G, 3G, 4G, or 5G), or combinations thereof. In one exemplary embodiment, communication component 816 receives broadcast signals or broadcast operation information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 816 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0264] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0265] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, which can be executed by a processor 820 of an electronic device 800 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0266] Figure 10This is a block diagram of an electronic device 1900 shown in this application. For example, the electronic device 1900 can be provided as a server.

[0267] Reference Figure 10 The electronic device 1900 includes a processing component 1922, which further includes one or more processors, and memory resources represented by memory 1932 for storing instructions, such as application programs, that can be executed by the processing component 1922. The application programs stored in memory 1932 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 1922 is configured to execute instructions to perform the methods described above.

[0268] Electronic device 1900 may also include a power supply component 1926 configured to perform power management of electronic device 1900, a wired or wireless network interface 1950 configured to connect electronic device 1900 to a network, and an input / output (I / O) interface 1958. Electronic device 1900 can operate on an operating system stored in memory 1932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0269] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0270] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0271] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0272] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0273] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0274] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0275] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0276] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0277] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0278] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for training a weather forecast wind element data correction model based on data decomposition, characterized in that, The method includes: Obtain the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine. Calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period. The calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data for the weather forecast of the first historical time period of the sample points. Based on the random forest algorithm, wind element sub-data related to wind speed are determined in the second wind element data; Obtain the historical measured wind speed of the sample points for the first historical time period; Using the fully adaptive noise ensemble empirical mode decomposition (CEEMDAN) algorithm, the historical measured wind speed of the sample points in the first historical time period is decomposed to obtain multiple intrinsic mode functions (IMF) wind speed components and a residual component. For any one of the multiple IMF wind speed components and a residual component, a wind element data correction sub-model corresponding to the component is trained using wind element sub-data related to wind speed as sample data and the component as labeled data. At least based on the wind element data corresponding to each component, a wind element data correction model for meteorological forecasts can be obtained.

2. The method according to claim 1, characterized in that, The step of obtaining a weather forecast wind element data correction model based on data decomposition, by modifying the sub-models according to the wind element data corresponding to each component, includes: Obtain the fully connected layer, which is used to overlay and reconstruct the data output from the wind element data correction sub-model corresponding to each component; Connect the output of the wind element data correction sub-model corresponding to each component to the input of the fully connected layer respectively. The output of the weather forecast wind element data correction model based on data decomposition includes the output of a fully connected layer.

3. The method according to claim 1, characterized in that, The method further includes: After obtaining the weather forecast wind element data correction model based on data decomposition, the second average value of the wind speed difference between the wind speed in the weather forecast wind element data of the third historical time period of the sample point and the historical measured wind speed of the sample point in the third historical time period is obtained; the third historical time period is later than the second historical time period. If the error between the second average value and the first average value is greater than the preset reference wind speed error, the preset reference wind speed error is updated to the second average value, and the process returns to the step of obtaining the first wind element data of the meteorological forecast for the first historical time period of the sample points.

4. A method for correcting meteorological forecast wind element data based on data decomposition, characterized in that, The method includes: Obtain the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure, and sunshine. Calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of the meteorological forecast for the point to be corrected and the historical measured wind speed of each second time period for the point to be corrected; the second time period is earlier than the first time period. The calculated wind speed difference is used to replace the wind speed in the first wind element data to obtain the second wind element data for the first time period of the weather forecast of the point to be corrected. Based on the random forest algorithm, wind element sub-data related to wind speed are determined in the second wind element data; Wind element sub-data related to wind speed is input into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on the output wind speed correction component. The meteorological forecast wind element data correction model based on data decomposition is trained by the method described in any one of claims 1 to 3.

5. The method according to claim 4, characterized in that, The method further includes: Obtain the second average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast for the third time period of the point to be corrected and the measured wind speed of the point to be corrected for the third time period; the third time period is later than the second time period. If the error between the second average value and the first average value is greater than the preset benchmark wind speed error, a prompt message is output. The prompt message is used to prompt the data decomposition-based meteorological forecast wind element data correction model to be retrained with the second average value as the preset benchmark wind speed error.

6. An apparatus for training a weather forecast wind element data correction model based on data decomposition, characterized in that, The device includes: The first acquisition module is used to acquire the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed and also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and sunshine. The first calculation module is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error. The preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of the meteorological forecast of multiple second historical time periods of the sample point and the historical measured wind speed of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period. The first replacement module is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the meteorological forecast for the first historical time period of the sample point. The first determination module is used to determine wind element sub-data related to wind speed from the second wind element data based on the random forest algorithm; The second acquisition module is used to acquire the historical measured wind speed of the sample points in the first historical time period. The decomposition module is used to decompose the historical measured wind speed of the sample points in the first historical time period using the fully adaptive noise set empirical mode decomposition (CEEMDAN) algorithm, and obtain multiple intrinsic mode functions (IMF) wind speed components and a residual component. The training module is used to train a wind element data correction sub-model corresponding to any one of multiple IMF wind speed components and a residual component, using wind element sub-data related to wind speed as sample data and the component as labeled data. The third acquisition module is used to obtain a weather forecast wind element data correction model based on data decomposition, at least according to the wind element data corresponding to each component.

7. A meteorological forecast wind element data correction device based on data decomposition, characterized in that, The device includes: The fifth acquisition module is used to acquire the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind speed, and the wind element sub-data also includes at least one of the following: wind direction, precipitation, temperature, humidity, air pressure and light intensity. The second calculation module is used to calculate the wind speed difference between the wind speed in the first wind element data and the preset benchmark wind speed error; the preset benchmark wind speed error includes: the first average value of the wind speed difference between the wind speed in the wind element data of multiple second time periods of the weather forecast of the point to be corrected and the historical measured wind speed of each second time period of the point to be corrected; the second time period is earlier than the first time period. The second replacement module is used to replace the wind speed in the first wind element data with the calculated wind speed difference to obtain the second wind element data of the weather forecast for the first time period of the point to be corrected. The second determination module is used to determine wind element sub-data related to wind speed from the second wind element data based on the random forest algorithm; The processing module is used to input wind element sub-data related to wind speed into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind speed to obtain its own output wind speed correction component. Then, the corrected wind speed of the point to be corrected in the first time period is obtained based on the output wind speed correction component. The meteorological forecast wind element data correction model based on data decomposition is trained by the method described in any one of claims 1 to 3.

8. A method for training a weather forecast wind element data correction model based on data decomposition, characterized in that, The method includes: Obtain the first wind element data of the meteorological forecast for the first historical time period of the sample point. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind direction and also includes at least one of the following: wind speed, precipitation, temperature, humidity, air pressure and sunshine. Calculate the wind direction difference between the wind direction in the first wind element data and the preset benchmark wind direction error; the preset benchmark wind direction error includes: the first average value of the wind direction difference between the wind direction in the wind element data of multiple second historical time periods of the sample point and the historical measured wind direction of each second historical time period of the sample point; the second historical time period is earlier than the first historical time period. The calculated wind direction difference is used to replace the wind direction in the first wind element data to obtain the second wind element data for the meteorological forecast of the first historical time period of the sample points. Based on the random forest algorithm, wind element sub-data related to wind direction are determined in the second wind element data; Obtain the historical measured wind direction for the first historical time period at the sample location; Using the fully adaptive noise ensemble empirical mode decomposition (CEEMDAN) algorithm, the historical measured wind direction of the sample points in the first historical time period is decomposed to obtain multiple intrinsic mode functions (IMF) wind direction components and a residual component. For any one of the multiple IMF wind direction components and a residual component, a wind element data correction sub-model corresponding to the component is trained using wind element sub-data related to wind direction as sample data and the component as labeled data. At least based on the wind element data corresponding to each component, a wind element data correction model for meteorological forecasts can be obtained.

9. A method for correcting meteorological forecast wind element data based on data decomposition, characterized in that, The method includes: Obtain the first wind element data of the meteorological forecast for the first time period of the point to be corrected. The first wind element data includes multiple wind element sub-data. The wind element sub-data includes at least wind direction and also includes at least one of the following: wind speed, precipitation, temperature, humidity, air pressure, and sunshine. Calculate the wind direction difference between the wind direction in the first wind element data and the preset benchmark wind direction error; the preset benchmark wind direction error includes: the first average value of the wind direction difference between the wind direction in the wind element data of multiple second time periods of meteorological forecasts for the point to be corrected and the historical measured wind direction of each second time period for the point to be corrected; the second time period is earlier than the first time period; The calculated wind direction difference is used to replace the wind direction in the first wind element data to obtain the second wind element data for the first time period of the weather forecast of the point to be corrected. Based on the random forest algorithm, wind element sub-data related to wind direction are determined in the second wind element data; Wind element sub-data related to wind direction is input into the weather forecast wind element data correction model based on data decomposition. The weather forecast wind element data correction model based on data decomposition includes multiple wind element data correction sub-models. Each wind element data correction sub-model processes the wind element sub-data related to wind direction to obtain its own output wind direction correction component. Then, the corrected wind direction of the point to be corrected in the first time period is obtained based on the output wind direction correction component. The meteorological forecast wind element data correction model based on data decomposition is obtained by training the method described in claim 8.

10. An electronic device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the method as described in any one of claims 1 to 5 and 8 to 9.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 5 and 8 to 9.

Images