Multi-source satellite offshore wind resource map analysis method, device and equipment

Abstract

The present invention provides a multi-source satellite offshore regional wind resource map analysis method, device, and equipment, comprising: preprocessing multi-source original sea surface wind fields to determine a multi-source target sea surface wind field corresponding to a first grid; constructing an average sea surface wind field set based on wind station data, the multi-source target sea surface wind field corresponding to the first grid, and reanalysis data; training a machine learning model using the average sea surface wind field set and sea surface feature data, so that the trained machine learning model can predict the satellite average sea surface wind field of a second grid based on the reanalysis data corresponding to the second grid; and generating a wind resource map corresponding to the study sea area based on the satellite average sea surface wind fields corresponding to the first and second grids, respectively. The present invention can effectively ensure the integrity of regional data, consistency with actual observations, and data accuracy, and can also reduce the difficulty of collecting data for wind resource assessment and prediction.

Description

Technical Field

[0001] The present invention relates to the field of microwave remote sensing technology, and in particular to a method, device and equipment for analyzing wind resource spectra in a multi-source satellite offshore area. Background Art

[0002] With the acceleration of industrialization, the demand for renewable energy continues to increase. Wind energy, a widely distributed, pollution-free renewable energy source with high utilization potential, is often used in wind farm construction. However, due to the highly uneven distribution of wind resources, regional wind resource assessments are essential before wind farm construction to better plan the layout of wind turbines and build appropriately sized wind power systems, thereby improving wind resource utilization. Due to the randomness and uncontrollability of wind speed, accurately assessing a region's wind energy resources requires a large amount of long-term observational data to reduce the uncertainty caused by wind speed variations. However, collecting large amounts of long-term observational data is not only difficult but also time-consuming. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a multi-source satellite offshore regional wind resource map analysis method, device and equipment, which can effectively ensure the integrity of regional data, consistency with actual observations and data accuracy, and can also reduce the difficulty of collecting data for wind resource assessment and prediction.

[0004] In a first aspect, the present invention provides a multi-source satellite offshore wind resource map analysis method, comprising:

[0005] Preprocessing the multi-source original sea surface wind field in the study sea area to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area, where the first grid is the grid whose corresponding multi-source original sea surface wind field has passed multi-level wind field quality control;

[0006] Based on the wind station data corresponding to the study sea area, as well as the multi-source target sea surface wind field and reanalysis data corresponding to the first grid, a mean sea surface wind field set is constructed. The mean sea surface wind field set includes the satellite mean sea surface wind field and the reanalysis mean sea surface wind field corresponding to different sea surface heights of the first grid;

[0007] The machine learning model is trained using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area, so that the trained machine learning model can predict the satellite average sea surface wind field corresponding to the second grid at different sea surface heights based on the reanalysis data corresponding to the second grid in the study sea area, where the second grid is a grid that lacks a multi-source original sea surface wind field or whose corresponding multi-source original sea surface wind field fails to pass the multi-level wind field quality control or has a small number of observations;

[0008] Based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights, a wind resource map corresponding to the study sea area is generated.

[0009] In one embodiment, preprocessing the multi-source original sea surface wind field in the study sea area to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area includes:

[0010] Perform spatiotemporal matching of the offshore buoy data and multi-source original sea surface wind fields corresponding to the study area;

[0011] Using offshore buoy data, multi-level wind field quality control is performed on the multi-source original sea surface wind field to extract the multi-source original sea surface wind field that has passed the multi-level wind field quality control;

[0012] The wind speed correction coefficient is fitted based on the offshore buoy data, and the wind speed correction coefficient is used to perform wind speed quality correction on the multi-source original sea surface wind field that has passed the multi-level wind field quality control to obtain the multi-source intermediate sea surface wind field.

[0013] The multi-source intermediate sea surface wind field is interpolated into a grid of specified resolution to obtain the multi-source target sea surface wind field corresponding to the first grid in the study sea area.

[0014] In one embodiment, based on the wind station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid, a mean sea surface wind field set is constructed, including:

[0015] Based on the wind measurement station data corresponding to the study sea area, the wind shear index corresponding to the study sea area at different time periods and different sea level heights is fitted;

[0016] The multi-source target sea surface wind field corresponding to the first grid is interpolated using the wind shear index to obtain the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights;

[0017] Perform spatiotemporal matching on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and the reanalysis data corresponding to the first grid;

[0018] Based on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and their matched reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate the average sea surface wind field set.

[0019] In one embodiment, based on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and their matching reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate an average sea surface wind field set, including:

[0020] Randomly combining the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights to obtain a sea surface wind field combination corresponding to the first grid at different sea level heights, wherein the sea surface wind field combination includes at least two source target sea surface wind fields;

[0021] Based on the sea surface wind field combination and its matching reanalysis data, a Weibull model of wind speed corresponding to the first grid is established;

[0022] According to the wind speed probability distribution described by the Weibull model, the satellite average sea surface wind speed and the satellite average sea surface wind direction are calculated to obtain the satellite average sea surface wind field, and the reanalysis average sea surface wind speed and the reanalysis average sea surface wind direction are calculated to obtain the reanalysis average sea surface wind field.

[0023] In one embodiment, training a machine learning model using the average sea surface wind field set and sea surface characteristic data corresponding to the study sea area includes:

[0024] For any first grid, the satellite average sea surface wind field corresponding to the first grid is taken as the learning target, and the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the first grid and its neighboring first grids are used as model inputs to train the machine learning model.

[0025] In one embodiment, the satellite-averaged sea surface wind field corresponding to the second grid at different sea level heights is predicted based on the reanalysis data corresponding to the second grid in the study sea area using the trained machine learning model, including:

[0026] Find the second grid within the research sea area;

[0027] For any second grid, the reanalysis average sea surface wind field corresponding to the second grid and its neighboring second grids is determined based on the reanalysis data. The reanalysis average sea surface wind field and sea surface characteristic data corresponding to the second grid and its neighboring second grids are input into the trained machine learning model to obtain the satellite average sea surface wind field corresponding to the second grid at different sea level heights.

[0028] In one embodiment, generating a wind resource map corresponding to the study sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights includes:

[0029] Based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea level heights, the satellite average sea surface wind field corresponding to the study sea area is obtained;

[0030] Based on the satellite average sea surface wind field corresponding to the study sea area, one or more of the effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution are calculated to obtain the wind resource map corresponding to the study sea area.

[0031] In a second aspect, the present invention further provides a multi-source satellite offshore wind resource map analysis device, comprising:

[0032] A preprocessing module is used to preprocess the multi-source original sea surface wind field in the study sea area to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area, where the first grid is the grid whose corresponding multi-source original sea surface wind field has passed the multi-level wind field quality control;

[0033] The mean sea surface wind field construction module is used to construct a mean sea surface wind field set based on the wind measurement station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid. The mean sea surface wind field set includes the satellite mean sea surface wind field and the reanalysis mean sea surface wind field corresponding to different sea surface heights of the first grid;

[0034] a missing sea surface wind field prediction module, configured to train a machine learning model using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area, so as to predict the satellite average sea surface wind field corresponding to the second grid at different sea surface heights based on the reanalysis data corresponding to the second grid in the study sea area through the trained machine learning model, where the second grid is a grid for which the multi-source original sea surface wind field is missing or for which the multi-source original sea surface wind field corresponding to the second grid fails the multi-level wind field quality control;

[0035] The wind resource map generation module is used to generate a wind resource map corresponding to the research sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights.

[0036] In a third aspect, the present invention further provides an electronic device comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement any one of the methods provided in the first aspect.

[0037] In a fourth aspect, the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect.

[0038] An embodiment of the present invention provides a multi-source satellite offshore wind resource map analysis method, device and equipment. First, the multi-source original sea surface wind field of the study sea area is pre-processed to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area. The first grid is a grid whose corresponding multi-source original sea surface wind field has passed multi-level wind field quality control. Then, based on the wind measurement station data corresponding to the study sea area, the multi-source target sea surface wind field corresponding to the first grid and the reanalysis data, an average sea surface wind field set is constructed. The average sea surface wind field set includes the satellite average sea surface wind field corresponding to the first grid at different sea level heights and the re-analysis data. The average sea surface wind field is analyzed; the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area are then used to train the machine learning model, so that the trained machine learning model can be used to predict the satellite average sea surface wind field corresponding to the second grid at different sea surface heights based on the reanalysis data corresponding to the second grid in the study sea area. The second grid is a grid that lacks the multi-source original sea surface wind field or whose corresponding multi-source original sea surface wind field fails to pass the multi-level wind field quality control; finally, the wind resource map corresponding to the study sea area is generated based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights. The above method trains a machine learning model based on the target sea surface wind field, reanalysis data, and sea surface characteristic data observed by multi-source satellites at the first grid, and then uses the machine learning model to calculate the satellite average sea surface wind field in the second grid with data in the area without satellite observation or missing data, thereby generating a wind resource map of the entire study sea area. On the one hand, the relatively easy-to-obtain reanalysis data and sea surface characteristic data are used to ensure the integrity of regional data, consistency with actual observations, and data accuracy through the machine learning model. On the other hand, it reduces the difficulty of collecting data during wind resource assessment and prediction, reduces the time cost of data collection, and provides a reference basis for reducing the construction risks of wind resource-related projects and improving the benefits of wind energy resource construction.

[0039] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 A schematic flow chart of a multi-source satellite offshore wind resource map analysis method provided in an embodiment of the present invention;

[0043] Figure 2 A technical framework diagram of a multi-source satellite offshore wind resource map analysis method provided by an embodiment of the present invention;

[0044] Figure 3 A schematic structural diagram of a multi-source satellite offshore wind resource map analysis device provided by an embodiment of the present invention;

[0045] Figure 4 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0047] At present, related technologies have the problems of high difficulty in operation and long time consumption. Based on this, the present invention provides a multi-source satellite offshore regional wind resource map analysis method, device and equipment, which can effectively ensure the integrity of regional data, consistency with actual observations and data accuracy, and can also reduce the difficulty of collecting data for wind resource assessment and prediction. At the same time, it has the advantages of simple algorithm, easy implementation, wide coverage and strong practicality.

[0048] To facilitate understanding of this embodiment, a multi-source satellite offshore wind resource map analysis method disclosed in an embodiment of the present invention is first described in detail. Figure 1 The flowchart of a multi-source satellite offshore wind resource map analysis method is shown, and the method mainly includes the following steps S102 to S108:

[0049] Step S102 : pre-processing the multi-source original sea surface wind field in the study sea area to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area.

[0050] The first grid is a grid whose corresponding multi-source original sea surface wind field has undergone multi-level wind field quality control. Preprocessing includes spatiotemporal matching, multi-level wind field quality control, wind field quality correction, and interpolation. In one example, the offshore buoy data corresponding to the study area can be spatiotemporally matched with the multi-source original sea surface wind field. Based on this, the offshore buoy data is used to perform multi-level wind field quality control and wind field quality correction on the multi-source original sea surface wind field data to obtain a multi-source intermediate sea surface wind field. Finally, the intermediate sea surface wind fields from different sources are interpolated to the same resolution grid to obtain the multi-source target sea surface wind field corresponding to the first grid.

[0051] Step S104 : constructing an average sea surface wind field set based on the wind station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid.

[0052] The mean sea surface wind field set includes the satellite-mean sea surface wind field and the reanalysis-mean sea surface wind field corresponding to the first grid at different sea surface heights. The satellite-mean sea surface wind field includes the satellite-mean sea surface wind speed and the satellite-mean sea surface wind direction, while the reanalysis-mean sea surface wind field includes the reanalysis-mean sea surface wind speed and the reanalysis-mean sea surface wind direction. In one example, wind station data can be used to fit the wind shear index corresponding to different time periods and sea surface heights in the study area. This wind shear index can be used to interpolate the multi-source target sea surface wind field to obtain the multi-source target sea surface wind field corresponding to different sea surface heights for the first grid. Based on this, the reanalysis data are temporally and spatially matched, and a wind speed Weibull model describing the wind speed probability distribution is constructed using the matched data set. Based on the wind speed probability distribution described by the wind speed Weibull model, the satellite-mean sea surface wind field and the reanalysis-mean sea surface wind field can be calculated, respectively.

[0053] Step S106, using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area to train the machine learning model, so that the trained machine learning model can predict the satellite average sea surface wind field corresponding to the second grid at different sea level heights based on the reanalysis data corresponding to the second grid in the study sea area.

[0054] The sea surface characteristic data includes sea surface height, atmospheric pressure, sea surface temperature, sea surface humidity, etc. The second grid is a grid that lacks a multi-source original sea surface wind field or whose corresponding multi-source original sea surface wind field fails multi-level wind field quality control. In one example, a machine learning model (such as XGBoost) is trained using the satellite-averaged sea surface wind field corresponding to the first grid as the learning target, and the reanalysis-averaged sea surface wind field and sea surface characteristic data corresponding to the first grid and its neighboring first grids as model inputs. For any second grid, the reanalysis-averaged sea surface wind field and sea surface characteristic data corresponding to the second grid and its neighboring second grids are input into the trained machine learning model (such as XGBoost) to obtain the satellite-averaged sea surface wind field corresponding to the second grid.

[0055] Step S108: Generate a wind resource map corresponding to the research sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights.

[0056] The wind resource map can describe the distribution of wind resources, such as effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution, for each grid cell within the study area. In one example, the satellite-averaged sea surface wind field corresponding to the first and second grid cells at different sea level heights can be used to obtain the satellite-averaged sea surface wind field for the entire study area. Wind resources, such as effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution, can then be calculated based on this satellite-averaged sea surface wind field to create a wind resource map.

[0057] The multi-source satellite offshore regional wind resource map analysis method provided by an embodiment of the present invention trains a machine learning model based on the target sea surface wind field, reanalysis data, and sea surface feature data observed by multi-source satellites at a first grid, and then uses the machine learning model to calculate the satellite average sea surface wind field in a second grid with data in a non-satellite observation area or missing data, thereby generating a wind resource map for the entire study sea area. On the one hand, the relatively easy-to-obtain reanalysis data and sea surface feature data are used to ensure the integrity of regional data, consistency with actual observations, and data accuracy through a machine learning model. On the other hand, the difficulty of collecting data during wind resource assessment and prediction is reduced, and the time cost of data collection is reduced, thereby providing a reference basis for reducing the construction risks of wind resource-related projects and improving the benefits of wind energy resource construction.

[0058] For ease of understanding, the present invention provides a specific implementation of a multi-source satellite offshore wind resource map analysis method, see Figure 2The technical framework diagram of a multi-source satellite offshore wind resource atlas analysis method shown in the figure is as follows: long-term series of SAR satellite data (i.e., multi-source original sea surface wind field), ERA5 reanalysis data, wind tower data and other sea surface feature data are collected. First, the SAR satellite data are preprocessed, mainly including multi-level quality control and wind speed correction; then, the wind shear index is fitted according to the wind tower data, and the SAR satellite data and ERA5 reanalysis data are temporally and spatially matched. Based on the matching data set, a wind speed Weibull distribution is established, and the observed average wind speed and average wind direction are calculated; based on the observed average wind speed and average wind direction, as well as the matching data set and sea surface feature data, an XGBoost wind field prediction model is established, and the model is used to predict the average wind field in the data-missing area to obtain the average wind field of the entire study sea area; finally, based on the average wind field of the entire study sea area, the wind energy resources are evaluated from the perspectives of effective wind speed, wind speed Weibull distribution analysis, effective wind energy density distribution, wind energy variation coefficient, wind direction frequency statistics, etc.

[0059] The specific implementation process is as follows:

[0060] (1) Data collection: Collect long-term SAR satellite data for the study area, including original sea surface wind field observations from multiple-source satellites such as Sentinel-1A, Sentinel-1B, GF3, and 1mCSAR; collect long-term ERA5 reanalysis data for the study area; collect offshore buoy data and wind tower data for the study area; collect sea surface characteristic data such as sea surface height, atmospheric pressure, sea surface temperature, and sea surface humidity for the study area.

[0061] (2) Preprocessing of the original sea surface wind field, including the following (2.1) to (2.4):

[0062] (2.1) Perform spatiotemporal matching of the offshore buoy data corresponding to the study area with the multi-source raw sea surface wind field. In one example, assuming the offshore buoy data represent true wind speed and direction, perform spatiotemporal matching of the multi-source raw sea surface wind field with the offshore buoy data, using a 30-minute temporal window and a 50-km spatial window.

[0063] (2.2) Using offshore buoy data, multi-level wind field quality control is performed on the multi-source original sea surface wind field to extract the multi-source original sea surface wind field that has passed the multi-level wind field quality control.

[0064] Among them, multi-level quality control includes data quality screening and anomaly detection quality control through the quality labels of various products. Not all wind farm data need to be involved in the calculation. When the deviation of a single data exceeds 3 times the mean absolute error of the overall data, it can be considered that this data is an anomaly and needs to be eliminated.

[0065] In one example, the wind direction error can be calculated as follows:

[0066] ;

[0067] Where, The first Wind direction values ​​of matching points, Indicates the The buoy wind direction value of the matching point, is the number of matching points.

[0068] In one example, the wind speed error can be calculated as follows:

[0069] ;

[0070] Where, The first Wind speed values ​​at matching points, Indicates the The wind speed value of the buoy at the assigned point, is the number of matching points.

[0071] Based on the above wind direction error and wind speed error, the wind direction mean absolute error and wind speed mean absolute error can be determined respectively for multi-level wind farm quality control.

[0072] (2.3) Based on the wind speed correction coefficient fitted by the offshore buoy data, the wind speed correction coefficient is used to perform wind speed quality correction on the multi-source original sea surface wind field that has passed the multi-level wind field quality control to obtain the multi-source intermediate sea surface wind field. In one example, the sea surface wind speed in the multi-source original sea surface wind field is corrected based on the spatiotemporal matching data results. The linear fitting is mainly performed based on the matched sea surface wind speed, and the fitting correction parameters of each satellite are obtained as follows: 、 、 、 、 、 、 、 , according to the correction parameters, the corrected sea surface wind speed is obtained. The calculation method of the corrected sea surface wind speed is as follows:

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] in, is the corrected sea surface wind speed of Sentinel-1A, is the sea surface wind speed before correction of Sentinel-1A, is the corrected sea surface wind speed from Sentinel-1B, is the sea surface wind speed before correction of Sentinel-1B, is the corrected sea surface wind speed of GF3, is the sea surface wind speed before correction of GF3, is the corrected sea surface wind speed for 1mCSAR, is the sea surface wind speed before correction for 1mCSAR.

[0078] (2.4) Interpolate the multi-source intermediate sea surface wind field to a grid of a specified resolution to obtain the multi-source target sea surface wind field corresponding to the first grid in the study area. In practical applications, due to the varying spatial resolutions of different SAR data, it is also necessary to interpolate the corrected multi-source intermediate sea surface wind field to a grid of uniform resolution.

[0079] (3) Fitting wind shear index: Based on the wind measuring station data corresponding to the study sea area, fit the wind shear index corresponding to the study sea area at different time periods and different sea level heights.

[0080] The wind shear index is used to measure the characteristics of wind speed changes with vertical height. It is related to sea level, surface characteristics, weather and climate, and seasonal changes. The wind shear index can be calculated using the exponential law method, and the expression is:

[0081] ;

[0082] in, and Height and The average sea surface wind speed at The wind shear index is different between the upper and lower layers of the ground during the day and at night, and there are differences between different altitude layers. The wind shear index of the day and night as well as the low, middle and high layers is fitted using the wind tower data at different altitude layers.

[0083] (IV) Wind field data interpolation, including the following (4.1) to (4.3):

[0084] (4.1) Using the wind shear index, interpolate the multi-source target sea surface wind field corresponding to the first grid to obtain the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights. In one example, for any sea surface height to be interpolated (e.g., any of the low, middle, or high layers), the sea surface wind speed included in the target sea surface wind field and the sea surface height at which it is located are known. Furthermore, the wind shear index between the sea surface height to be interpolated and the sea surface height at which the target sea surface wind field is located is known. Using the formula in (3), the target sea surface wind speed corresponding to the sea surface height to be interpolated can be obtained, and thus the target sea surface wind field corresponding to the sea surface height to be interpolated can be obtained.

[0085] (4.2) Temporally and spatially match the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights with the reanalysis data corresponding to the first grid. In one example, the ERA5 reanalysis data were interpolated to the same resolution based on the observation time and latitude and longitude data of the multi-source target sea surface wind fields to obtain the matched ERA5 reanalysis data.

[0086] (4.3) Based on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and their matching reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate the average sea surface wind field set.

[0087] In a specific implementation, first, the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights are randomly combined to obtain a sea surface wind field combination corresponding to the first grid at different sea level heights, and the sea surface wind field combination includes at least two source target sea surface wind fields; then, based on the sea surface wind field combination and its matched reanalysis data, a wind speed Weibull model corresponding to the first grid is established; finally, according to the wind speed probability distribution described by the wind speed Weibull model, the satellite sea surface average wind speed and the satellite sea surface average wind direction are calculated to obtain the satellite average sea surface wind field, and the reanalysis sea surface average wind speed and the reanalysis sea surface average wind direction are calculated to obtain the reanalysis average sea surface wind field.

[0088] (V) Construction of XGBoost wind field prediction model: For any first grid, the satellite average sea surface wind field corresponding to the first grid is used as the learning target, and the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the first grid and its neighboring first grids are used as model input to train the machine learning model. Specifically:

[0089] (5.1) Sample Data Generation: The satellite-mean and reanalysis-mean sea surface wind fields are converted into UV components of the wind field, namely U_mean_sar, V_mean_sar, U_mean_era5, and V_mean_era5. XGBoost training samples are generated based on the UV components of the wind field, as well as atmospheric pressure, temperature, and humidity data. U_mean_sar and V_mean_sar are used as learning targets, and the U_mean_era5 and V_mean_era5 data at the U_mean_sar and V_mean_sar grid points and their surrounding n*n (n=1, 3, 5, 7, etc.) grids, along with the atmospheric pressure, sea surface temperature, and sea surface humidity of the grids, are used as training features. A sample dataset is obtained, of which 80% is randomly selected as the training set and 20% as the validation set.

[0090] (5.2) Set different surrounding n*n grids, where n is 3-21. Use K-fold cross-validation to adjust XGBoost parameters and select the optimal model parameters and the optimal n value. K-fold cross-validation divides the sample data set into K parts, one of which is used as the validation set and the remaining K-1 parts as the training set. Each time, the model is trained with the training set and tested with the validation set to calculate the model error. Cross-validation is repeated K times, and K model validation results are returned. The average of the K results is used as the final model error. Here, K=10. The performance of the wind field prediction model is evaluated using multiple error statistical methods, including correlation coefficient (R), standard deviation (STD), absolute error (MAE), root mean square error (RMSE), and central root mean square error (E').

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] .

[0096] (5.3) Construct the parameter space of the XGBoost model. The XGBoost model is an additive training model based on the Boosting ensemble concept. It uses the forward distribution algorithm for greedy learning. Each iteration learns a CART tree to fit the prediction results of the previous k-1 trees and the residual of the sample value. The final prediction value is the result of adding the prediction values ​​of each decision tree: = .

[0097] The XGBoost model parameters mainly include the tree structure depth max_depth, the minimum child node weight threshold min_child_weight, if the weight of a tree node is less than this threshold, the subtree will not be split again, the loss reduction threshold gamma caused by the XGBoost decision tree split, and if the model overfits, it is necessary to adjust the subsampling parameter subsample, the feature sampling ratio of the entire tree colsample_bytree, the feature sampling ratio of a certain layer colsample_bylevel, the feature sampling ratio of a certain tree node colsample_bynode, and the L1L2 regularization parameters reg_alpha and reg_lambda. By adjusting the XGBoost parameters, the final XGBoost model, that is, the average wind field prediction model, is obtained.

[0098] (6) Average wind field calculation: Find the second grid in the study sea area. For any second grid, determine the reanalysis average sea surface wind field corresponding to the second grid and its neighboring second grids based on the reanalysis data. Input the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the second grid and its neighboring second grids into the trained machine learning model to obtain the satellite average sea surface wind field corresponding to the second grid at different sea level heights.

[0099] Specifically: Since the sea surface wind field has undergone multi-level quality control and there is a lot of missing data, the locations of the grid points with missing data and the ERA5 reanalysis data, atmospheric pressure, sea surface temperature, sea surface humidity and other sea surface characteristic data of the corresponding locations and surrounding grid points are found. The XGBoost wind field prediction model constructed in the previous step is used to calculate the satellite average sea surface wind field of the grid points with missing data, and the satellite average sea surface wind field of the complete area is obtained for regional wind resource assessment.

[0100] (VII) Calculation and evaluation of wind resource maps: First, based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea level heights, the satellite average sea surface wind field corresponding to the study sea area is obtained; then, based on the satellite average sea surface wind field corresponding to the study sea area, one or more of the effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution are calculated to obtain the wind resource map corresponding to the study sea area.

[0101] Specifically: Based on the satellite-averaged sea surface wind field obtained in the previous step, calculate the effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution; evaluate the wind energy size based on the effective wind energy density; evaluate the wind energy stability based on the wind energy variation coefficient; obtain the dominant wind direction in the target area and changes in wind direction stability through the wind direction and speed frequency distribution, thereby comprehensively evaluating the richness and stability of regional wind resources.

[0102] In summary, wind energy, as a common clean energy, has great development potential, especially in the field of wind power generation. However, the assessment of wind energy resources often requires several years or even decades of observation data. In actual construction applications, data collection is difficult and the data integrity is low. To address this defect, the embodiment of the present invention uses the relatively easy-to-obtain ERA5 reanalysis data, fully considers the sea surface characteristic data related to the wind farm and the relationship between the wind unit and the surrounding units, and establishes the XGBoost wind farm prediction model to ensure the integrity, rationality and consistency of regional data, thereby reducing the difficulty of collecting data during wind resource assessment and prediction, and reducing the time cost of data collection, providing a reference basis for reducing the construction risk of wind resource-related projects and improving the efficiency and utilization rate of wind energy resource construction.

[0103] Based on the above embodiment, the present invention provides a multi-source satellite offshore wind resource map analysis device, see Figure 3 The structure diagram of a multi-source satellite offshore wind resource map analysis device is shown, which mainly includes the following parts:

[0104] A preprocessing module 302 is configured to preprocess the multi-source original sea surface wind field in the study sea area to determine a multi-source target sea surface wind field corresponding to a first grid in the study sea area, where the first grid is a grid for which the multi-source original sea surface wind field has passed multi-level wind field quality control;

[0105] The mean sea surface wind field construction module 304 is configured to construct a mean sea surface wind field set based on the wind station data corresponding to the study sea area, the multi-source target sea surface wind field data corresponding to the first grid, and the reanalysis data. The mean sea surface wind field set includes the satellite mean sea surface wind field and the reanalysis mean sea surface wind field corresponding to different sea level heights of the first grid.

[0106] a missing sea surface wind field prediction module 306 for training a machine learning model using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area, so as to predict, through the trained machine learning model, the satellite average sea surface wind field corresponding to the second grid at different sea level heights based on the reanalysis data corresponding to the second grid in the study sea area, where the second grid is a grid for which the multi-source original sea surface wind field is missing or for which the multi-source original sea surface wind field corresponding to the second grid fails the multi-level wind field quality control;

[0107] The wind resource map generation module 308 is used to generate a wind resource map corresponding to the study sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights.

[0108] The multi-source satellite offshore regional wind resource map analysis device provided by an embodiment of the present invention trains a machine learning model based on the target sea surface wind field, reanalysis data, and sea surface feature data observed by multi-source satellites at a first grid, and then uses the machine learning model to calculate the satellite average sea surface wind field in a second grid with data in a non-satellite observation area or where data is missing, thereby generating a wind resource map for the entire study sea area. On the one hand, the relatively easy-to-obtain reanalysis data and sea surface feature data are used to ensure the integrity of regional data, consistency with actual observations, and data accuracy through a machine learning model. On the other hand, the difficulty of collecting data during wind resource assessment and prediction is reduced, and the time cost of data collection is reduced, thereby providing a reference basis for reducing the construction risks of wind resource-related projects and improving the benefits of wind energy resource construction.

[0109] In one embodiment, the pre-processing module 302 is specifically configured to:

[0110] Perform spatiotemporal matching of the offshore buoy data and multi-source original sea surface wind fields corresponding to the study area;

[0111] Using offshore buoy data, multi-level wind field quality control is performed on the multi-source original sea surface wind field to extract the multi-source original sea surface wind field that has passed the multi-level wind field quality control;

[0112] The wind speed correction coefficient is fitted based on the offshore buoy data, and the wind speed correction coefficient is used to perform wind speed quality correction on the multi-source original sea surface wind field that has passed the multi-level wind field quality control to obtain the multi-source intermediate sea surface wind field.

[0113] The multi-source intermediate sea surface wind field is interpolated into a grid of specified resolution to obtain the multi-source target sea surface wind field corresponding to the first grid in the study sea area.

[0114] In one embodiment, the average sea surface wind field construction module 304 is specifically configured to:

[0115] Based on the wind measurement station data corresponding to the study sea area, the wind shear index corresponding to the study sea area at different time periods and different sea level heights is fitted;

[0116] The multi-source target sea surface wind field corresponding to the first grid is interpolated using the wind shear index to obtain the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights;

[0117] Perform spatiotemporal matching on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and the reanalysis data corresponding to the first grid;

[0118] Based on the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights and their matched reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate the average sea surface wind field set.

[0119] In one embodiment, the average sea surface wind field construction module 304 is specifically configured to:

[0120] Randomly combining the multi-source target sea surface wind fields corresponding to the first grid at different sea level heights to obtain a sea surface wind field combination corresponding to the first grid at different sea level heights, wherein the sea surface wind field combination includes at least two source target sea surface wind fields;

[0121] Based on the sea surface wind field combination and its matching reanalysis data, a Weibull model of wind speed corresponding to the first grid is established;

[0122] According to the wind speed probability distribution described by the Weibull model, the satellite average sea surface wind speed and the satellite average sea surface wind direction are calculated to obtain the satellite average sea surface wind field, and the reanalysis average sea surface wind speed and the reanalysis average sea surface wind direction are calculated to obtain the reanalysis average sea surface wind field.

[0123] In one embodiment, the missing sea surface wind field prediction module 306 is specifically configured to:

[0124] For any first grid, the satellite average sea surface wind field corresponding to the first grid is taken as the learning target, and the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the first grid and its neighboring first grids are used as model inputs to train the machine learning model.

[0125] In one embodiment, the missing sea surface wind field prediction module 306 is specifically configured to:

[0126] Find the second grid within the research sea area;

[0127] For any second grid, the reanalysis average sea surface wind field corresponding to the second grid and its neighboring second grids is determined based on the reanalysis data. The reanalysis average sea surface wind field and sea surface characteristic data corresponding to the second grid and its neighboring second grids are input into the trained machine learning model to obtain the satellite average sea surface wind field corresponding to the second grid at different sea level heights.

[0128] In one embodiment, the wind resource map generation module 308 is specifically configured to:

[0129] Based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea level heights, the satellite average sea surface wind field corresponding to the study sea area is obtained;

[0130] Based on the satellite average sea surface wind field corresponding to the study sea area, one or more of the effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution are calculated to obtain the wind resource map corresponding to the study sea area.

[0131] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those in the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference can be made to the corresponding content in the aforementioned method embodiment.

[0132] An embodiment of the present invention provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned embodiments.

[0133] Figure 4 This is a structural diagram of an electronic device provided in an embodiment of the present invention. The electronic device 100 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43. The processor 40, the communication interface 43 and the memory 41 are connected via the bus 42; the processor 40 is used to execute an executable module stored in the memory 41, such as a computer program.

[0134] Memory 41 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between the system network element and at least one other network element is achieved through at least one communication interface 43 (which may be wired or wireless), and may utilize the Internet, a wide area network, a local area network, a metropolitan area network, or the like.

[0135] The bus 42 may be an ISA bus, a PCI bus, or an EISA bus. The bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0136] Among them, the memory 41 is used to store programs, and the processor 40 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0137] Processor 40 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method may be completed by hardware integrated logic circuits or software instructions in processor 40. The above processor 40 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processing unit (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It may implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention may be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory 41 , and the processor 40 reads the information in the memory 41 and completes the steps of the above method in combination with its hardware.

[0138] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be referred to the previous method embodiment and will not be repeated here.

[0139] If the 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 the present invention, or the portion 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 for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0140] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A multi-source satellite offshore wind resource map analysis method, characterized in that: include: Preprocessing the multi-source original sea surface wind field in the study sea area to determine a multi-source target sea surface wind field corresponding to a first grid in the study sea area, where the first grid is a grid to which the multi-source original sea surface wind field corresponding to the first grid passes multi-level wind field quality control; Based on the wind measurement station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid, constructing an average sea surface wind field set, wherein the average sea surface wind field set includes the satellite average sea surface wind field and the reanalysis average sea surface wind field corresponding to the first grid at different sea level heights; The machine learning model is trained using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area, so that the trained machine learning model can predict the satellite average sea surface wind field corresponding to the second grid in the study sea area at different sea surface heights based on the reanalysis data corresponding to the second grid, where the second grid is a grid that lacks the multi-source original sea surface wind field or whose corresponding multi-source original sea surface wind field fails to pass the multi-level wind field quality control. Generate a wind resource map corresponding to the research sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights; Based on the wind measurement station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid, constructing an average sea surface wind field set, including: fitting the wind shear index corresponding to the study sea area at different time periods and different sea surface heights based on the wind measurement station data corresponding to the study sea area; The multi-source target sea surface wind field corresponding to the first grid is interpolated using the wind shear index to obtain the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights; the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights is temporally and spatially matched with the reanalysis data corresponding to the first grid; based on the multi-source target sea surface wind field corresponding to the first grid at different sea surface heights and the matched reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate an average sea surface wind field set.

2. The multi-source satellite offshore wind resource map analysis method according to claim 1, characterized in that: Preprocessing the multi-source original sea surface wind field of the study sea area to determine the multi-source target sea surface wind field corresponding to the first grid in the study sea area includes: Performing spatiotemporal matching between the offshore buoy data corresponding to the study sea area and the multi-source original sea surface wind field; Using the offshore buoy data, performing multi-level wind field quality control on the multi-source original sea surface wind field to extract the multi-source original sea surface wind field that passes the multi-level wind field quality control; Fitting a wind speed correction coefficient based on the offshore buoy data, and using the wind speed correction coefficient to perform wind speed quality correction on the multi-source original sea surface wind field that has passed the multi-level wind field quality control to obtain a multi-source intermediate sea surface wind field; The multi-source intermediate sea surface wind field is interpolated into a grid of a specified resolution to obtain a multi-source target sea surface wind field corresponding to the first grid in the study sea area.

3. The multi-source satellite offshore wind resource map analysis method according to claim 1, characterized in that: Based on the multi-source target sea surface wind fields corresponding to the first grid at different sea surface heights and the matched reanalysis data, a wind speed Weibull model corresponding to the first grid is established to generate an average sea surface wind field set, including: Randomly combining the multiple source target sea surface wind fields corresponding to the first grid at different sea level heights to obtain a sea surface wind field combination corresponding to the first grid at different sea level heights, wherein the sea surface wind field combination includes at least two source target sea surface wind fields; Establishing a wind speed Weibull model corresponding to the first grid based on the sea surface wind field combination and the matched reanalysis data; According to the wind speed probability distribution described by the wind speed Weibull model, the satellite sea surface average wind speed and the satellite sea surface average wind direction are calculated to obtain the satellite average sea surface wind field, and the reanalysis sea surface average wind speed and the reanalysis sea surface average wind direction are calculated to obtain the reanalysis average sea surface wind field.

4. The multi-source satellite offshore wind resource map analysis method according to claim 1, characterized in that: Training a machine learning model using the average sea surface wind field set and sea surface characteristic data corresponding to the study sea area includes: For any of the first grids, the satellite average sea surface wind field corresponding to the first grid is used as the learning target, and the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the first grid and its neighboring first grids are used as model inputs to train the machine learning model.

5. The multi-source satellite offshore wind resource map analysis method according to claim 1, characterized in that: Predicting the satellite-averaged sea surface wind field corresponding to the second grid at different sea level heights by the trained machine learning model based on the reanalysis data corresponding to the second grid in the study sea area includes: Searching for a second grid within the research sea area; For any second grid, the reanalysis average sea surface wind field corresponding to the second grid and its neighboring second grids is determined based on the reanalysis data, and the reanalysis average sea surface wind field and sea surface characteristic data corresponding to the second grid and its neighboring second grids are input into the trained machine learning model to obtain the satellite average sea surface wind field corresponding to the second grid at different sea surface heights.

6. The multi-source satellite offshore wind resource map analysis method according to claim 1, characterized in that: Generating a wind resource map corresponding to the study sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights, including: Obtaining the satellite average sea surface wind field corresponding to the research sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights; According to the satellite average sea surface wind field corresponding to the study sea area, one or more of the effective wind energy density, maximum wind speed, effective wind hours, wind energy variation coefficient, and wind direction and speed frequency distribution are calculated to obtain a wind resource map corresponding to the study sea area.

7. A multi-source satellite offshore wind resource map analysis device, characterized in that: include: a preprocessing module, configured to preprocess the multi-source original sea surface wind field in the study sea area to determine a multi-source target sea surface wind field corresponding to a first grid in the study sea area, wherein the first grid is a grid to which the multi-source original sea surface wind field corresponding to the first grid passes multi-level wind field quality control; a mean sea surface wind field construction module, configured to construct a mean sea surface wind field set based on the wind measuring station data corresponding to the study sea area, and the multi-source target sea surface wind field and reanalysis data corresponding to the first grid, wherein the mean sea surface wind field set includes the satellite mean sea surface wind field and the reanalysis mean sea surface wind field corresponding to the first grid at different sea level heights; a missing sea surface wind field prediction module, configured to train a machine learning model using the average sea surface wind field set and the sea surface characteristic data corresponding to the study sea area, so as to predict, through the trained machine learning model, the satellite average sea surface wind field corresponding to a second grid in the study sea area at different sea surface heights based on the reanalysis data corresponding to the second grid, where the second grid is a grid that lacks the multi-source original sea surface wind field or whose corresponding multi-source original sea surface wind field fails the multi-level wind field quality control or has a small number of observations; a wind resource map generating module, configured to generate a wind resource map corresponding to the study sea area based on the satellite average sea surface wind fields corresponding to the first grid and the second grid at different sea surface heights; The average sea surface wind field construction module is specifically used to: fit the wind shear index corresponding to the research sea area at different time periods and different sea level heights based on the wind measurement station data corresponding to the research sea area; use the wind shear index to interpolate the multi-source target sea surface wind field corresponding to the first grid to obtain the multi-source target sea surface wind field corresponding to the first grid at different sea level heights; perform spatiotemporal matching on the multi-source target sea surface wind field corresponding to the first grid at different sea level heights with the reanalysis data corresponding to the first grid; establish a wind speed Weibull model corresponding to the first grid based on the multi-source target sea surface wind field corresponding to the first grid at different sea level heights and the matched reanalysis data, so as to generate an average sea surface wind field set.

8. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 6.

Images