Wheel hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching

Abstract

The application discloses a hub height wind speed prediction method based on three-dimensional coupling dynamic reference layer matching, and relates to the technical field of wind power wind speed prediction. The method comprises the following steps: obtaining meteorological input data output by a multi-layer numerical model, obtaining wind speeds of each standard layer and atmospheric stability parameters; performing multi-dimensional coupling classification according to a classification parameter and the atmospheric stability parameters, and obtaining several atmospheric turbulence state categories; for each atmospheric turbulence state, the optimal reference layer is determined in combination with the wind speeds of each standard layer and the target height observation wind speed; based on the optimal reference layer and the target height observation wind speed, a wind shear index parameter library of each category is constructed; according to the meteorological data at a to-be-predicted moment, the atmospheric turbulence state category is matched, the corresponding optimal reference layer wind speed and the wind shear index are called, and the target height prediction wind speed is output through power law extrapolation. The scheme is favorable for reducing the conversion deviation of static fixed parameters and improving the hub height wind speed deduction adaptability.

Description

Technical Field

[0001] This application relates to the field of wind power wind speed forecasting technology, and in particular to a hub height wind speed forecasting method based on three-dimensional coupled dynamic reference layer matching. Background Technology

[0002] Wind speed forecasting at hub height is a crucial foundational technology for wind farm power prediction and wind power resource assessment. Numerical model outputs are widely used for hub height wind speed extrapolation. Existing technologies often directly use a fixed wind shear index to convert wind speeds between different heights, or select a single fixed model layer as a reference layer for wind speed extrapolation, with wind speed conversion parameters remaining unchanged regardless of atmospheric conditions. Under actual atmospheric conditions, near-surface thermal, radiation, and wind field conditions collectively alter atmospheric turbulence. Fixed reference layers and fixed wind shear indices are ill-suited to the variable atmospheric turbulence conditions. The correlation between model layer wind speed and actual wind speed at hub height changes with the atmospheric environment. Fixed-parameter extrapolation methods are prone to introducing systematic biases, making it difficult to meet the needs of hub height wind speed extrapolation under different turbulent conditions, resulting in significant deviations between forecast results and actual conditions. Summary of the Invention

[0003] The purpose of this application is to provide a hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching, which aims to solve the technical problem of static solidification of wind speed extrapolation parameters in the prior art.

[0004] To achieve the above objectives, this application provides a method for predicting hub height wind speed based on three-dimensional coupled dynamic reference layer matching, comprising: Obtain meteorological input data from multi-level numerical models, and obtain wind speed and atmospheric stability parameters for each standard layer based on the meteorological input data; Based on the classification parameters of the meteorological input data and the atmospheric stability parameters, a multi-dimensional coupled classification is performed to obtain several atmospheric turbulence state categories; For each category of atmospheric turbulence, the optimal reference layer for that category is determined by combining the wind speed at each standard layer with the wind speed observed at the target height. For each of the atmospheric turbulence state categories, a corresponding wind shear index parameter library is established based on the wind speed observed at the optimal reference layer and target height. Based on the meteorological input data at the time to be predicted, the atmospheric turbulence state category is matched, the optimal reference layer wind speed and corresponding wind shear index under that category are retrieved, and the target height forecast wind speed is output through power law extrapolation.

[0005] In some optional embodiments, meteorological input data from multi-level numerical models are acquired, and wind speed and atmospheric stability parameters for each standard layer are obtained based on the meteorological input data, including: Read the meteorological input data output by the multi-level numerical model, perform format verification and quality screening on the meteorological input data, and output the valid meteorological input data; By analyzing the valid meteorological input data, standard layer meteorological element data, near-surface wind speed, and surface downward solar radiation flux are obtained. Based on the meteorological element data of the standard layer, the original wind field information of each standard layer is extracted, and the wind speed of each standard layer is generated. Based on the layer height and temperature parameters in the meteorological element data of the standard layer, the atmospheric stability parameters are solved to obtain the atmospheric stability parameters.

[0006] In some optional embodiments, multi-dimensional coupled classification is performed based on the classification parameters of the meteorological input data and the atmospheric stability parameters to obtain several atmospheric turbulence state categories, including: Extract the classification parameters from the meteorological input data to obtain the near-surface wind speed and the downward solar radiation flux from the Earth's surface; The near-surface wind speed, the downward solar radiation flux from the Earth's surface, and the atmospheric stability parameters are classified into different levels to obtain their respective level labels. The level identifiers of the near-surface wind speed, the level identifiers of the downward solar radiation flux, and the level identifiers of the atmospheric stability parameters at the same time are combined to generate a combined label. Based on the combined labels, several atmospheric turbulence state categories are obtained.

[0007] In some optional embodiments, the near-surface wind speed, the downward solar radiation flux, and the atmospheric stability parameters are respectively classified into levels to obtain their corresponding level labels, including: Pre-configure thresholds for near-surface wind speed, surface-to-surface solar radiation flux, and atmospheric stability parameters; The near-surface wind speed is compared with the corresponding classification threshold, and the corresponding level label is assigned; The surface solar radiation flux downwards is compared with the corresponding classification threshold, and the corresponding level label is assigned. The atmospheric stability parameters are compared with the corresponding classification thresholds, and the corresponding level identifiers are assigned.

[0008] In some optional embodiments, classification processing is performed based on the combined labels to obtain several atmospheric turbulence state categories, including: The combined tags generated at all times are traversed, and abnormal combined tags are removed; The remaining combined tags are aggregated and merged to form a set of non-overlapping categories; Each category set is defined as a category of atmospheric turbulence states.

[0009] In some optional embodiments, for each of the atmospheric turbulence state categories, the optimal reference layer for the corresponding category is determined by combining the wind speed at each standard layer with the wind speed observed at the target height, including: Filter the sample times belonging to the same atmospheric turbulence state category, and extract the wind speed of each standard layer and the wind speed observed at the target height at the corresponding time. For each standard layer, based on the sample time under the same atmospheric turbulence state category, the correlation between the wind speed of each standard layer and the wind speed observed at the target height is calculated to obtain the correlation results corresponding to each standard layer. The obtained correlation results are sorted to obtain a sorted sequence of correlation results; Based on the sorting sequence, the optimal reference layer corresponding to the atmospheric turbulence state category is selected.

[0010] In some optional embodiments, for each standard layer, based on the sample time under the same atmospheric turbulence state category, the correlation between the wind speed of each standard layer and the wind speed observed at the target height is calculated to obtain the correlation results corresponding to each standard layer, including: Obtain the wind speed sequence of each standard layer and the wind speed sequence observed at the target height for all sample times under the same atmospheric turbulence state category; The wind speed sequence and the wind speed sequence observed at the target height are checked for consistency, and samples with mismatched times are removed. Based on the verified wind speed sequence and the observed wind speed sequence at the target height, the correlation coefficient is calculated to obtain the correlation results for the corresponding standard layer.

[0011] In some optional embodiments, for each of the atmospheric turbulence state categories, a corresponding wind shear index parameter library is established based on the wind speed observed at the optimal reference layer and target height, including: Select samples belonging to the same atmospheric turbulence state category, and extract the wind speed of the optimal reference layer, the layer height of the optimal reference layer, and the wind speed observed at the target height; Based on the extracted parameters, the wind shear index is back-calculated to obtain multiple sets of original samples of wind shear index under the atmospheric turbulence state category. Data cleaning processing was performed on multiple sets of the original wind shear index samples to obtain effective wind shear index samples; Based on the effective wind shear index samples, a wind shear index parameter library corresponding to the atmospheric turbulence state category is constructed.

[0012] In some optional embodiments, data cleaning processing is performed on multiple sets of the original wind shear index samples to obtain effective wind shear index samples, including: Read all original samples of the wind shear index under the same atmospheric turbulence state category; Configure the value range of the wind shear index, and compare the original samples of each wind shear index with the value range; Remove the original samples of wind shear index that are outside the range of values, and keep the samples that fall within the range of values. The retained samples are deduplicated to output the effective wind shear index samples.

[0013] In some optional embodiments, the atmospheric turbulence state category is matched based on the meteorological input data at the time to be predicted, the optimal reference layer wind speed and corresponding wind shear index under that category are retrieved, and the target height forecast wind speed is output by power-law extrapolation, including: Read the meteorological input data for the time to be predicted, and parse it to obtain the classification parameters and atmospheric stability parameters for the time to be predicted; Based on the classification parameters of the time to be predicted and the atmospheric stability parameters, the corresponding atmospheric turbulence state category is obtained by matching; Retrieve the optimal reference layer wind speed and corresponding wind shear index corresponding to the atmospheric turbulence state category from the wind shear index parameter library; Obtain the layer height of the optimal reference layer and the predicted target height; Based on the retrieved wind speed, wind shear index, the layer height of the optimal reference layer, and the predicted target height, a power-law extrapolation calculation is performed to obtain the predicted wind speed at the target height.

[0014] The above-mentioned technical solution of this application has at least the following beneficial technical effects: The technical solution of this application obtains wind speed and atmospheric stability parameters for each standard layer using meteorological input data. It then performs multi-dimensional coupled classification using the classification parameters of the meteorological input data and atmospheric stability parameters to obtain atmospheric turbulence state categories. This enables the classification of different atmospheric turbulence conditions. Furthermore, it determines the optimal reference layer for each atmospheric turbulence state category and establishes a corresponding wind shear index parameter library. This facilitates the adaptation of the reference layer, wind shear index, and corresponding atmospheric turbulence state category. Subsequently, based on the meteorological input data at the forecast time, it matches the corresponding atmospheric turbulence state category and calls the wind speed and wind shear index of the matched reference layer for power-law extrapolation. This allows the parameters used for extrapolation to be dynamically selected according to the real-time atmospheric turbulence state, which helps reduce the conversion deviation caused by static fixed parameters and improves the adaptability of hub height wind speed extrapolation. Attached Figure Description

[0015] Figure 1This is a flowchart illustrating an embodiment of the hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching provided in this application; Figure 2 This application provides a specific embodiment of the novel method and a root mean square error heatmap of the forecasts from the European Centre for Central Mechanics and the old method. Figure 3 This is a scatter plot of different predicted hub height wind speeds versus observed wind speeds using a novel method according to a specific embodiment of this application. Figure 4 This is a scatter plot of wind speed at different forecast hub heights from the European Centre for Medium-Range Weather Forecasts versus observed wind speeds, according to a specific embodiment provided in this application. Figure 5 This is a scatter plot of different predicted hub height wind speeds versus observed wind speeds using the original method of a specific embodiment provided in this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to specific embodiments and accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of this application. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0017] The embodiments described in this application are only some, not all, of the embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments described herein without inventive effort are within the scope of protection of this application.

[0018] Wind speed forecasting at hub height is a crucial foundational technology for wind farm power prediction and wind power resource assessment. Numerical model outputs are widely used for hub height wind speed extrapolation. Existing technologies often directly use a fixed wind shear index to convert wind speeds between different heights, or select a single fixed model layer as a reference layer for wind speed extrapolation, with wind speed conversion parameters remaining unchanged regardless of atmospheric conditions. Under actual atmospheric conditions, near-surface thermal, radiation, and wind field conditions collectively alter atmospheric turbulence. Fixed reference layers and fixed wind shear indices are ill-suited to the variable atmospheric turbulence conditions. The correlation between model layer wind speed and actual wind speed at hub height changes with the atmospheric environment. Fixed-parameter extrapolation methods are prone to introducing systematic biases, making it difficult to meet the needs of hub height wind speed extrapolation under different turbulent conditions, resulting in significant deviations between forecast results and actual conditions.

[0019] To address the aforementioned technical issues, this application provides a method for predicting hub height and wind speed based on three-dimensional coupled dynamic reference layer matching.

[0020] The technical solution of this application obtains wind speed and atmospheric stability parameters for each standard layer using meteorological input data. It then performs multi-dimensional coupled classification using the classification parameters of the meteorological input data and atmospheric stability parameters to obtain atmospheric turbulence state categories. This enables the classification of different atmospheric turbulence conditions. Furthermore, it determines the optimal reference layer for each atmospheric turbulence state category and establishes a corresponding wind shear index parameter library. This facilitates the adaptation of the reference layer, wind shear index, and corresponding atmospheric turbulence state category. Subsequently, based on the meteorological input data at the forecast time, it matches the corresponding atmospheric turbulence state category and calls the wind speed and wind shear index of the matched reference layer for power-law extrapolation. This allows the parameters used for extrapolation to be dynamically selected according to the real-time atmospheric turbulence state, which helps reduce the conversion deviation caused by static fixed parameters and improves the adaptability of hub height wind speed extrapolation.

[0021] In some alternative embodiments, please refer to Figure 1 The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching includes the following steps: Step 1: Obtain meteorological input data from the multi-level numerical model output, and obtain the wind speed and atmospheric stability parameters of each standard layer based on the meteorological input data; In some optional embodiments, step 1 includes the following specific steps: Step 11: Read the meteorological input data output by the multi-layer numerical model, perform format verification and quality screening on the meteorological input data, and output valid meteorological input data; specifically, read the meteorological input data output by the RMAPS-WIND model, perform format verification and quality screening on the meteorological input data, and output valid meteorological input data. The meteorological input data includes zonal wind speed U, meridional wind speed V, perturbation potential temperature T, geopotential height perturbation PH, reference geopotential height PHB, and surface downward solar radiation flux SWDOWN. During the training phase, it also includes wind speed observed at the target height of the wind tower.

[0022] Step 12: Analyze the effective meteorological input data to separate the standard layer meteorological element data, near-surface wind speed, and surface downward solar radiation flux; specifically, analyze the effective meteorological input data to separate the standard layer meteorological element data, near-surface wind speed, and surface downward solar radiation flux; among which, the standard layer meteorological element data includes the zonal wind speed U, meridional wind speed V, disturbance potential temperature T, geopotential height disturbance PH, and reference geopotential height PHB corresponding to each standard pressure layer.

[0023] Step 13: Based on the meteorological element data of the standard layers, extract the original wind field information of each standard layer and generate the wind speed of each standard layer; specifically, based on the meteorological element data of the standard layers, extract the original wind field information of each standard layer, and generate the wind speed of each standard layer using the formula... The wind speeds at each standard level were calculated, where, Let be the wind speed at the i-th standard floor. Let be the zonal wind speed at the i-th standard layer. Let be the meridional wind speed at the i-th standard layer. Simultaneously, the intermediate height of each standard layer is calculated using geopotential height-related parameters; the calculation formula is: , In the formula, The height is the median height of the i-th standard floor. The potential height perturbation of the i-th layer. Let be the reference potential height of the i-th layer. For the potential height perturbation of the (i+1)th layer, Let i+1 be the reference potential height of the base layer. The first layer potential height perturbation, The height of the first-level reference potential is given by , and g is the gravitational acceleration, with a value of 9.8.

[0024] Step 14: Based on the layer height and temperature parameters in the standard layer meteorological data, solve for the atmospheric stability parameters to obtain the atmospheric stability parameters. Specifically, based on the layer height and temperature parameters in the standard layer meteorological data, solve for the atmospheric stability parameters to obtain the atmospheric stability parameters; the atmospheric stability parameter is the Richardson number Ri, and the formula for calculating the Richardson number Ri is: , In the formula, Ri is the Richardson number, and g is the acceleration due to gravity. For potential temperature, For potential temperature gradient, For wind speed gradient, For the height difference; prioritize the selection of parameters from the first and third layers of the model to complete the Richardson number Ri calculation.

[0025] Step 2: Perform multi-dimensional coupled classification based on the classification parameters of the meteorological input data and the atmospheric stability parameters to obtain several atmospheric turbulence state categories; In some optional embodiments, step 2 includes the following specific steps: Step 21: Extract the classification parameters from the meteorological input data to obtain the near-surface wind speed and the downward solar radiation flux from the ground surface. The near-surface wind speed is taken from the output wind speed of the first layer of the model.

[0026] Step 22: Classify the near-surface wind speed, surface-to-surface solar radiation flux, and atmospheric stability parameters into levels and obtain their corresponding level labels; In some optional embodiments, step 22 further includes the following specific steps: Step 221: Pre-configure the near-surface wind speed classification threshold, the surface-to-surface solar radiation flux classification threshold, and the atmospheric stability parameter classification threshold; specifically, the near-surface wind speed classification threshold, the surface-to-surface solar radiation flux classification threshold, and the atmospheric stability parameter classification threshold are pre-configured as shown in Table 1, Table 2, and Table 3, respectively.

[0027] Table 1. Criteria for Classifying Atmospheric Stability Table 1 presents the atmospheric stability classification criteria, where Ri is the Richardson number, used as an atmospheric stability parameter to characterize the stability of atmospheric stratification. Ri < 0 corresponds to an unstable atmospheric stratification, 0 ≤ Ri ≤ 0.25 corresponds to a neutral atmospheric stratification, and Ri > 0.25 corresponds to a stable atmospheric stratification. During implementation, the calculated Ri value is read and compared with each threshold in the table. Based on the comparison results, a corresponding atmospheric stability category label is assigned to the current time. This category label serves as one dimension of the three-dimensional coupled classification, used to subsequently generate combined labels to complete the classification of atmospheric turbulence states.

[0028] Table 2 Classification of Near-Surface Wind Speed ​​Levels Table 2 shows the classification of near-surface wind speeds. The near-surface wind speeds are taken from the output wind speed of the first layer of the model and are used to characterize the intensity of near-surface winds. Extremely low wind speeds correspond to wind speeds less than 2 m / s, low wind speeds correspond to wind speeds in the range of 2–4 m / s, medium wind speeds correspond to wind speeds in the range of 4–7 m / s, high wind speeds correspond to wind speeds in the range of 7–12 m / s, and extremely high wind speeds correspond to wind speeds greater than or equal to 12 m / s. During implementation, the parsed near-surface wind speeds are read and compared sequentially with the wind speed thresholds in the table, assigning a corresponding wind speed level label. This wind speed level label serves as one dimension of the three-dimensional coupled classification, used to subsequently generate combined labels to complete the classification of atmospheric turbulence states.

[0029] Table 3. Criteria for Diurnal Division of Solar Radiation Flux Downward from Earth's Surface Table 3 shows the criteria for dividing the day and night of the surface-downward solar radiation flux, where SWDOWN is the surface-downward solar radiation flux, used to characterize the intensity of solar radiation received by the Earth's surface. SWDOWN > 10 For daytime operating conditions, SWDOWN ≤ 10 For nighttime operation, during implementation, the surface downward solar radiation flux obtained from the analysis is read, SWDOWN is compared with the radiation threshold in the table, and the corresponding radiation category identifier is assigned. This radiation category identifier serves as one dimension input for the three-dimensional coupled classification, and is used to generate combined labels to complete the classification of atmospheric turbulence state categories.

[0030] Step 222: Compare the near-surface wind speed with the corresponding classification threshold and assign the corresponding level label; Step 223: Compare the downward solar radiation flux from the Earth's surface with the corresponding classification threshold and assign the corresponding level label; Step 224: Compare the atmospheric stability parameters with the corresponding classification thresholds and assign the corresponding level labels.

[0031] Step 23: Combine the level labels of near-surface wind speed, downward solar radiation flux, and atmospheric stability parameters corresponding to the same moment to generate combined labels; through three-dimensional coupling, a total of 3×5×2=30 combined labels are generated.

[0032] Step 24: Classify the data based on the combined labels to obtain several atmospheric turbulence state categories.

[0033] In some optional embodiments, step 24 further includes the following specific steps: Step 241: Iterate through all the combined tags generated at all time points and remove abnormal combined tags; Step 242: Aggregate and merge the remaining combined tags to form a set of non-overlapping categories; Step 243: Define each category set as a class of atmospheric turbulence states.

[0034] Step 3: For each atmospheric turbulence state category, combine the wind speed at each standard layer with the wind speed observed at the target height to determine the optimal reference layer for that category; In some optional embodiments, step 3 includes the following specific steps: Step 31: Select sample times belonging to the same atmospheric turbulence state category, and extract the wind speed of each standard layer and the wind speed observed at the target height at the corresponding time; limit the candidate reference layer to layers L1 to L4 to avoid the amplification of extrapolation error caused by layers L5 and above.

[0035] Step 32: For each standard layer, based on the sample time under the same atmospheric turbulence state category, calculate the correlation between the wind speed of each standard layer and the wind speed observed at the target height to obtain the correlation results corresponding to each standard layer. In some optional embodiments, step 32 further includes the following specific steps: Step 321: Obtain the wind speed sequence of each standard layer and the wind speed sequence observed at the target height for all sample times under the same atmospheric turbulence state category; Step 322: Perform a sample consistency check between the wind speed sequence and the wind speed sequence observed at the target height, and remove samples with mismatched times; Step 323: Based on the verified wind speed sequence and the wind speed sequence observed at the target height, calculate the correlation coefficient to obtain the correlation results of the corresponding standard layer.

[0036] Specifically, based on the verified wind speed sequence and the observed wind speed sequence at the target height, the correlation coefficient is calculated to obtain the correlation results for the corresponding standard layer; the Pearson correlation coefficient is used, and the calculation formula is as follows: , In the formula, Let Pearson correlation coefficient be the value of the i-th standard layer. Let be the wind speed at the i-th standard floor. Let be the sample mean of the wind speed at the i-th standard layer. Observe wind speed at target altitude. The mean of wind speed observed at the target altitude. Summation operation for the sample.

[0037] Step 33: Sort the obtained correlation results to obtain a sorted sequence of correlation results; Step 34: Based on the sorting sequence, select the optimal reference layer for the corresponding atmospheric turbulence state category. Specifically, based on the sorting sequence, select the optimal reference layer for the corresponding atmospheric turbulence state category; select the standard layer corresponding to the maximum Pearson correlation coefficient as the optimal reference layer, satisfying... ,in, As the optimal reference layer number, To perform the maximum value operation, Let be the Pearson correlation coefficient corresponding to the i-th standard layer.

[0038] Step 4: For each atmospheric turbulence state category, establish a corresponding wind shear index parameter library based on the wind speed observed at the optimal reference layer and target height; In some optional embodiments, step 4 includes the following specific steps: Step 41: Select samples belonging to the same atmospheric turbulence state category, and extract the wind speed of the optimal reference layer, the layer height of the optimal reference layer, and the wind speed observed at the target height; Step 42: Perform inverse calculation of the wind shear index based on the extracted parameters to obtain multiple sets of original wind shear index samples under this atmospheric turbulence state category; specifically, perform inverse calculation of the wind shear index based on the extracted parameters to obtain multiple sets of original wind shear index samples under this atmospheric turbulence state category; wind shear index The inverse calculation formula is: , In the formula, The wind shear index. For natural logarithm operations, Observe wind speed at target altitude. The optimal reference layer wind speed is given by z, where z is the target height. This is the optimal reference layer height.

[0039] Step 43: Perform data cleaning on multiple sets of original wind shear index samples to obtain valid wind shear index samples; In some optional embodiments, step 43 further includes the following specific steps: Step 431: Read all original samples of wind shear indices under the same atmospheric turbulence state category; Step 432: Configure the value range of the wind shear index, and compare the original samples of each wind shear index with the value range; the value range is set to -2 to 2.

[0040] Step 433: Remove the original samples of wind shear index that are outside the range of values, and keep the samples that fall within the range of values; Step 434: Perform duplicate sample removal on the retained samples and output the valid wind shear index samples.

[0041] Step 44: Based on the valid wind shear index samples, construct the wind shear index parameter library corresponding to this atmospheric turbulence state category. The field structure of the wind shear index parameter library is shown in Table 4.

[0042] Table 4. Description of Fields in the Wind Shear Index Parameter Library Table 4 is a field description table for the wind shear index parameter library. Among them, stability, wind_class, and rad_class are three-dimensional classification labels. stability represents the atmospheric stability category, wind_class represents the near-surface wind speed level, and rad_class represents the downward solar radiation flux category. The combination of these three uniquely identifies a class of atmospheric turbulence. best_layer is the optimal reference layer for this classification, used to store the optimal standard layer number obtained through correlation calculation and filtering under the current atmospheric turbulence state category. n_samples is the number of samples, recording the total number of valid samples participating in the wind shear index statistical calculation under the corresponding atmospheric turbulence state category. ~ Wind shear index The 10% to 90% quantiles are used to remove extreme outliers and can also be used for probability prediction interval output. , These are the wind shear indexes. The mean and standard deviation of the effective wind shear index are stored, and the statistical mean and standard deviation of the effective wind shear index samples under this atmospheric turbulence state category are selected during the forecast stage. The mean is used as the forecast value for the wind shear index. During implementation, after classifying atmospheric turbulence states and determining the optimal reference layer for each class, for each set of 3D classification labels, all valid samples under that class are extracted. The total number of samples is entered into the `n_samples` field, and the optimal reference layer number for that class is written into the `best_layer` field. Based on all interval-washed wind shear index samples for that class, the 10th and 90th quantiles of the wind shear index are calculated and assigned to [the appropriate value]. ~ Simultaneously calculate the statistical mean and statistical standard deviation of this group of wind shear index samples, and fill them in. , All fields are associated and stored to form a wind shear index parameter library. During the forecasting phase, the corresponding three-dimensional classification labels are obtained by matching the three-dimensional parameters at the time to be forecasted. The best_layer is retrieved from this table to obtain the optimal reference layer. As the value for wind shear index forecasting, deterministic wind speed forecasting is carried out by reading... ~ Used for probabilistic forecast interval output, to complete hub height and wind speed forecasts.

[0043] Step 5: Match the atmospheric turbulence state category based on the meteorological input data at the time to be predicted, retrieve the optimal reference layer wind speed and corresponding wind shear index under that category, and output the target height forecast wind speed through power law extrapolation.

[0044] In some optional embodiments, step 5 includes the following specific steps: Step 51: Read the meteorological input data for the time to be predicted, and parse it to obtain the classification parameters and atmospheric stability parameters for the time to be predicted; Step 52: Based on the classification parameters and atmospheric stability parameters of the time to be predicted, the corresponding atmospheric turbulence state category is matched and obtained; Step 53: Retrieve the optimal reference layer wind speed and corresponding wind shear index for this atmospheric turbulence state category from the wind shear index parameter library. As a wind shear index used for forecasting.

[0045] Step 54: Obtain the optimal reference layer height and the predicted target height; Step 55: Based on the retrieved wind speed, wind shear index, the height of the optimal reference layer, and the predicted target height, perform a power-law extrapolation to obtain the predicted wind speed at the target height.

[0046] Specifically, based on the retrieved wind speed, wind shear index, the optimal reference layer height, and the predicted target height, a power-law extrapolation calculation is performed to obtain the predicted wind speed at the target height; the power-law extrapolation calculation formula is: , In the formula, Forecast wind speed at target altitude. For the optimal reference layer wind speed, For the target height of the wheel hub, To be the optimal reference layer height, The 50th percentile of the wind shear index.

[0047] The technical solution of this application will be described in detail below with reference to a specific embodiment.

[0048] In this embodiment, measured data from 9 meteorological towers in February and March were selected as the target height wind speed. Data output from the RMAPS-WIND model was used as the meteorological input data for the multi-level numerical model. Comparative experiments on hub height wind speed forecasts were conducted using the new method of this application, the European Centre for Medium-Range Weather Forecasts (ECMWF) forecast, and the old method. All experimental procedures were strictly performed in accordance with the aforementioned steps 1 to 5.

[0049] Please see Figure 2 , Figure 2 A heatmap comparing the root mean square error (RMSE) of different methods at multiple sites is shown. The vertical axis represents each meteorological tower site and its corresponding month, while the horizontal axis arranges three forecast schemes: the new method, the European Centre for Medium-Range Weather Forecast (ECMWF) forecast, and the old method. The legend color represents the RMSE value; darker colors indicate smaller RMSE values ​​and better forecasts, while redder colors indicate larger RMSE values ​​and greater forecast bias. Figure 2 As can be seen, for most stations in the current month, the cells corresponding to the new method in this application are darker in color, and the root mean square error is generally lower than that of the European Centre for Central Mechanics forecast and the old method. There are only a few exceptions. The old method shows dark red cells at some stations, and the corresponding forecast error is significantly higher.

[0050] Please see Figure 3 , Figure 3 The diagram shows a scatter plot of the predicted wind speed versus the observed wind speed using the novel method of this application. The horizontal axis represents the observed wind speed, the vertical axis represents the predicted wind speed, and the dashed line represents the ideal prediction reference line. The red scatter points in the diagram represent sample data points. In this embodiment, the novel method yields a root mean square error of 1.96 m / s and a prediction deviation of 0.25 m / s. A large number of scatter points are close to the ideal prediction reference line, resulting in a relatively smaller degree of dispersion.

[0051] Please see Figure 4 , Figure 4The image shows a scatter plot of predicted wind speed versus observed wind speed from the European Centre for Medium-Range Nuclear Forces (ECMWF). The horizontal axis represents observed wind speed, the vertical axis represents predicted wind speed, and the dashed line is the ideal forecast reference line. The blue squares in the plot represent sample data points. The ECMWF forecast yielded a root mean square error of 2.26 m / s and a forecast bias of 0.82 m / s. Compared to... Figure 3 The degree to which the sample scatter points deviate from the ideal reference line increases.

[0052] Please see Figure 5 , Figure 5 The old method is shown as a scatter plot of predicted wind speed versus observed wind speed. The horizontal axis represents observed wind speed, the vertical axis represents predicted wind speed, the dashed line represents the ideal prediction reference line, and the green triangles in the figure represent sample data points. The old method yielded a root mean square error of 2.48 m / s and a prediction bias of 1.05 m / s. The dispersion of the sample scatter plot from the ideal reference line was the greatest among the three schemes.

[0053] The specific execution process of this embodiment is as follows: First, execute step 1 to obtain the meteorological input data output by the Ruitu-Fengguang model, and at the same time obtain the wind speed observed at the target height of 9 wind measurement towers in February and March. Then, execute steps 11 to 14 in sequence to complete the data quality screening and parameter analysis, and calculate the wind speed of each standard layer, the intermediate height of each standard layer, and the atmospheric stability parameters in the form of Richardson numbers.

[0054] Then, step 2 is executed, followed by steps 21 to 243, to extract near-surface wind speed and surface downward solar radiation flux. The three dimensions are assigned level labels according to the preset classification thresholds, and combined to generate combined labels. After abnormal label removal, aggregation and merging, several atmospheric turbulence state categories are obtained.

[0055] Next, proceed to step 3, steps 31 to 34, to limit the candidate reference layers to standard layers L1 to L4, filter sample times under the same atmospheric turbulence state category, complete the sample time consistency verification, calculate the Pearson correlation coefficient, and determine the optimal reference layer corresponding to each atmospheric turbulence state category based on the correlation coefficient ranking results.

[0056] Next, execute step 4, steps 41 to 44, extract the wind speed of the optimal reference layer, the layer height, and the wind speed observed at the target height, and back-calculate multiple sets of original wind shear index samples. Complete sample cleaning and deduplication according to the value range of -2 to 2. Based on the effective wind shear index samples, construct a wind shear index parameter library containing three-dimensional classification labels, optimal reference layer, number of samples, wind shear index quantiles, mean, and standard deviation fields.

[0057] Finally, step 5 is executed, following steps 51 to 55. Meteorological input data for the forecast time is read, and the corresponding atmospheric turbulence state category is matched. The optimal reference layer wind speed and the 50th quantile of the wind shear index corresponding to that category are retrieved. The optimal reference layer height and hub target height are obtained. The target height is then extrapolated using power-law calculations to forecast the wind speed. For all stations and months, forecast results are output using the new method of this application, the European Centre for Central Meteorological Observatory (ECHO) forecast, and the old method, respectively. The root mean square error and forecast bias index of each scheme are statistically obtained, generating... Figures 2 to 5 The corresponding experimental results charts.

[0058] Combination Figure 2 , Figure 3 , Figure 4 , Figure 5 The experimental results show that, compared with the European Centre for Central Weather Forecasts and the old methods, the technical solution of this application is adopted in this embodiment, which reduces the root mean square error of most stations, makes the scatter points closer to the ideal forecast reference line, and reduces the forecast deviation.

[0059] This application aims to protect a hub height wind speed forecasting method based on three-dimensional coupled dynamic reference layer matching. The technical solution of this application obtains the wind speed and atmospheric stability parameters of each standard layer using meteorological input data. It combines the classification parameters of the meteorological input data with the atmospheric stability parameters to carry out multi-dimensional coupled classification to obtain atmospheric turbulence state categories, which can realize the classification of different atmospheric turbulence conditions. Then, for each atmospheric turbulence state category, the optimal reference layer is determined and a corresponding wind shear index parameter library is established. This is conducive to the adaptation of the reference layer, wind shear index and corresponding atmospheric turbulence state category. Subsequently, the corresponding atmospheric turbulence state category is matched based on the meteorological input data at the time to be forecasted. The wind speed and wind shear index of the reference layer matched for this category are called for power-law extrapolation. This allows the parameters used for extrapolation to be dynamically selected according to the real-time atmospheric turbulence state, which helps to reduce the conversion deviation caused by static fixed parameters and improve the adaptability of hub height wind speed extrapolation.

[0060] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of this application and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of this application should be included within the protection scope of this application. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A method for predicting hub height wind speed based on three-dimensional coupled dynamic reference layer matching, characterized in that, include: Obtain meteorological input data from multi-level numerical models, and obtain wind speed and atmospheric stability parameters for each standard layer based on the meteorological input data; Based on the classification parameters of the meteorological input data and the atmospheric stability parameters, a multi-dimensional coupled classification is performed to obtain several atmospheric turbulence state categories; For each category of atmospheric turbulence, the optimal reference layer for that category is determined by combining the wind speed at each standard layer with the wind speed observed at the target height. For each of the atmospheric turbulence state categories, a corresponding wind shear index parameter library is established based on the wind speed observed at the optimal reference layer and target height. Based on the meteorological input data at the time to be predicted, the atmospheric turbulence state category is matched, the optimal reference layer wind speed and corresponding wind shear index under that category are retrieved, and the target height forecast wind speed is output through power law extrapolation.

2. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 1, characterized in that, Acquire meteorological input data from multi-level numerical models, and obtain wind speed and atmospheric stability parameters for each standard layer based on the meteorological input data, including: Read the meteorological input data output by the multi-level numerical model, perform format verification and quality screening on the meteorological input data, and output the valid meteorological input data; By analyzing the valid meteorological input data, standard layer meteorological element data, near-surface wind speed, and surface downward solar radiation flux are obtained. Based on the meteorological element data of the standard layer, the original wind field information of each standard layer is extracted, and the wind speed of each standard layer is generated. Based on the layer height and temperature parameters in the meteorological element data of the standard layer, the atmospheric stability parameters are solved to obtain the atmospheric stability parameters.

3. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 2, characterized in that, Based on the classification parameters of the meteorological input data and the atmospheric stability parameters, a multi-dimensional coupled classification is performed to obtain several atmospheric turbulence state categories, including: Extract the classification parameters from the meteorological input data to obtain the near-surface wind speed and the downward solar radiation flux from the Earth's surface; The near-surface wind speed, the downward solar radiation flux from the Earth's surface, and the atmospheric stability parameters are classified into different levels to obtain their respective level labels. The level identifiers of the near-surface wind speed, the level identifiers of the downward solar radiation flux, and the level identifiers of the atmospheric stability parameters at the same time are combined to generate a combined label. Based on the combined labels, several atmospheric turbulence state categories are obtained.

4. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 3, characterized in that, The near-surface wind speed, the downward solar radiation flux, and the atmospheric stability parameters are classified into levels to obtain their respective level labels, including: Pre-configure thresholds for near-surface wind speed, surface-to-surface solar radiation flux, and atmospheric stability parameters; The near-surface wind speed is compared with the corresponding classification threshold, and the corresponding level label is assigned; The surface solar radiation flux downwards is compared with the corresponding classification threshold, and the corresponding level label is assigned. The atmospheric stability parameters are compared with the corresponding classification thresholds, and the corresponding level identifiers are assigned.

5. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 3, characterized in that, Based on the combined labels, several atmospheric turbulence state categories are obtained, including: The combined tags generated at all times are traversed, and abnormal combined tags are removed; The remaining combined tags are aggregated and merged to form a set of non-overlapping categories; Each category set is defined as a category of atmospheric turbulence states.

6. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 1, characterized in that, For each of the aforementioned atmospheric turbulence state categories, the optimal reference layer for the corresponding category is determined by combining the wind speed at each standard layer with the wind speed observed at the target altitude, including: Filter the sample times belonging to the same atmospheric turbulence state category, and extract the wind speed of each standard layer and the wind speed observed at the target height at the corresponding time. For each standard layer, based on the sample time under the same atmospheric turbulence state category, the correlation between the wind speed of each standard layer and the wind speed observed at the target height is calculated to obtain the correlation results corresponding to each standard layer. The obtained correlation results are sorted to obtain a sorted sequence of correlation results; Based on the sorting sequence, the optimal reference layer corresponding to the atmospheric turbulence state category is selected.

7. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 6, characterized in that, For each standard layer, based on the sample time under the same atmospheric turbulence state category, the correlation between the wind speed of each standard layer and the wind speed observed at the target height is calculated to obtain the correlation results corresponding to each standard layer, including: Obtain the wind speed sequence of each standard layer and the wind speed sequence observed at the target height for all sample times under the same atmospheric turbulence state category; The wind speed sequence and the wind speed sequence observed at the target height are checked for consistency, and samples with mismatched times are removed. Based on the verified wind speed sequence and the observed wind speed sequence at the target height, the correlation coefficient is calculated to obtain the correlation results for the corresponding standard layer.

8. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 1, characterized in that, For each of the aforementioned atmospheric turbulence state categories, a corresponding wind shear index parameter library is established based on the wind speed observed at the optimal reference layer and target height, including: Select samples belonging to the same atmospheric turbulence state category, and extract the wind speed of the optimal reference layer, the layer height of the optimal reference layer, and the wind speed observed at the target height; Based on the extracted parameters, the wind shear index is back-calculated to obtain multiple sets of original samples of wind shear index under the atmospheric turbulence state category. Data cleaning processing was performed on multiple sets of the original wind shear index samples to obtain effective wind shear index samples; Based on the effective wind shear index samples, a wind shear index parameter library corresponding to the atmospheric turbulence state category is constructed.

9. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 8, characterized in that, Data cleaning and processing were performed on multiple sets of the original wind shear index samples to obtain valid wind shear index samples, including: Read all original samples of the wind shear index under the same atmospheric turbulence state category; Configure the value range of the wind shear index, and compare the original samples of each wind shear index with the value range; Remove the original samples of wind shear index that are outside the range of values, and keep the samples that fall within the range of values. The retained samples are deduplicated to output the effective wind shear index samples.

10. The hub height wind speed prediction method based on three-dimensional coupled dynamic reference layer matching according to claim 1, characterized in that, Based on the meteorological input data for the forecast time, the atmospheric turbulence state category is matched, the optimal reference layer wind speed and corresponding wind shear index under that category are retrieved, and the target height forecast wind speed is output via power-law extrapolation, including: Read the meteorological input data for the time to be predicted, and parse it to obtain the classification parameters and atmospheric stability parameters for the time to be predicted; Based on the classification parameters of the time to be predicted and the atmospheric stability parameters, the corresponding atmospheric turbulence state category is obtained by matching; Retrieve the optimal reference layer wind speed and corresponding wind shear index corresponding to the atmospheric turbulence state category from the wind shear index parameter library; Obtain the layer height of the optimal reference layer and the predicted target height; Based on the retrieved wind speed, wind shear index, the layer height of the optimal reference layer, and the predicted target height, a power-law extrapolation calculation is performed to obtain the predicted wind speed at the target height.

Images