Ground wind speed measurement and calculation correction method considering convective complex environment characteristics

Abstract

The invention provides a ground wind speed measurement and calculation correction method and device considering convective complex environment characteristics. The method comprises the following steps: acquiring original historical data of a target station; preprocessing the original historical data to generate standardized historical data; according to the threshold values of the wind speed, the duration time and the convection parameters, the type of the gale process is judged on the basis of standardized historical data such as meteorological observation and forecast, early warning information and disaster-causing events, and a common case, a cold air systematic gale case and a convective gale case are screened out; the method comprises the following steps: training and debugging LSTM, XGboost and Stacking models under each category based on standardized historical data such as weather forecast and weather observation according to different gale process categories, and generating a wind speed measurement and calculation classification correction model; and for real-time wind speed measurement and calculation, judging the type of a gale process, and automatically matching a proper wind speed measurement and calculation correction model in the wind speed measurement and calculation classification correction model to correct wind speed measurement and calculation.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a correction method and device for ground wind speed measurement taking into account complex convective environment characteristics. Background Art

[0002] At present, the correction of ground wind speed measurement data is usually obtained based on topography, meteorological observations, etc. using error statistical analysis methods or artificial intelligence models. For example, combining high-resolution terrain elevation data and geomorphological data, using the empirical relationship of the influence of topography on wind speed, constructing the preliminary correction results of high-resolution wind speed at the target point or local area, and further based on the historical observation of wind speed at the target point, using statistical analysis methods to calculate the wind speed measurement error at the target point, and combining the preliminary wind speed correction results to obtain the final wind speed correction results; or, using high-resolution topography data, meteorological forecast data, and historical observation data to train artificial intelligence models, obtain a correction model for ground wind speed measurement, which is used to correct wind speed measurement. The above wind speed measurement correction models usually do not consider different weather processes and meteorological characteristics under different weather backgrounds. In the specific practice process, it is found that although these wind speed measurement correction models can reduce the overall error level of wind speed measurement, the wind speed correction effect is poor under strong wind process conditions, especially convective strong wind process conditions. Therefore, it is necessary to construct a correction method for ground wind speed measurement that considers the characteristics of complex convective environments. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a correction method for ground wind speed measurement taking into account the complex convective environment characteristics, so as to improve the problem that the error of the wind speed measurement at the current site is still large after correction.

[0004] The present application provides a method for calculating and correcting ground wind speed taking into account the complex convective environment characteristics, which includes: Obtain the original historical data of the target site, including weather forecasts, weather observations, warning information, and disaster-causing events; Preprocess the original historical data of the target site obtained above to generate standardized historical data; determine the category of the strong wind process based on the standardized historical data according to the thresholds including wind speed, duration, and convection parameters, and screen out common cases, cold air systemic strong wind cases, and convective strong wind cases; Based on standardized historical data and according to different types of strong wind processes, LSTM, XGboost, and Stacking models are trained and debugged to generate a wind speed measurement classification correction model; For real-time wind speed measurement, the category of strong wind process is judged by combining the thresholds including wind speed, duration and convection parameters, and the appropriate wind speed measurement correction model is automatically matched in the wind speed measurement classification correction model to correct the wind speed measurement results.

[0005] Specifically, the generation of the wind speed measurement correction model A for a common case is as follows: The generation of wind speed calculation correction model B for cold air systematic gale cases is as follows: The generation of wind speed calculation correction model C for convective gale cases is as follows: Among them, Xa and Xb are feature tensors constructed with the standardized historical data of conventional meteorological element forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for common cases and cold air systematic gale cases, respectively; Xc is a feature tensor constructed with the standardized historical data of conventional meteorological element forecasts, standardized historical data of convective parameter forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for convective gale cases; Y is a label tensor constructed with the standardized historical data of wind speed measurement errors of the target forecast batch; T is the wind speed measurement error threshold; Y is the label tensor constructed with the standardized historical data of wind speed measurement errors of the target forecast batch; predict is the wind speed measurement error prediction value, WS orig Standardized data for wind speed measurement, WS corr is the wind speed measurement standard data correction value, Y′ predict is the predicted value of wind speed measurement error after threshold control.

[0006] Specifically, the specific process of the Stacking model includes: Based on the standardized historical data of wind speed measurement and the standardized historical data of wind speed observation, the standardized historical data of wind speed measurement error is calculated; The feature tensor is constructed using the standardized historical data of meteorological forecasts including conventional meteorological elements such as wind speed, temperature, humidity, and air pressure, as well as the standardized historical data of meteorological observations of wind speed, and the label tensor is constructed using the standardized historical data of wind speed error; the feature tensor and label tensor are divided into a training set and a test set; The training set and the test set are input into the artificial intelligence model for training and the model parameters are adjusted to obtain multiple wind speed measurement error prediction models as a Stacking first-level learner; Input the outputs of the multiple wind speed measurement error prediction models as new features into the secondary learner, train the Stacking secondary learner and adjust the parameters; A wind speed measurement error prediction model is constructed based on the Stacking first-level learner and the Stacking second-level learner.

[0007] Specifically, the feature tensor is constructed using the standardized historical data of weather forecasts and the standardized historical data of wind speed meteorological observations, including: Based on the standardized historical data of wind speed measurement and the standardized historical data of wind speed observation, the standardized historical data of wind speed measurement error is calculated; Based on the standardized historical data of wind speed calculation, wind speed observation, and wind speed calculation error of the same forecast batch in the past few days, feature X is constructed. di ; where X di The dimension is R×C, the rows R represent different forecast time steps, and the columns C represent different meteorological elements; Based on the feature X di , concatenate the features of the past few days to form a feature X with a dimension of R×(nC) dn , where n represents the number of days in the past; based on the X dn , the input features of the machine learning model are obtained using the scaling method; the label feature is the error of the historical wind speed forecast of the current batch; The number of days n is adjusted, and a sensitivity experiment is performed to obtain the optimal value of n, and thereby the input feature X and label feature Y of the machine learning model are constructed.

[0008] Specifically, the wind speed estimation correction model is generated, including: Calculate the wind speed measurement error prediction value by using the wind speed measurement error prediction model; Using the wind speed measurement error prediction value and the wind speed measurement standardized historical data, a preliminary correction value of the wind speed measurement standardized historical data is calculated; Based on the standardized historical data of wind speed observation, the preliminary correction value of the standardized historical data of wind speed measurement is checked, and the wind speed measurement error threshold is set according to the test result. This process is repeated to obtain the optimal threshold of wind speed measurement error, so as to control the range of wind speed measurement error prediction value; Based on the wind speed calculation error prediction model and the wind speed calculation error optimal threshold, a wind speed calculation correction model is constructed.

[0009] Specifically, the categories of gale processes are determined based on standardized historical data including meteorological observations, warning information, and disaster-causing events, including: Based on standardized historical data including warning information and disaster-causing events, combined with national and industry standards, thresholds for wind speed, duration of high winds, and convection parameters are established; Based on standardized historical data of meteorological observations, combined with thresholds including wind speed, duration of strong winds, and convection parameters, the category of strong wind processes is determined.

[0010] Specifically, the determination of the type of the gale process includes: Based on the standardized historical data of meteorological observations, combined with thresholds including wind speed and duration of strong winds, ordinary cases and strong wind cases are determined; Based on the standardized historical data of meteorological forecasts, convective parameters are used in strong wind cases, and Pearson correlation analysis is applied to obtain highly correlated convective parameters. Based on the highly correlated convective parameters, the k-means clustering method is further used to screen out cold air systematic strong wind cases and convective strong wind cases.

[0011] Specifically, based on the highly correlated convective parameters, the k-means clustering method is used to screen out cold air systemic gale cases and convective gale cases, including: The data including highly relevant convective parameters and historical data of meteorological elements are collated and formed into a training data set that meets the k-means input format; The k-means model is trained based on the training data set that meets the k-means input format, and the number of centers is determined through sensitivity experiments and used as the number of case types; Based on the results of the sensitivity experiment, cases of cold air systemic strong winds and convective strong winds were screened out.

[0012] Specifically, for real-time wind speed measurement, the category of the strong wind process is determined by combining the thresholds including wind speed, duration, and convection parameters, and the appropriate wind speed measurement correction model is automatically matched in the wind speed measurement classification correction model to correct the wind speed measurement, including: Obtain real-time wind speed measurement and preprocess it into standardized real-time data for weather forecast; Determine the category of the strong wind process based on thresholds including wind speed, duration, and convective parameters; According to the category of strong wind process, the wind speed measurement is corrected by matching the appropriate wind speed measurement correction model from the wind speed measurement classification correction model.

[0013] The present application also provides a ground wind speed correction device taking into account the complex convective environment characteristics, which includes: Data acquisition module, which is used for the original historical data of target sites, including weather forecasts, weather observations, warning information, and disaster-causing events; A data preprocessing module is used to preprocess the original historical data of the target site obtained above to generate standardized historical data; The gale process classification module is used to determine the gale process category based on thresholds including wind speed, duration, and convection parameters, and based on standardized historical data including meteorological observations, warning information, and disaster-causing events, and to screen out common cases, cold air systemic gale cases, and convective gale cases; The wind speed estimation correction model training module is used to train and debug LSTM, XGboost, and Stacking models according to different categories of strong wind processes based on standardized historical data including weather forecasts and meteorological observations, and generate a wind speed estimation classification correction model; The real-time wind speed measurement and correction module is used to judge the category of the strong wind process based on the thresholds including wind speed, duration, and convection parameters, and select the appropriate wind speed correction model to correct the real-time measured wind speed accordingly.

[0014] Beneficial technical effects: This method can significantly improve the accuracy of wind speed measurement. After correction, the average absolute error of wind speed measurement is greatly reduced, which means that the average deviation between the measured value and the actual wind speed value is significantly reduced; this method can enhance the timeliness and adaptability of wind speed measurement. This method designs a set of efficient real-time wind speed measurement and correction methods. By combining wind speed, duration, and convective parameter thresholds, it can quickly determine the category of strong wind processes, and automatically match the most suitable model in the wind speed measurement classification correction model to instantly optimize the wind speed measurement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of the revised method for surface wind speed calculation considering the complex convective environment characteristics; Figure 2 A schematic diagram of a device for a correction method for surface wind speed measurement taking into account the complex convective environment characteristics; Figure 3 This is a schematic diagram of the correction effect of the wind speed measurement at a meteorological station in mid-January 2023; Figure 4 Schematic diagram of the correction effect of wind speed measurement at a meteorological station in mid-July 2023. DETAILED DESCRIPTION

[0016] like Figure 1 As shown, the embodiment of the present application provides a method for calculating and correcting ground wind speed taking into account the complex convective environment characteristics, including: Step S110: Acquire original historical data such as weather forecasts, weather observations, warning information, and disaster-causing events of the target site.

[0017] Original historical weather forecast data, including: historical forecast data of conventional meteorological elements such as temperature, humidity, air pressure, precipitation, dew point temperature, wind speed, wind direction, and historical forecast data of convective parameters such as convective effective potential energy (CAPE), convective inhibition energy (CIN), and K index; Original historical data of meteorological observations, including historical observation data of meteorological elements such as gusts, wind speed, and wind direction; Original historical data of warning information, including: historical warning information issued by meteorological departments at all levels and historical warning information of the industry; Original historical data of disaster-causing events, including disaster records from various levels of civil affairs, disaster relief, meteorology and other government departments and industries.

[0018] Step S120: pre-processing the acquired original historical data rows of the target site (quality control, filling or discarding default values, standardization, etc.).

[0019] The original historical data of weather forecasts are extracted using a spatial interpolation method for forecasts of various meteorological elements at target points, forecast batches are sorted, forecast batches with more missing forecast times are discarded, and forecast batches with fewer missing forecast times are filled with time interpolation, and then standardized historical data of weather forecasts are generated according to a prefabricated format; Performing quality control (e.g., limit value check, internal consistency check, spatial consistency check, temporal consistency check, etc.) on the original historical data of meteorological observations, discarding observation periods with more consecutive missing measurements, filling observation periods with fewer consecutive missing measurements with interpolation, extracting required meteorological elements, and generating standardized historical data of meteorological observations in a pre-fabricated format; Extract the warning information related to the target site from the original historical data of the warning information, remove the duplicate warnings in the national, provincial, municipal and county multi-level warnings, and generate standardized historical data of meteorological warning information; For the original historical data of the disaster-causing events, the disaster-causing information related to the target site is extracted, and the duplicate records in the national, provincial, municipal and county multi-level disaster-causing event records and industry disaster records are eliminated to generate standardized historical data of disaster-causing events.

[0020] Step S130: According to the thresholds of wind speed, duration, and convective parameters, the category of the gale process is determined based on standardized historical data such as meteorological observations, warning information, and disaster-causing events, and ordinary cases, cold air systemic gale cases, and convective gale cases are selected.

[0021] Based on standardized historical data of early warning information, disaster-causing events, and meteorological observations, combined with national standards and industry standards, thresholds for wind speed, duration of high winds, and convection parameters are established; Based on the thresholds such as wind speed and duration of strong winds and standardized historical data of meteorological observations, the category of the strong wind process is judged, and ordinary cases and strong wind cases are determined.

[0022] In the case of strong winds, based on the standardized historical data of weather forecast, a variety of convective parameters were analyzed using Pearson correlation to obtain highly correlated convective parameters; Furthermore, based on the highly correlated convective parameters, the k-means clustering method is further used to screen out cases of systematic strong winds due to cold air and cases of convective strong winds, including: organizing the highly correlated convective parameters, historical data of meteorological elements and other information, and forming a training data set that meets the k-means input format; training the k-means model based on the training data set that meets the k-means input format, determining the number of centers through sensitivity experiments and using it as the number of case types; based on the results of the sensitivity experiments, screening out cases of systematic strong winds due to cold air and cases of convective strong winds.

[0023] Step S140: Based on standardized historical data such as weather forecasts and meteorological observations, and according to different categories of strong wind processes, train and debug LSTM, XGboost, and Stacking models to generate a wind speed measurement classification correction model.

[0024] For the common case, based on the standardized historical data of meteorological forecast and meteorological observation of conventional meteorological elements such as wind speed, temperature, humidity, and air pressure, the LSTM model is trained to obtain the wind speed measurement correction model A; For the cold air systematic gale case, based on the standardized historical data of meteorological forecast and meteorological observation of conventional meteorological elements such as wind speed, temperature, humidity, and air pressure, LSTM was trained to obtain wind speed calculation correction model 1, and LSTM and XGboost were trained to obtain wind speed calculation correction model 2, and the Stacking method was further used to obtain wind speed calculation correction model B; For the case of convective strong winds, based on the standardized historical data of meteorological forecasts and standardized historical data of meteorological observations of conventional meteorological parameters such as wind speed, temperature, humidity, and air pressure and highly correlated convective parameters, LSTM is trained to obtain wind speed calculation correction model 3, LSTM and XGboost models are trained to obtain wind speed calculation correction model 4, and the Stacking method is further used to obtain wind speed calculation correction model C.

[0025] The wind speed calculation correction model A, wind speed calculation correction model B and wind speed calculation correction model C constitute a wind speed calculation classification correction model.

[0026] Step S150: For real-time wind speed measurement, the category of the strong wind process is determined by combining the wind speed, duration, highly correlated convection parameters and corresponding thresholds, and accordingly, a suitable wind speed correction model is automatically matched in the wind speed measurement classification correction model to correct the wind speed measurement.

[0027] Obtain real-time wind speed measurement and process it into standardized weather forecast data; Determine the category of the strong wind process based on wind speed, duration, high-correlation convective parameters and corresponding thresholds; According to the category of strong wind processes, the wind speed measurement is corrected by matching the appropriate wind speed measurement correction model from the wind speed measurement classification correction models (wind speed measurement correction model A, wind speed measurement correction model B, wind speed measurement correction model C).

[0028] In the process of implementing the above scheme, by utilizing the historical data of weather forecasts, meteorological observations, warning information, disaster-causing events, etc. of the target site to judge different strong wind weather processes, and establishing wind speed measurement and correction models respectively, the ground wind speed measurement and correction model can be adapted to the changes in different weather processes, effectively improving the accuracy of the ground wind speed measurement data.

[0029] As an optional implementation method of the above-mentioned step S110, the original historical data of weather forecasts of the above-mentioned target site can be obtained from the China Meteorological Administration Mesoscale Numerical Weather Prediction Operational System (CMA-MESO); the original historical data of meteorological observations of the above-mentioned target site can be obtained from the national ground meteorological observation station network of the China Meteorological Administration; the original historical data of warning information of the above-mentioned target site can be obtained from the national meteorological warning information database and industry warning information database of the China Meteorological Administration; the original historical data of disaster-causing events of the above-mentioned target site can be collected from government departments and industry records such as civil affairs, disaster relief, and meteorology at all levels.

[0030] As an optional implementation of the above step S120, preprocessing (quality control, default value filling or discarding, standardization, etc.) the original historical data of the target site obtained above includes: Step S121: extracting forecasts of various meteorological elements at target points from the original historical weather forecast data using a spatial interpolation method; sorting forecast batches, discarding forecast batches with more missing forecast times, and performing time interpolation filling on forecast batches with fewer missing forecast times; and then generating standardized historical weather forecast data according to a prefabricated format; An implementation example of the above-mentioned step S121 is as follows: the original historical data of weather forecast comes from numerical forecasts such as CMA-MESO, which is grid data, and a spatial interpolation method (such as nearest point interpolation, bilinear interpolation, machine learning statistical regression, etc.) is used to obtain historical forecast data of various meteorological elements at the target point; the number of forecast times contained in each forecast batch is counted, and the forecast batches with the number of forecast times lower than the preset threshold are discarded, and the forecast batches with the number of forecast times higher than the preset threshold are retained; for the retained forecast batches, if some forecast times are missing, the historical forecast data of various meteorological elements at the target point are filled using a time interpolation method (such as linear interpolation, time series statistics, machine learning regression, etc.); and then the standardized historical data of weather forecast is generated according to a prefabricated format.

[0031] Step S122: performing quality control on the original historical data of meteorological observations, discarding observation periods with more consecutive missing observations, filling observation periods with fewer consecutive missing observations with interpolation, extracting required meteorological elements and generating standardized historical data of meteorological observations in a pre-made format; An implementation method of the above-mentioned step S122 is, for example: organizing historical meteorological observation data through quality control methods (such as limit value check, internal consistency check, spatial consistency check, time consistency check, etc.), discarding observation periods with more consecutive missing measurements, filling observation periods with fewer consecutive missing measurements through interpolation (such as time series analysis interpolation, linear interpolation, machine learning regression interpolation, etc.), and then extracting the required meteorological elements and generating historical meteorological observation standardized data according to a pre-made format.

[0032] Step S123: extracting the warning information related to the target site from the original historical data of the warning information, removing duplicate warnings in the national, provincial, municipal and county multi-level warnings and industry warnings, and generating standardized historical data of the warning information; An implementation method of the above-mentioned step S123 is, for example: extracting important fields (such as time period, location, warning category, warning level, etc.) through a matching algorithm, integrating national, provincial, municipal and county multi-level warnings and industry warnings through a deduplication algorithm, obtaining preliminary and effective warning information data, and then generating standardized historical data of warning information according to a preset format.

[0033] Step S124: extract the high wind disaster information related to the target site from the original historical data of the disaster event, remove the duplicate records in the national, provincial, municipal and county multi-level disaster event records and industry disaster records, and generate standardized data of historical disaster events.

[0034] An implementation method of the above-mentioned step S124 is, for example: extracting important fields (such as time period, impact range, disaster category, disaster level, etc.) through a matching algorithm, integrating national, provincial, municipal and county multi-level disaster event records and industry disaster records through a deduplication algorithm, obtaining preliminary and effective disaster event data, and then generating standardized historical data of disaster events according to a preset format.

[0035] As an optional implementation of the above step S130, judging the category of the gale process based on thresholds such as wind speed and duration and convective parameters, based on standardized historical data such as meteorological observations, warning information, and disaster-causing events, may include: Step S131: Based on standardized historical data such as warning information, disaster-causing events, and meteorological observations, combined with national standards, industry standards, etc., thresholds for wind speed, duration of strong winds, etc. are established.

[0036] An implementation method of the above-mentioned step S131 is, for example: for the target point, target industry, target operation scene, threshold 1 of wind speed, duration of strong winds, etc. is obtained by combining national standards, industry standards, etc.; for standardized historical data such as warning information and disaster-causing events, standardized historical data of meteorological observations are matched to obtain threshold 2; threshold 1 is corrected with threshold 2 to determine thresholds such as wind speed and duration of strong winds.

[0037] Step S132: Based on the thresholds such as wind speed and gale duration and the standardized historical data of meteorological observations, the category of the gale process is determined, and common cases and gale cases are determined.

[0038] An implementation example of the above-mentioned step S132 is as follows: using thresholds such as wind speed and duration of strong winds to match the corresponding observation time periods in the standardized historical data of meteorological observations, and obtaining the standardized historical data of meteorological observations during strong winds and the standardized historical data of meteorological observations during ordinary times; based on the above-mentioned strong wind time periods and ordinary times, in the standardized historical data of meteorological forecasts, selecting the corresponding standardized historical data of strong wind time period forecasts and the standardized historical data of meteorological forecasts during ordinary times; forming ordinary cases from the standardized historical data of meteorological observations during ordinary times and the standardized historical data of meteorological forecasts during ordinary times; forming strong wind cases from the standardized historical data of meteorological observations during strong winds and the standardized historical data of meteorological forecasts during strong winds.

[0039] Step S133: In the high wind cases, based on the standardized historical data of the meteorological forecast and the convective parameter forecast therein, the Pearson correlation analysis is used to obtain the highly correlated convective parameters, and based on the highly correlated convective parameters, the k-means clustering method is used to further screen out the cold air systematic high wind cases and the convective high wind cases.

[0040] The implementation method of the above step S133 is, for example, to establish the correlation between the convective effective potential energy, convective inhibition energy, K index, helicity and other convective parameters CV and wind speed U using the Pearson correlation analysis method in a high wind case. , select high-correlation convection parameters; for the selected high-correlation convection parameters, use the k-means clustering method to further screen the strong wind cases, and obtain cold air systemic strong wind cases and convective strong wind cases, including: organizing the highly correlated convection parameters, historical data of meteorological elements and other information, and forming a training data set that meets the k-means input format; training the k-means model based on the training data set that meets the k-means input format, and determining the number of centers through sensitivity experiments and using it as the number of case types; based on the results of the sensitivity experiments, screen out cold air systemic strong wind cases and convective strong wind cases. As an optional implementation of the above step S140, the above implementation of establishing a wind speed measurement classification correction model may include: Step S141: Produce feature tensors and label tensors based on the standardized historical data of weather forecasts and the standardized historical data of weather observations.

[0041] The implementation method of the above step S141 is, for example: based on the wind speed measurement standardized historical data and the wind speed observation standardized historical data, the wind speed measurement error standardized historical data is calculated; based on the wind speed measurement standardized historical data, the wind speed observation standardized historical data, and the wind speed measurement error standardized historical data of the same forecast batch (08:00 batch, 20:00 batch, etc.) in the past several days (-3d, -4d, -5d, etc.), the feature X is constructed. di ; where X di The dimension is R*C, the row R represents different forecast time steps, and the column C represents different meteorological elements (forecast, convection parameters, observation, wind speed measurement error); based on the feature X di , concatenate the features of the past few days to form a feature X with a dimension of R×(nC) dn , where n represents the number of days in the past; based on the X dn , the input features of the machine learning model are obtained by using a scaling method (normalization method, standardization method, etc.); the label feature is the error of the historical wind speed forecast of the current batch; the number of days n is adjusted, and a sensitivity experiment is performed to obtain the optimal value of n, and thereby the input features X and label features Y of the machine learning model are constructed.

[0042] Step S142: For the common case, based on the feature tensor and the label tensor, the LSTM model is trained to obtain the wind speed measurement correction model A.

[0043] An implementation example of the above-mentioned step S142 is as follows: in a common case, a feature tensor Xa is constructed using the standardized historical data of forecasts of conventional meteorological elements (wind speed, temperature, humidity, air pressure, etc.) of the same forecast batch in the past few days, the standardized historical data of meteorological observations of wind speed, and the standardized historical data of wind speed measurement errors; a label tensor Ya is constructed using the standardized historical data of wind speed measurement errors of the target forecast batch; (feature tensor Xa, label tensor Ya) is divided into a training set and a test set; the training set and the test set are input into the LSTM long short-term memory neural network model for training and the model parameters are adjusted to obtain a wind speed measurement error prediction model A; a preliminary correction value of the wind speed measurement standardized historical data is calculated using the wind speed error prediction value and the wind speed measurement standardized historical data; the preliminary correction value of the wind speed measurement standardized historical data is checked based on the wind speed observation standardized historical data, and a wind speed measurement error threshold is set according to the check result, and this process is repeated to obtain the optimal threshold of the wind speed measurement error, which is used to control the range of the wind speed measurement error prediction value; the above constitutes a wind speed measurement correction model A. The specific process is as follows: Among them, Xa is a feature tensor constructed with the standardized historical data of regular meteorological element forecasts, the standardized historical data of wind speed meteorological observations, and the standardized historical data of wind speed measurement errors for ordinary cases; Ya is a label tensor constructed with the standardized historical data of wind speed measurement errors for target forecast batches for ordinary cases; Ta is the wind speed measurement error threshold for ordinary cases; Ya predict is the wind speed measurement error prediction value for a common case, WS orig Standardized data for wind speed measurement, WS corr is the wind speed measurement standard data correction value, Ya′ predict It is the predicted value of wind speed measurement error after threshold control for a common case.

[0044] Step S143: For the case of systematic strong winds caused by cold air, based on the feature tensor and the label tensor, LSTM is trained to obtain wind speed correction model 1, LSTM and XGboost are trained to obtain wind speed correction model 2, and the Stacking method is further used to obtain wind speed measurement correction model B.

[0045] An example of an implementation method of the above step S143 is as follows: in an individual case of systematic strong winds caused by cold air, a feature tensor Xb is constructed using the standardized historical data of forecasts of conventional meteorological elements (wind speed, temperature, humidity, air pressure, etc.) of the same forecast batch in the past few days, the standardized historical data of meteorological observations of wind speed, and the standardized historical data of wind speed measurement errors; a label tensor Yb is constructed using the standardized historical data of wind speed measurement errors of the target forecast batch; (feature tensor Xb, label tensor Yb) is divided into a training set and a test set; the training set and the test set are input into an LSTM long short-term memory neural network model for training and the model parameters are adjusted to obtain a wind speed measurement error prediction model B1; the training set and the test set are input into the LSTM long short-term memory neural network model, and the input result Ob_lstm and the label tensor Yb of the LSTM model constitute a new input data access extreme gradient boosting model XGBoost, and the LSTM and XGBoost model and adjust model parameters to obtain wind speed measurement error prediction model B2; wind speed measurement error prediction model B1 and wind speed measurement error prediction model B2 are used as Stacking first-level learners, and their outputs are used as new feature inputs to train Stacking second-level learners (linear models or other machine learning models). Wind speed measurement error prediction model B is obtained by the above Stacking first-level learners and Stacking second-level learners; using wind speed error prediction values ​​and wind speed measurement standardized historical data, the preliminary correction value of wind speed measurement standardized historical data is calculated; based on wind speed observation standardized historical data, the preliminary correction value of wind speed measurement standardized historical data is tested, and the wind speed measurement error threshold is set according to the test results. This process is repeated to obtain the optimal threshold of wind speed measurement error, which is used to control the range of wind speed error prediction values; the above constitutes wind speed measurement correction model B. The specific process is: Among them, Xb is a feature tensor constructed with standardized historical data of conventional meteorological element forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for individual cases of systematic cold air winds; Yb is a label tensor constructed with standardized historical data of wind speed measurement errors for target forecast batches of individual cases of systematic cold air winds; Tb is the wind speed measurement error threshold for individual cases of systematic cold air winds; Yb predict is the wind speed measurement error prediction value of the cold air systematic strong wind case, WS orig Standardized data for wind speed measurement, WS corr is the wind speed measurement standard data correction value, Yb′ predict It is the predicted value of wind speed measurement error after threshold control for a case of systematic strong winds caused by cold air.

[0046] Step S144: For the convective strong wind case, based on the feature tensor and the label tensor, the LSTM is trained to obtain the wind speed correction model 3, the LSTM and XGboost models are trained to obtain the wind speed correction model 4, and the Stacking method is further used to obtain the wind speed measurement correction model C.

[0047] The implementation method of the above step S143 is, for example: in the case of convective gale, the feature tensor Xc is constructed with the standardized historical data of the forecast of conventional meteorological elements (wind speed, temperature, humidity, air pressure, etc.) of the same forecast batch in the past few days, the standardized historical data of the forecast of convective parameters, the standardized historical data of meteorological observations of wind speed, and the standardized historical data of wind speed measurement errors; the label tensor Yc is constructed with the standardized historical data of wind speed measurement errors of the target forecast batch; (feature tensor Xc, label tensor Yc) is divided into a training set and a test set; the training set and the test set are input into the LSTM long short-term memory neural network model for training and the model parameters are adjusted to obtain a wind speed measurement error prediction model C1; the training set and the test set are input into the LSTM long short-term memory neural network model, and the input result Oc_lstm and the label tensor Yc of the LSTM model constitute a new input data access extreme gradient boosting model XGBoost, Train the LSTM and XGBoost models and adjust the model parameters to obtain the wind speed measurement error prediction model C2; use the wind speed measurement error prediction model C1 and the wind speed measurement error prediction model C2 as Stacking first-level learners, and use their outputs as new feature inputs to train the Stacking second-level learner (linear model or other machine learning model), and obtain the wind speed measurement error prediction model C from the above Stacking first-level learner and Stacking second-level learner; use the wind speed error prediction value and the wind speed measurement standardized historical data to calculate the preliminary correction value of the wind speed measurement standardized historical data; test the preliminary correction value of the wind speed measurement standardized historical data based on the wind speed observation standardized historical data, and set the wind speed measurement error threshold according to the test results, and repeat this process to obtain the optimal threshold of the wind speed measurement error to control the range of the wind speed error prediction value; the above constitutes the wind speed measurement correction model C. The specific process is:

[0048] Among them, Xc is a feature tensor constructed with standardized historical data of conventional meteorological element forecasts, standardized historical data of convective parameter forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for convective strong wind cases; Yc is a label tensor constructed with standardized historical data of wind speed measurement errors for target forecast batches of convective strong wind cases; Tc is the wind speed measurement error threshold for convective strong wind cases; Yc predict is the wind speed measurement error prediction value for a convective gale case, WS origStandardized data for wind speed measurement, WS corr Yc′ is the wind speed measurement standard data correction value, predict It is the predicted value of wind speed measurement error after threshold control for a case of convective strong wind.

[0049] As an optional implementation of the above step S150, the above implementation of correcting the real-time wind speed measurement may include: Step S151: Obtain real-time wind speed measurement and process it into standardized real-time data for weather forecast.

[0050] The implementation method of the above-mentioned step S151 is, for example: obtaining real-time forecast data from the China Meteorological Administration Mesoscale Numerical Weather Prediction Operational System (CMA-MESO); extracting forecast data of conventional meteorological elements and highly correlated convective parameters such as wind speed, temperature, humidity, and air pressure at the target site; and generating standardized real-time data of meteorological forecast according to a pre-made format.

[0051] Step S152: Based on thresholds such as wind speed, duration of strong winds, and convection parameters, determine the category of the strong wind process described in the real-time forecast.

[0052] An implementation example of the above step S152 is: based on thresholds such as wind speed, duration, and convection parameters, the wind speed, gale duration, and convection parameters in the standardized real-time data of the meteorological forecast are compared to determine the gale process category described in the real-time forecast.

[0053] Step S153: According to the category of the strong wind process, a suitable wind speed measurement correction model is matched from the wind speed measurement classification correction model to correct the wind speed measurement.

[0054] An implementation method of the above-mentioned step S153 is, for example: according to the above-determined category of strong wind processes, a suitable wind speed measurement correction model is matched from the wind speed measurement classification correction model (wind speed measurement correction model A, wind speed measurement correction model B, wind speed measurement correction model C), and a corresponding input feature tensor is established according to the selected wind speed measurement correction model. After the feature tensor is input into the model, a corrected wind speed measurement is generated.

[0055] like Figure 2 As shown, the embodiment of the present application provides a ground wind speed correction device 200 that takes into account the complex convective environment characteristics, including: The data acquisition module 210 is used to acquire original historical data such as historical weather forecasts, historical weather observations, historical warning information, and historical disaster-causing events of the target site.

[0056] The data preprocessing module 220 is used to perform preprocessing such as cleaning, quality control, default value filling or discarding, and standardization on the various historical data of the target site obtained above to generate standardized historical data.

[0057] The strong wind process classification module 230 is used to judge the category of the strong wind process based on thresholds such as wind speed and duration, based on historical meteorological observations, historical warning information, historical disaster-causing events and other data, and to screen out ordinary cases and strong wind cases. Among the strong wind cases, the k-means method is further used to screen out cold air systematic strong wind cases and convective strong wind cases based on convection parameters (CAPE, etc.).

[0058] The wind speed measurement classification correction model training module 240 is used to train and debug LSTM, XGboost, and Stacking models according to the category of strong wind processes, based on historical meteorological forecast data and historical meteorological observation data, to generate a wind speed measurement classification correction model.

[0059] The real-time wind speed measurement and correction module 250 is used to determine the type of the strong wind process according to thresholds such as wind speed and duration and convection parameters, and select an appropriate wind speed correction model to correct the real-time wind speed measurement accordingly.

[0060] As an optional implementation of the above device, the original historical data includes: original historical data of weather forecasts, original historical data of weather observations, original historical data of early warning information, and original historical data of disaster-causing events; the data acquisition module includes: The original historical weather forecast data acquisition module is obtained from the China Meteorological Administration Mesoscale Numerical Weather Prediction Operational System (CMA-MESO).

[0061] The module for acquiring the original historical data of meteorological observations is obtained from the national ground meteorological observation station network of the China Meteorological Administration.

[0062] The original historical data acquisition module of the warning information is obtained from the national meteorological warning information database and industry warning information database of the China Meteorological Administration.

[0063] The original historical data acquisition module of disaster-causing events is obtained from government departments and industry records such as civil affairs, disaster relief, and meteorology at all levels.

[0064] As an optional implementation of the above device, the data preprocessing module includes: The weather forecast data preprocessing module is used to extract the original historical forecasts of various meteorological elements at the target point, fill in the missing forecast times, and generate standardized historical data of weather forecasts in a pre-made format; The meteorological observation data preprocessing module is used to perform quality control on the original historical data of meteorological observations, discard observation data with more consecutive missing measurements, interpolate and fill observation data with less consecutive missing measurements, extract the required meteorological elements and generate standardized historical data of meteorological observations in a pre-made format; The warning information preprocessing module extracts the original historical data of warning information related to the target site, removes duplicate warnings from the national, provincial, municipal and county multi-level warning information and industry warning information, and generates standardized historical data of meteorological warning information; The disaster event preprocessing module extracts the original historical data of disaster information related to the target site, eliminates duplicate records in the national, provincial, municipal and county multi-level disaster event records and industry disaster records, and generates standardized historical data of disaster events.

[0065] As an optional implementation of the above device, the strong wind process classification module includes:

[0066] The threshold determination module is used to establish thresholds such as wind speed and gale duration by using standardized historical data of early warning information, standardized historical data of disaster-causing events, and standardized historical data of meteorological observations, in combination with national standards and industry standards;

[0067] The preliminary classification module is used to determine the category of strong wind processes and identify common cases and strong wind cases by using thresholds such as wind speed and duration of strong winds and standardized historical data of meteorological observations.

[0068] The refined classification module is used to use the convective parameter data in the standardized historical data of the meteorological forecast to obtain highly correlated convective parameters by Pearson correlation analysis within the high wind cases, and further screen out cold air systematic strong wind cases and convective strong wind cases based on the highly correlated convective parameters by using the k-means clustering method.

[0069] As an optional implementation of the above device, the wind speed measurement classification correction model training module includes: The general wind speed measurement and correction model training module is used for general cases. Based on the standardized historical data of meteorological forecasts and meteorological observations of conventional meteorological elements such as wind speed, temperature, humidity, and air pressure, the LSTM model is trained to obtain the wind speed correction model A. The module for training the systematic gale wind speed calculation and correction model is used for the case of systematic gale in cold air. Based on the standardized historical data of meteorological forecast and meteorological observation of conventional meteorological elements such as wind speed, temperature, humidity, and air pressure, LSTM is trained to obtain wind speed correction model 1. LSTM and XGboost are trained to obtain wind speed correction model 2. The stacking method is further used to obtain the final wind speed correction model B. The convective gale wind speed estimation and correction model training module is used for convective gale cases. Based on the standardized historical data of meteorological forecasts and meteorological observations of conventional meteorological parameters such as wind speed, temperature, humidity, and air pressure and highly correlated convective parameters, the LSTM is trained to obtain the wind speed correction model 3. The LSTM and XGboost models are trained to obtain the wind speed correction model 4. The Stacking method is further used to obtain the final wind speed correction model C.

[0070] As an optional implementation of the above device, the real-time wind speed measurement and correction module includes: Real-time data standardization module, used to obtain real-time wind speed measurement and process it into standardized real-time data for weather forecast; Real-time high wind process classification module, used to determine the high wind process category using thresholds such as wind speed, duration, and convection parameters; The wind speed measurement classification correction module is used to select a suitable wind speed measurement correction model from the wind speed measurement classification correction models (wind speed measurement correction model A, wind speed measurement correction model B, wind speed measurement correction model C) to correct the wind speed measurement according to the category of the strong wind process.

[0071] It should be understood that the device corresponds to the above-mentioned wind farm forecast data correction method embodiment and can execute the various steps involved in the above-mentioned method embodiment. The specific functions of the device can be found in the above description, and the detailed description is appropriately omitted here. The device includes at least one software function module that can be stored in a memory in the form of software or firmware or solidified in the operating system (OS) of the device.

[0072] The following are examples of application scenarios to which the method claimed for protection in this application is applicable: energy and power construction and operation, transportation construction and operation, engineering construction, etc. Specifically, in the application scenario of energy and power construction and operation, wind farm operators can use the revised wind farm forecast data to more accurately predict future wind speed and wind direction, reasonably arrange wind turbine start and stop, maintenance plans, and revise power output, which will help wind power companies participate in power market bidding and improve wind power utilization and economic benefits. In the application scenario of transportation construction and operation, railway or light rail operating departments can use the revised wind farm forecast data to issue wind speed warnings in time to improve operational efficiency and safety. In the application scenario of engineering construction, wind farm forecasts are crucial for safe construction and safe hoisting. The construction party can use the revised wind farm forecast data to issue wind speed warnings in advance, adjust construction plans, and ensure construction safety. In addition, the meteorological department can use refined wind farm forecast data and revised wind farm forecast data to predict and warn natural disasters caused by strong winds on the local ground in advance.

[0073] The correction effect was tested on multiple strong wind cases in Beijing from January to October 2023. The comprehensive test results of multiple meteorological stations showed that after correction by the ground wind speed measurement correction model, the average absolute error of the wind speed forecast (1.64m / s) was reduced by 0.58m / s compared with the original forecast (2.22m / s); the average error (0.02m / s) was reduced by 1.39m / s compared with the original forecast (1.41m / s), and the optimization effect of the wind speed forecast was obvious.

[0074] like Figure 3 , 4 As shown in the figure, the case test results of the wind speed measurement and correction at a certain meteorological station in mid-January and mid-July 2023 show that after correction by the ground wind speed measurement and correction model, the average wind speed forecast at each moment (blue line) is significantly lower than the original forecast (gray line), and the time series is closer to actual observations (gray shadow).

Claims

1. A method for calculating and correcting ground wind speed considering complex convective environment characteristics, characterized in that: include: Obtain the original historical data of the target site, including weather forecasts, weather observations, warning information, and disaster-causing events; Preprocessing the original historical data of the target site obtained above to generate standardized historical data; According to the thresholds including wind speed, duration, and convection parameters, the category of the gale process is determined based on standardized historical data, and ordinary cases, cold air systemic gale cases, and convective gale cases are screened out; Based on standardized historical data and according to different types of strong wind processes, LSTM, XGboost, and Stacking models are trained and debugged to generate a wind speed measurement classification correction model; For real-time wind speed measurement, the category of strong wind process is judged by combining the thresholds including wind speed, duration and convection parameters, and the appropriate wind speed measurement correction model is automatically matched in the wind speed measurement classification correction model to correct the wind speed measurement results.

2. The method according to claim 1, characterized in that: The generation of the wind speed measurement correction model A for a common case is as follows: The generation of wind speed calculation correction model B for cold air systematic gale cases is as follows: The generation of wind speed calculation correction model C for convective gale cases is as follows: Among them, Xa and Xb are feature tensors constructed with the standardized historical data of conventional meteorological element forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for common cases and cold air systematic gale cases, respectively; Xc is a feature tensor constructed with the standardized historical data of conventional meteorological element forecasts, standardized historical data of convective parameter forecasts, standardized historical data of meteorological observations of wind speed, and standardized historical data of wind speed measurement errors for convective gale cases; Y is a label tensor constructed with the standardized historical data of wind speed measurement errors of the target forecast batch; T is the wind speed measurement error threshold; Y is the label tensor constructed with the standardized historical data of wind speed measurement errors of the target forecast batch; predict is the wind speed measurement error prediction value, WS orig Standardized data for wind speed measurement, WS corr Y′ is the corrected value of wind speed measurement standard data, predict is the predicted value of wind speed measurement error after threshold control.

3. The method according to claim 1, characterized in that The specific process of the Stacking model includes: Based on the standardized historical data of wind speed measurement and the standardized historical data of wind speed observation, the standardized historical data of wind speed measurement error is calculated; The feature tensor is constructed using the standardized historical data of meteorological forecasts of conventional meteorological elements including wind speed, temperature, humidity, and air pressure, as well as the standardized historical data of meteorological observations of wind speed, and the label tensor is constructed using the standardized historical data of wind speed errors; the feature tensor and the label tensor are divided into a training set and a test set; The training set and the test set are input into the artificial intelligence model for training and the model parameters are adjusted to obtain multiple wind speed measurement error prediction models as a Stacking first-level learner; Input the outputs of the multiple wind speed measurement error prediction models as new features into the secondary learner, train the Stacking secondary learner and adjust the parameters; A wind speed measurement error prediction model is constructed based on the Stacking first-level learner and the Stacking second-level learner.

4. The method according to claim 3, characterized in that: The feature tensor is constructed using the standardized historical data of weather forecasts and the standardized historical data of wind speed meteorological observations, including: Based on the standardized historical data of wind speed measurement and the standardized historical data of wind speed observation, the standardized historical data of wind speed measurement error is calculated; Based on the standardized historical data of wind speed calculation, wind speed observation, and wind speed calculation error of the same forecast batch in the past few days, feature X is constructed. di ; where X di The dimension is R×C, the rows R represent different forecast time steps, and the columns C represent different meteorological elements; Based on the feature X di , concatenate the features of the past few days to form a feature X with a dimension of R×(nC) dn , where n represents the number of days in the past; based on the X dn , the input features of the machine learning model are obtained using the scaling method; the label feature is the error of the historical wind speed forecast of the current batch; The number of days n is adjusted, and a sensitivity experiment is performed to obtain the optimal value of n, and thereby the input feature X and label feature Y of the machine learning model are constructed.

5. The method according to claim 2, characterized in that: Generate wind speed estimation correction model, including: Calculate the wind speed measurement error prediction value by using the wind speed measurement error prediction model; Using the wind speed measurement error prediction value and the wind speed measurement standardized historical data, a preliminary correction value of the wind speed measurement standardized historical data is calculated; Based on the standardized historical data of wind speed observation, the preliminary correction value of the standardized historical data of wind speed measurement is checked, and the wind speed measurement error threshold is set according to the test result. This process is repeated to obtain the optimal threshold of wind speed measurement error, so as to control the range of wind speed measurement error prediction value; Based on the wind speed calculation error prediction model and the wind speed calculation error optimal threshold, a wind speed calculation correction model is constructed.

6. The method according to claim 1, characterized in that The categories of gale processes are determined based on standardized historical data including meteorological observations, warning information, and disaster-causing events, including: Based on standardized historical data including warning information and disaster-causing events, combined with national and industry standards, thresholds for wind speed, duration of high winds, and convection parameters are established; Based on standardized historical data of meteorological observations, combined with thresholds including wind speed, duration of strong winds, and convection parameters, the category of strong wind processes is determined.

7. The method according to claim 6, characterized in that The categories of the strong wind process include: Based on the standardized historical data of meteorological observations, combined with thresholds including wind speed and duration of strong winds, ordinary cases and strong wind cases are determined; Based on the standardized historical data of meteorological forecasts, convective parameters are used in strong wind cases, and Pearson correlation analysis is applied to obtain highly correlated convective parameters. Based on the highly correlated convective parameters, the k-means clustering method is further used to screen out cold air systematic strong wind cases and convective strong wind cases.

8. The method according to claim 7, characterized in that Based on the highly correlated convective parameters, the k-means clustering method was used to screen out cold air systemic gale cases and convective gale cases, including: The data including highly relevant convective parameters and historical data of meteorological elements are collated and formed into a training data set that meets the k-means input format; The k-means model is trained based on the training data set that meets the k-means input format, and the number of centers is determined through sensitivity experiments and used as the number of case types; Based on the results of the sensitivity experiment, cases of cold air systemic strong winds and convective strong winds were screened out.

9. The method according to claim 1, characterized in that: For real-time wind speed measurement, the wind speed, duration, and convection parameter thresholds are combined to determine the category of the strong wind process, and the appropriate wind speed measurement correction model is automatically matched in the wind speed measurement classification correction model to correct the wind speed measurement, including: Obtain real-time wind speed measurement and preprocess it into standardized real-time data for weather forecast; Determine the category of the strong wind process based on thresholds including wind speed, duration, and convective parameters; According to the category of strong wind process, the wind speed measurement is corrected by matching the appropriate wind speed measurement correction model from the wind speed measurement classification correction model.

10. A surface wind speed correction device considering complex convective environment characteristics, characterized in that: include: Data acquisition module, which is used for the original historical data of target sites, including weather forecasts, weather observations, warning information, and disaster-causing events; A data preprocessing module is used to preprocess the original historical data of the target site obtained above to generate standardized historical data; The gale process classification module is used to determine the gale process category based on thresholds including wind speed, duration, and convection parameters, and based on standardized historical data including meteorological observations, warning information, and disaster-causing events, and to screen out common cases, cold air systemic gale cases, and convective gale cases; The wind speed estimation correction model training module is used to train and debug LSTM, XGboost, and Stacking models according to different categories of strong wind processes based on standardized historical data including weather forecasts and meteorological observations, and generate a wind speed estimation classification correction model; The real-time wind speed measurement and correction module is used to judge the category of the strong wind process based on the thresholds including wind speed, duration, and convection parameters, and select the appropriate wind speed correction model to correct the real-time measured wind speed accordingly.

Images