Method, system, medium and equipment for controlling quality of anemometer data of anemometer tower

Abstract

The invention belongs to the field of anemometer data processing, and provides an anemometer tower anemometer data quality control method and system, a medium and equipment. The anemometer tower wind measurement data quality control method comprises the steps that current wind measurement data are obtained, the format of the current wind measurement data is recognized in a self-adaptive mode, matching analysis is carried out, and the analyzed current wind measurement data are packaged according to a set unified format; according to the current wind measurement data packaged in the set unified format and a pre-constructed quality evaluation model, performing quality evaluation on the current wind measurement data packaged in the set unified format; according to an evaluation result of the quality evaluation model and a pre-trained machine learning model, performing corresponding wind measurement data interpolation; and verifying the interpolated wind measurement data through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets a set application requirement.

Description

Technical Field

[0001] The present invention belongs to the field of wind measurement data processing, and in particular relates to a method, system, medium and equipment for quality control of wind measurement data of a wind measurement tower. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In the actual wind measurement and collection process, wind tower wind data may be intermittent or erroneous due to factors such as environmental conditions and equipment failures. In addition, wind tower wind data may be affected by network delays, packet loss and other issues during transmission, further affecting the integrity and timeliness of the data. Inaccurate, incomplete or outlier wind resource data may lead to misleading results and decisions. Although there are some data quality inspection and cleaning methods in the prior art, they are not suitable for the full process monitoring of the quality of wind tower wind data, and cannot fully meet the wind energy industry's demand for high-quality data, and thus cannot provide an accurate data basis for wind farm layout and power generation efficiency optimization. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method, system, medium and equipment for quality control of wind measurement data of a wind measurement tower, which performs quality control throughout the entire process from data collection and processing to quality assessment, improvement and confirmation to ensure the quality and reliability of wind measurement data.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: A first aspect of the present invention provides a method for controlling the quality of wind measurement data from a wind tower.

[0006] In one or more embodiments, a method for controlling the quality of wind data measured by a wind tower is provided, comprising: Obtain the current wind measurement data and adaptively identify its format and match the analysis, and encapsulate the parsed current wind measurement data according to the set unified format; According to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model, quality assessment is performed on the current wind measurement data packaged in a set unified format; According to the evaluation results of the quality assessment model and the pre-trained machine learning model, the corresponding wind measurement data is interpolated; The interpolated wind measurement data is verified through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

[0007] As an implementation method, for wind measurement data in NRG format, the password of the wind measurement data is first checked, and the legitimacy of the wind measurement data and the credibility of the source are confirmed by comparing it with a preset password library; when the password matches successfully, further analysis is performed.

[0008] As an implementation mode, the wind measurement data in non-NRG format is processed according to preset parsing rules.

[0009] As an implementation mode, the quality assessment model includes a rule verification model based on a priori rule setting and a data assessment model constructed based on historical data; The rule verification model based on prior rule setting is used to perform a primary detection on the input wind measurement data and initially mark the abnormal data therein; the data evaluation model constructed based on historical data is used to perform a secondary quality evaluation on the initially marked abnormal data to determine the type of abnormal data.

[0010] As an implementation method, when the evaluation result of the quality evaluation model is that the wind measurement data at different heights on the same tower is missing, a pre-trained machine learning model is used to perform same-tower interpolation.

[0011] As an implementation method, when the evaluation result of the quality assessment model is whether the wind direction and / or wind speed in the wind measurement data of the target wind measurement tower is missing, in the process of interpolation between towers using a pre-trained machine learning model, the wind direction of the wind measurement data is interpolated between towers first, and then the wind speed is interpolated between towers.

[0012] As an implementation mode, before verifying the interpolated wind measurement data through a preset verification mechanism, the method further includes: The interpolated wind measurement data are further corrected for the representative year.

[0013] A second aspect of the present invention provides a wind tower wind measurement data quality control system.

[0014] In one or more embodiments, a wind tower wind data quality control system includes: A unified format encapsulation module is used to obtain the current wind measurement data and adaptively identify its format and match the analysis, and encapsulate the parsed current wind measurement data according to the set unified format; A quality assessment module, which is used to perform quality assessment on the current wind measurement data packaged in a set unified format according to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model; A data interpolation module is used to interpolate corresponding wind measurement data according to the evaluation results of the quality evaluation model and the pre-trained machine learning model; The data verification module is used to verify the interpolated wind measurement data through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

[0015] A third aspect of the present invention provides a computer-readable storage medium.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the above-mentioned method for controlling the quality of wind measurement data from a wind tower.

[0017] A fourth aspect of the present invention provides an electronic device.

[0018] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the above-mentioned method for quality control of wind measurement data from a wind tower are implemented.

[0019] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention utilizes the current wind measurement data packaged in a unified format and a pre-built quality assessment model to perform quality assessment on the current wind measurement data packaged in a unified format. Through the automated data processing, quality assessment and confirmation mechanism, the work efficiency is greatly improved, the manpower and time costs are reduced, and the quality control of the entire process from data collection, processing to quality assessment, improvement and confirmation is achieved. At the same time, data quality problems can be discovered and resolved in a timely manner, thereby avoiding additional costs and losses caused by poor data quality.

[0020] (2) The quality assessment model of the present invention is composed of a rule verification model based on a priori rules and a data assessment model constructed based on historical data. The rule verification model based on a priori rules is used to perform a primary detection on the input wind measurement data, and the data assessment model constructed based on historical data is used to perform a secondary quality assessment on the initially marked abnormal data. The two-detection mechanism is used to timely discover and solve data quality problems, thereby improving the integrity, accuracy and credibility of the data, thereby improving the quality and reliability of the data.

[0021] (3) The high-quality wind measurement data obtained by the present invention will provide reliable data support for the planning, design, construction and operation of the wind energy industry, help optimize the layout of wind farms, improve power generation efficiency, and thus support the development and application of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0023] Figure 1 1 is a flow chart of a method for controlling wind tower wind data quality according to an embodiment of the present invention; Figure 2 2 is a schematic diagram of the structure of a wind tower wind data quality control system according to an embodiment of the present invention; Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0025] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0026] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0027] Figure 1 is a flow chart of a method for controlling wind tower wind data quality in an embodiment of the present invention. Figure 1 The wind tower wind measurement data quality control method in the embodiment shown may include: S101, obtaining current wind measurement data and adaptively identifying its format and matching analysis, and packaging the parsed current wind measurement data according to a set unified format; S102, performing quality assessment on the current wind measurement data packaged in a set unified format according to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model; S103, performing corresponding wind measurement data interpolation according to the evaluation result of the quality evaluation model and the pre-trained machine learning model; S104, verifying the interpolated wind measurement data through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

[0028] This embodiment uses the current wind measurement data packaged in a unified format and a pre-built quality assessment model to perform quality assessment on the current wind measurement data packaged in a unified format. Through automated data processing, quality assessment and confirmation mechanisms, work efficiency is greatly improved, manpower and time costs are reduced, and quality control of the entire process from data collection, processing to quality assessment, improvement and confirmation is achieved. At the same time, data quality issues can be discovered and resolved in a timely manner to avoid additional costs and losses caused by poor data quality.

[0029] In step S101, in order to be compatible with multiple wind measurement data formats and ensure that data from different sources can be effectively processed and parsed, for data in NRG (NRG Systems) format, the data password must be checked first. Password verification is a key step to ensure data security and integrity. By comparing with the preset password library, the legitimacy of the data and the credibility of the source are confirmed. Only after the password matches successfully, further parsing and processing are performed.

[0030] For data that is not in NRG format, the preset parsing rules are used for processing. The preset parsing rules include identification of data in different formats, parsing rules, data type definition, etc. It is necessary to parse the input data header information and each column of information to extract necessary information, such as wind measurement height, sensor type, data type, etc. The parsing process needs to extract the corresponding information according to the characteristics of different data formats and preset rules, and perform normalization processing.

[0031] First, the data type is automatically identified based on the format of the input file (such as file extension, file header information, etc.). For NRG data, the data source is confirmed by the file extension (.rld, .rwd). For NRG data, the password information is obtained from the input and compared with the system's preset password library. After the comparison is passed, the data parsing phase is entered; otherwise, an error message is returned, indicating that the password does not match.

[0032] Preset parsing rules for different data formats, such as the number of columns, field names, units, etc. The parsing process uses regular expressions or special parsing algorithms to extract and identify information column by column. During the parsing process, the installation information of the sensor will be obtained from the file header information, which usually includes the installation height, direction, type of the sensor, and the longitude and latitude of the wind tower, altitude information, etc. The parsing process mainly relies on the data column header information to identify the wind measurement height, sensor type and data type. Confirm the sensor type by matching the preset sensor library. For example, the header of a column of data is "WS150m-avg", where WS indicates that the sensor type is a wind speed sensor (wind speed), the installation height is 150 meters, and avg indicates that the data is the average data of 10 minutes. For some data, when the system cannot parse it correctly, manual identification is required to mark the data height, sensor type, data type and other information.

[0033] After parsing, all kinds of information will be reorganized and encapsulated in the predefined JSON format. The JSON object includes key fields such as basic data information (such as timestamp, station number, etc.), wind measurement height, sensor type, and measurement value. The standardized JSON data format output by the module has good compatibility and scalability, which is convenient for subsequent data processing and system integration. The module also provides detailed parsing logs to record the key steps and results in the data parsing and conversion process for easy traceability and debugging. These log information includes data source information, parsing steps, conversion results, and potential errors or exception prompts to ensure that problems can be quickly located and resolved when they occur during data processing.

[0034] In step S102, the quality assessment model includes a rule verification model based on prior rule settings and a data assessment model built based on historical data; The rule verification model based on prior rule setting is used to perform a primary detection on the input wind measurement data and initially mark the abnormal data therein; the data evaluation model constructed based on historical data is used to perform a secondary quality evaluation on the initially marked abnormal data to determine the type of abnormal data.

[0035] In step S102, the quality assessment model is trained according to the wind measurement data quality management knowledge base.

[0036] Among them, a wind measurement data quality management knowledge base is built based on historical data: historical wind measurement data is collected and analyzed, and a quality management knowledge base including quality standards, evaluation indicators, best practices, etc. is established to provide a reference basis for subsequent quality control.

[0037] Historical wind measurement data refers to wind resource-related data collected by wind towers, wind radars, meteorological stations, etc. in history, including conventional observation variables such as wind speed, wind direction, temperature, air pressure, as well as statistical data such as turbulence intensity, wind energy, and wind power density.

[0038] In step S102, the wind measurement data quality standard and evaluation index system in the wind measurement data quality management knowledge base is constructed in the following process: First, based on the reference to the national and industry-related wind data quality standards, the quality assessment indicators, qualification standards and other contents are extracted. And based on actual experience, the indicator system and evaluation rules for data quality assessment are supplemented and improved, and a set of wind data quality knowledge bases containing problems encountered in wind data quality control and rules for evaluation are initially established. For example, for invalid data, based on the 2019 version of the "Technical Specifications for Review, Interpolation and Correction of Meteorological Observation Data in Wind Farms", a reasonable range of wind speed and wind direction can be initially set, such as the reasonable range of wind speed should be between 0 and 75m / s, and the reasonable range of wind direction should be between 0 and 360 degrees. For example, considering the freezing situation, at low temperatures, the wind sensor may freeze, resulting in data distortion. According to experience, if the average wind speed change for three hours is less than 0.2m / s, it can be regarded as frozen data.

[0039] Next, we will further build a quality assessment model based on historical wind measurement data. First, we need to collect and prepare historical wind measurement data and related meteorological data. Historical wind measurement data, which has been processed for quality assessment, contains quality assessment labels, such as invalid data, low-quality data, tower shadow data, frozen data, etc.

[0040] After data preprocessing, feature engineering is performed. Features are extracted from wind measurement data, such as wind speed, wind direction, wind speed difference at different heights, and wind speed standard deviation. Combined with time features, such as year, month, day, hour, season, etc., environmental features can also be extracted, such as geographic location and surrounding terrain. Based on these features, new combined features are generated, such as wind speed change rate and wind direction change rate. All numerical features are standardized to ensure that they are of the same magnitude.

[0041] After the dataset is prepared, it is divided into training set, validation set, and test set. The ratio is 8:1:1 to ensure that each dataset contains data of each quality label. The training set is used to train the random forest model. Grid search is used for hyperparameter tuning to adjust the hyperparameters for optimal performance.

[0042] The following evaluation indicators are selected to evaluate the trained model: Accuracy: the ratio of correctly predicted samples to the total number of samples; Precision: the ratio of correctly predicted positive samples; Recall: the ratio of correctly predicted positive samples to the actual number of positive samples. F1 score: the harmonic mean of precision and recall.

[0043] Predict the validation set and calculate the above evaluation indicators. According to the evaluation results, adjust the model hyperparameters (for example, the number of trees, maximum depth, minimum sample split number, minimum sample leaf number, etc.) to optimize the model performance.

[0044] The final trained evaluation model is saved in the wind measurement data quality evaluation knowledge base for subsequent data evaluation.

[0045] The following example takes the identification and processing of frozen data as an example: Collect wind speed, wind direction, temperature, air pressure and other wind measurement data for a certain area over several years. Mark the frozen data based on the three-hour wind speed change method.

[0046] Next, feature extraction is performed to extract multi-dimensional features from the data, and these features are used to automatically identify freezing phenomena in the machine learning model: Wind speed time series features: Calculate the mean and standard deviation of wind speed within a certain time window (such as 1 hour, 3 hours, 6 hours). The wind speed fluctuations in freezing data are usually very small, so the standard deviation will be lower during freezing. Calculate the autocorrelation of the wind speed time series. In the frozen state, the wind speed has a strong similarity over multiple consecutive time steps, and the autocorrelation will be high.

[0047] Temperature and humidity characteristics: Freezing often occurs in low temperature environments. Although it does not rely on a fixed temperature threshold, the time variation pattern of temperature can provide important information. When the temperature is low and the humidity is high, freezing is more likely.

[0048] Generate interactive variables through feature combination, such as the combined features of temperature change rate and wind speed change rate, the interactive effect of humidity and wind speed, etc. These combined features can help capture the impact of environmental conditions on wind speed data.

[0049] Use support vector machines (SVM) to learn patterns from features to identify frozen data.

[0050] The labeled historical data is divided into training set, validation set, and test set in a ratio of 8:1:1. The training set is used for model training, the validation set is used for parameter adjustment and to avoid model overfitting, and the test set is used to finally evaluate the performance of the model on unseen data.

[0051] The trained model can automatically identify frozen data based on multiple features without the need for fixed wind speed or temperature thresholds.

[0052] The real-time observed wind speed, temperature and other data are input, and features such as the sliding window mean, standard deviation, and autocorrelation coefficient are extracted from them. The trained model is used to analyze the data and output the prediction label.

[0053] For example, suppose that on a certain day, the wind data is: Wind speed: The wind speed changes little over several consecutive hours, the sliding window mean and standard deviation within 1 hour are very low, and the autocorrelation coefficient is high. Temperature: Between -3°C and -1°C, with small fluctuations. Relative humidity: Above 80%, and no significant drop.

[0054] After feature extraction, the data is input into the model, which combines this information to determine that the wind speed data during this period is frozen and marks it as frozen data. Different from simple threshold judgment, the model can more accurately identify the freezing phenomenon based on multi-dimensional information.

[0055] Record the feature combinations of ice data identification (such as wind speed time series characteristics, autocorrelation coefficient, etc.) and their performance in the machine learning model; regularly update the ice identification model in the knowledge base, and save the hyperparameter configuration of each model tuning to ensure the robustness of identification under different meteorological conditions.

[0056] In step S102, the quality assessment model aims to use the assessment indicators and rules in the quality management knowledge base to automatically assess the quality of the input wind measurement data and identify abnormalities in the data. By identifying and assessing different types of quality issues, it can provide users with accurate and comprehensive data quality assessment results, providing strong support for subsequent data processing and decision-making.

[0057] First, we obtain predefined evaluation indicators and rules from the quality management knowledge base. These indicators and rules include standards and methods for evaluating various aspects of wind data quality. For the input wind data, we will perform automated quality evaluation based on these indicators and rules to identify possible anomalies in the data.

[0058] During the quality assessment process, different types of quality issues will be identified and evaluated. These quality issues include but are not limited to extreme values, outliers, missing values, consistency issues, etc. For each type of quality issue, the module will give the corresponding evaluation results based on the evaluation indicators and rules defined in the quality management knowledge base. For example, for extreme value issues, the module will check whether there are values ​​in the data that deviate significantly from other data points. A method based on the rate of change method based on time series and a definition based on business rules can be used. Based on business rule definition: For the physical limits of wind speed and wind direction, business rules can be used to judge significant deviations. For example, the wind speed should usually be between 0 and 75m / s. If it exceeds this range, it can be directly regarded as an extreme value. For example, the wind direction should be between 0 and 360. This method is based on actual physical conditions and is an effective basis for identifying extreme values. It is suitable for data points that are obviously beyond the reasonable range. For wind speed data that does not exceed the physical limit but has drastic changes, the rate of change method in the time series can be used to judge significant deviations: Measure the change in wind speed at adjacent time points. If the rate of change is significantly larger (for example, several times beyond the normal range), the data may be an extreme value. This can be determined by calculating the rate of change within the sliding window:

[0059] When the rate of change of a data point is far beyond the norm (such as multiple times the mean or median), it can be considered an extreme deviation. For missing value problems, the number and distribution of missing values ​​in the data are evaluated; for consistency issues, the module checks the logical consistency between different data points in the data set or between data sets. The process of data quality assessment is as follows: The indicators and rules required for data quality assessment are extracted from the constructed knowledge base. These indicators and rules include the evaluation standards and methods for various aspects of wind measurement data quality: rule verification based on prior rule settings, and data evaluation models built based on historical data.

[0060] The input wind measurement data will first be pre-tested using the inspection rules extracted based on national standards and industry rules, and abnormal data will be marked. After that, based on the data evaluation model, detailed quality assessment and inspection will be carried out again, including assessment and inspection of extreme values, outliers, missing values, and data consistency.

[0061] The screening rules for wind resource data established based on existing standards cover key aspects such as data accuracy, completeness, and consistency. The json format data processed by the input module is screened and filtered based on the screening rules. According to the rules, data items that may have problems are screened out and marked instead of being deleted. The specific contents of rule screening include: support for judging the validity and rationality of wind measurement data, and support for chart output. The wind speed, wind direction, temperature, and air pressure of the original data of the wind tower are inspected, including integrity and rationality tests. The integrity test includes the inspection of quantity and time sequence. The number of data should be equal to the expected number of data records; the time sequence of the data should conform to the expected start and end time, and should be continuous in the middle. The rationality test of the data includes range test, correlation test, and trend test.

[0062] The data after preliminary evaluation is then input into the pre-trained quality evaluation model in the quality evaluation knowledge base for further screening and labeling. The data quality is evaluated using the random forest model. The model will conduct a comprehensive analysis and detection of the data based on the patterns and rules learned during previous training.

[0063] The model will identify extreme values, outliers, and missing values ​​based on the characteristics and attributes of the data, mark them, and calculate the effective completeness rate of the data (effective completeness rate = (total number - missing data - abnormal data) / total number × 100%) to indicate the overall quality of the data.

[0064] Quality assessment result output: The system will output a copy of the data containing abnormal data and labels for backup, and a copy of the data with outliers deleted for the next step.

[0065] For the rule screening part, the system provides detailed evaluation results in the form of a table. In some embodiments, outliers and missing values ​​of the data are displayed in the form of a graph, such as a scatter plot to show the degree of outliers, or a heat map to show missing data.

[0066] In step S103, when the evaluation result of the quality evaluation model is that the wind measurement data at different heights on the same tower is missing, the pre-trained machine learning model is used to perform same-tower interpolation; wherein, The process of interpolating wind speed data at different heights on the same tower is as follows: When there is at least one effective wind speed at the target height at a certain moment, the average of these effective wind speeds is taken as the wind speed value of the target height at that moment; When there are effective wind speeds at at least two heights other than the target height at that moment, the Pearson correlation coefficient between the target height and the channels with effective wind speeds at that moment is calculated respectively, and the effective wind speeds of a set number of channels with the highest correlation are selected for same-tower interpolation of the wind speed data of the target height at that moment; when interpolating at the same tower, the wind shear coefficient at that moment is calculated using the selected effective wind speed, and the wind shear profile is calculated based on the wind shear coefficient at that moment, thereby inferring the wind speed data of the target height at that moment; When there is an effective wind speed at a height other than the target height at that moment, the wind shear profile is calculated using the average wind shear coefficient observed in the known wind resource data set at similar times and dates in a day, and then the wind speed data at the target height at that moment is deduced; The process of interpolating wind direction data / temperature data at different heights on the same tower is as follows: When the number of wind direction data / temperature data at the target height at a certain moment is at least two, the average of these wind direction data / temperature data is calculated as the wind direction data / temperature data at the target height at that moment; When there is only one wind direction data / temperature data at the target height at a certain moment, the wind direction data / temperature data at the target height at that moment are inferred by using the average wind direction deflection / temperature gradient observed at similar times and days in a known wind resource data set.

[0067] In step S103, when the evaluation result of the quality evaluation model is whether the wind direction and / or wind speed in the wind measurement data of the target wind measurement tower are missing, in the process of interpolation between towers using the pre-trained machine learning model, the wind direction of the wind measurement data is interpolated between towers first, and then the wind speed is interpolated between towers; The process of interpolating wind direction from different towers based on the wind measurement data of adjacent towers is as follows: The neighboring wind tower is used as the reference wind tower, and the wind direction data of the target wind tower is divided into different sectors according to the wind direction data of the reference wind tower; the change from the reference wind direction to the target wind direction at each moment in each sector is calculated respectively, and the average offset of each sector is calculated; the interpolation value at the corresponding moment can be obtained by adding the average offset of the sector to the reference wind direction; The process of interpolating wind speed from different towers based on the wind resource data of neighboring wind towers is as follows: The wind speed data of the target wind tower are divided into several sectors according to the interpolated wind direction data of the target wind tower. The neighboring wind towers are used as reference wind towers, and the relationship between the wind speed data of the target wind tower and the reference wind speed of the reference wind tower is fitted sector by sector. Based on the fitted wind speed relationship of each sector, the wind speed data in each sector of the target wind tower are interpolated.

[0068] In S103, the missing parts are filled by using the data from the same tower and different towers to improve data quality. The missing values ​​are intelligently interpolated based on historical data and adjacent data points using machine learning models to improve data integrity. The wind measurement data is dynamically corrected by combining multiple meteorological models and measured data to adapt to complex and changing meteorological conditions.

[0069] In step S103, random forest is selected as the interpolation model, and the model is trained using the prepared feature data. During the training process, the hyperparameters of the random forest model, such as the number of trees, the depth of the trees, and other related parameters, need to be set to ensure the performance and generalization ability of the model. The model training and evaluation here adopt the same method as before.

[0070] The samples containing missing values ​​are used as input to the trained random forest model. The model will predict the possible values ​​of the missing values ​​based on the existing feature data. In the random forest model, each decision tree will give a prediction result, and the final interpolation result can be obtained by weighted averaging the prediction results of multiple decision trees.

[0071] Finally, the interpolation results are evaluated and verified. At this stage, the interpolated data needs to be evaluated to check whether the interpolation effect meets the requirements. Evaluation indicators such as root mean square error (RMSE) and mean absolute error (MAE) can be used to evaluate the accuracy and reliability of the interpolation results. If the interpolation effect is not ideal, it may be necessary to further adjust the model parameters or improve the interpolation strategy.

[0072] Through the above detailed steps and processes, the machine learning model can be effectively used to intelligently interpolate the missing values ​​in the wind measurement data, improve the integrity and availability of the data, and provide more reliable data support for wind energy resource assessment.

[0073] In this step, the main purpose is to improve the data quality through dynamic processing and intelligent interpolation of wind measurement data. The specific methods are as follows: Interpolation using data from the same tower: Extract key features such as wind speed and wind direction at different heights on the wind tower, and analyze the correlation of data at each height. Wind speed and wind direction are usually strongly correlated at different heights on the same tower. Calculate the Pearson correlation coefficient between heights to evaluate this relationship. Extract time features, including season, month, and time period (such as morning, evening, etc.), because wind speed and wind direction usually change periodically over time. Use the random forest model to train the model using the existing complete data. Use the trained model to predict missing values. Input the missing wind speed or wind direction corresponding to the height, time characteristics and other information into the model. The model will predict the missing data values ​​based on the input data. For the missing situations in different time periods or seasons, dynamically adjusted weights can be set according to different features to make the interpolation results closer to the actual situation. For example, the wind speed changes differently at night and during the day, and the model can adjust the interpolation strategy based on time characteristics.

[0074] Interpolation using data from different towers: Interpolation using data from different towers means supplementing the missing data of a certain tower with the wind data from neighboring towers (talls close in space). This method is based on the spatial correlation between different towers, especially in variables such as wind speed and wind direction, to perform intelligent interpolation.

[0075] Extract relevant features from the data of neighboring wind towers for modeling and interpolation. Focus on wind speed, wind direction, and other features that may affect wind speed changes. Calculate the Pearson correlation coefficient to analyze the spatial correlation of wind speed and wind direction between the target wind tower and neighboring wind towers. Usually, wind speeds between wind towers that are close to each other are highly correlated, so wind tower data with strong correlation are selected as interpolation candidates.

[0076] The closer the neighboring masts are to the target mast, the stronger the wind speed correlation is likely to be. Use a weighted analysis to give higher weights to masts that are closer.

[0077] Extract the topographic and environmental features around the wind tower, such as elevation, vegetation, water bodies, etc. These features will affect the wind speed. If the topographic features of neighboring wind towers are similar, the correlation of wind speed changes may be higher.

[0078] Use the random forest regression model for interpolation. The input features include wind speed and direction data of multiple neighboring wind towers, wind sensor height, and auxiliary features such as terrain and environment. Use the complete data of neighboring wind towers for model training. Use the trained model to predict the missing values ​​of the target wind tower. Input the characteristic data (wind speed, wind direction, height, time characteristics, etc.) of the target wind tower and the neighboring wind towers into the model, and the model will predict the missing values ​​of wind speed or wind direction of the target wind tower.

[0079] The model can dynamically adjust weights or interpolation strategies for interpolation situations in different seasons and weather conditions. For example, the wind speed change pattern is different in certain seasons, and the model can combine time characteristics for dynamic prediction.

[0080] In step S104, a data quality confirmation mechanism involving multiple parties is established, including manual review, system verification, etc., to ensure that the data quality meets the application requirements.

[0081] In step S104, before verifying the interpolated wind measurement data by a preset verification mechanism, the method further includes: The interpolated wind measurement data are further corrected for the representative year.

[0082] The representative year correction is the process of adjusting the wind energy assessment data for long-term consistency. This process aims to ensure that short-term observation data can reflect the average characteristics of long-term meteorological conditions, so that the assessment results are more representative and reliable. Especially for the design and operation of wind farms, the data needs to reflect the average wind resource conditions over many years, while short-term wind speed measurement data is often affected by climate anomalies in a specific year and cannot be directly used as a representative of the long-term climate. Therefore, these short-term data must be corrected through the representative year correction method.

[0083] Before making representative year corrections, the required data include: actual observation data from wind towers, long-term meteorological data (meteorological station observation data or reanalysis data such as ERA5 and MERRA2 data can be used).

[0084] Since the wind speed and direction heights of the reanalysis data and the wind tower data are generally different, the wind measurement data must first be extrapolated to a height consistent with the long-term data based on wind shear.

[0085] Next, the wind data and long-term data are spatially matched. When ERA5 or MERRA2 is selected as the long-term data, the Kriging interpolation method is used to interpolate the grid data to the location of the wind tower.

[0086] Next, the short-term wind tower data and long-term meteorological data are jointly divided by time (such as month) and wind direction sector (such as north, east, south, etc.). This will form several "time + sector" data subsets, such as "January-north wind direction sector", "April-south wind direction sector", etc. Through this division, the wind speed characteristics of different time periods and wind directions can be captured, avoiding representative distortion caused by not considering the difference in wind direction.

[0087] Correction factors are calculated for each "time+sector" subset to adjust short-term observation data to be consistent with long-term meteorological conditions. The specific steps are as follows: For wind direction data, the short-term wind direction data average and long-term wind direction average are calculated in the time period and sector respectively. The short-term wind direction average is subtracted from the long-term wind direction average to obtain the wind direction offset factor, and the corresponding offset factor is added to the wind direction data in the sector to complete the correction.

[0088] For wind speed correction, fit y=ax+b sector by sector, and calculate the correlation coefficient R between the reference wind speed and the target wind speed in each sector; Based on the value of the correlation coefficient R and the slope a, it is determined whether to correct the sector; For the sectors that need to be corrected, calculate the difference Δx between the average value of the reference data and the average value of the long sequence reference data, and the correction amount of each sector is aΔx. For the sectors that need to be corrected, use the target data + aΔx to get the corrected representative data.

[0089] In step S104, wind measurement data quality confirmation mechanism: establish a data quality confirmation mechanism involving multiple parties, including manual review, system verification, etc., to ensure that the data quality meets the application requirements.

[0090] The wind data quality confirmation mechanism is an important step to ensure that the data quality meets the application requirements. This mechanism covers data review and verification measures involving multiple parties.

[0091] In terms of manual review, professionals regularly conduct sampling analysis of data to promptly identify and handle anomalies and errors in the data to ensure the accuracy and reliability of the data.

[0092] Verification is another key component of data quality confirmation. Through pre-set rules and algorithms, the system can automatically check and verify data and quickly find anomalies in the data. This includes steps such as data format check, range check, and logical consistency check, which can effectively reduce the workload of manual review and improve the efficiency of data processing.

[0093] The data quality confirmation mechanism requires the participation of multiple parties, including data collectors, data analysts, system administrators, etc. Each participant has the responsibility to ensure the quality of the data and to promptly discover and handle problems in the data. Through the participation of multiple parties, a collaborative mechanism can be formed to effectively improve the efficiency and accuracy of data quality confirmation and ensure the objectivity and reliability of data quality.

[0094] The problems found in the inspection are crucial to the update of the quality control knowledge base. When manual review or system verification finds problems with data quality, these problems should be recorded and analyzed in a timely manner. These problems can involve anomalies in data format, scope, logical consistency, etc.

[0095] Recording these problems in the quality control knowledge base will help to continuously improve the quality control process and rules. By analyzing the root causes of the problems, possible problems in data collection, transmission or processing can be discovered, and corresponding measures can be taken to improve them.

[0096] Figure 2 Schematic diagram of the structure of a wind tower wind data quality control system in an embodiment of the present invention. Figure 1 The wind tower wind data quality control method corresponds to the wind tower wind data quality control method, such as Figure 2 As shown, the wind tower wind measurement data quality control system in this embodiment may include: The unified format encapsulation module 201 is used to obtain the current wind measurement data and adaptively identify its format and perform matching analysis, and encapsulate the parsed current wind measurement data according to a set unified format; A quality assessment module 202, which is used to perform quality assessment on the current wind measurement data packaged in a set unified format according to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model; A data interpolation module 203, which is used to interpolate corresponding wind measurement data according to the evaluation results of the quality evaluation model and the pre-trained machine learning model; The data verification module 204 is used to verify the interpolated wind measurement data through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

[0097] It should be noted here that Figure 2 Each module in the wind tower wind data quality control system in Figure 1 The various steps in the wind tower wind measurement data quality control method in the specification correspond to each other, and the specific implementation process is the same, so it will not be repeated here.

[0098] Reference Figure 3 , a schematic diagram of an electronic device is given. It should be noted that, Figure 3 The electronic device 300 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0099] like Figure 3 As shown, the electronic device 300 includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303. In the RAM 303, various programs and data required for system operation are also stored. The central processing unit 301, the ROM 302 and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0100] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a local area network (LAN) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed, so that a computer program read therefrom is installed into the storage section 308 as needed.

[0101] When the central processing unit 301 in the electronic device of this embodiment executes the program, the following is achieved: Figure 1 The steps in the quality control method of wind tower wind measurement data are shown.

[0102] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, the computer program including a computer program for executing Figure 1 In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 309, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit 301, various functions defined in the apparatus of the present application are executed.

[0103] in, Figure 1 The computer program instructions corresponding to the method shown may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0104] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0105] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for controlling the quality of wind measurement data from a wind tower, characterized in that: include: Obtain the current wind measurement data and adaptively identify its format and match the analysis, and encapsulate the parsed current wind measurement data according to the set unified format; According to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model, quality assessment is performed on the current wind measurement data packaged in a set unified format; According to the evaluation results of the quality assessment model and the pre-trained machine learning model, the corresponding wind measurement data is interpolated; The interpolated wind measurement data is verified through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

2. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: For wind measurement data in NRG format, first check the password of the wind measurement data and confirm the legitimacy of the wind measurement data and the credibility of the source by comparing it with the preset password library; when the password matches successfully, further analysis is performed.

3. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: For wind measurement data not in NRG format, they are processed according to the preset parsing rules.

4. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: The quality assessment model includes a rule verification model based on prior rule settings and a data assessment model built based on historical data; The rule verification model based on prior rule setting is used to perform a primary detection on the input wind measurement data and initially mark the abnormal data therein; the data evaluation model constructed based on historical data is used to perform a secondary quality evaluation on the initially marked abnormal data to determine the type of abnormal data.

5. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: When the evaluation result of the quality assessment model is that the wind measurement data at different heights on the same tower are missing, the pre-trained machine learning model is used to perform same-tower interpolation.

6. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: When the evaluation result of the quality assessment model is whether the wind direction and / or wind speed in the wind measurement data of the target wind measurement tower is missing, in the process of interpolation between different towers using the pre-trained machine learning model, the wind direction of the wind measurement data is interpolated between different towers first, and then the wind speed is interpolated between different towers.

7. The method for controlling the quality of wind data measured by a wind tower according to claim 1, characterized in that: Before verifying the interpolated wind data through the preset verification mechanism, it also includes: The interpolated wind measurement data are further corrected for the representative year.

8. A wind tower wind data quality control system, characterized in that: include: A unified format encapsulation module is used to obtain the current wind measurement data and adaptively identify its format and match the analysis, and encapsulate the parsed current wind measurement data according to the set unified format; A quality assessment module, which is used to perform quality assessment on the current wind measurement data packaged in a set unified format according to the current wind measurement data packaged in a set unified format and a pre-built quality assessment model; A data interpolation module is used to interpolate corresponding wind measurement data according to the evaluation results of the quality evaluation model and the pre-trained machine learning model; The data verification module is used to verify the interpolated wind measurement data through a preset verification mechanism to determine whether the quality of the interpolated wind measurement data meets the set application requirements.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method for quality control of wind measurement data of a wind tower as described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for quality control of wind measurement data of a wind tower according to any one of claims 1 to 7 are implemented.

Images