Accurate correction method for wind resource representative year of mountain wind power plant
Through the quadrant correction method, adaptive weight adjustment and sliding average processing, the problem of inaccurate data in the assessment of wind resources in mountain wind farms is solved, the accuracy and stability of wind speed data is improved, and reliable data support is provided for wind farm planning. .
Patent Information
- Application Number
- CN202411260394.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-12-03
AI Technical Summary
There are problems with inaccurate and incomplete data assessment of wind resource in mountain wind farms. The traditional wind speed data correction method cannot fully consider various factors such as wind direction, temperature, and air pressure, resulting in low accuracy and insufficient data cleaning and quality control, which affects Accuracy and reliability of the assessment.
The accuracy and stability of wind speed data are gradually improved through data collection, cleaning, quality control, machine learning model application and verification optimization.
It significantly improves the accuracy and stability of wind speed data, reduces the random error and systematic error of the data, and provides reliable data support for wind resource evaluation of mountain wind farms.
Smart Images

Figure CN119067399A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of wind resource assessment, and in particular to a method for accurately correcting a representative year of wind resources in a mountainous wind farm. Background Art
[0002] With the continuous growth of global energy demand and the enhancement of environmental protection awareness, wind energy, as a clean and renewable form of energy, has received widespread attention and application. The construction of wind farms has developed rapidly around the world, especially in mountainous areas, which often have good wind resource conditions due to their unique terrain conditions. However, the wind resource assessment of mountain wind farms is more complicated than that of plains and hilly areas, mainly due to the diversity of mountain terrain and the frequent climate change. Accurately assessing the wind resources of mountain wind farms is of great significance to the planning, design and operation of wind farms. Therefore, the data revision of the representative year of wind resources has become a key link.
[0003] At present, representative annual data for wind resource assessment is usually obtained through observation data from wind towers. However, due to the temporal and spatial limitations of observation data from wind towers, coupled with the complex terrain and climate change in mountainous areas, these data are often inaccurate and incomplete. Traditional wind speed data correction methods, such as simple statistical methods and linear interpolation methods, cannot fully consider the influence of multiple factors such as wind direction, temperature, and air pressure, resulting in low accuracy of the correction results. In addition, the existing technology also has many deficiencies in data cleaning, quality control, and verification optimization, and fails to effectively eliminate outliers and fill in missing data, affecting the reliability of the final data. These technical problems seriously restrict the accuracy and reliability of wind resource assessment in mountain wind farms, and a more accurate and comprehensive data correction method is urgently needed. Summary of the invention
[0004] Based on the above purpose, the present invention provides a method for accurately correcting the representative year of wind resources in a mountain wind farm.
[0005] The accurate correction method of the wind resource representative year of a mountain wind farm includes the following steps:
[0006] S1, data collection and preprocessing: collect historical (at least 3 years) observation data from the wind tower, including timestamp, wind speed, wind direction, temperature and air pressure, obtain historical observation data from the reference station, including timestamp, wind speed, wind direction, temperature and air pressure, and conduct preliminary format review and integrity review;
[0007] S2, preliminary quadrant correction: using the quadrant correction method, the wind speed of each wind direction quadrant is corrected separately to obtain preliminary correction results;
[0008] S3, data cleaning and quality control: data cleaning and quality control are performed on the data after preliminary correction, outliers and missing values are eliminated, and missing data are filled;
[0009] S4, machine learning model application: the machine learning model is trained using cleaned and quality-controlled historical observation data, with input features including wind speed, wind direction, temperature and air pressure, and the output is the further corrected wind speed value;
[0010] S5, adaptive weight adjustment: according to the changes in different wind directions and time periods, the weight factor is calculated, the weight factor is introduced into the quadrant correction method, and correction is performed according to the weighted average method to further improve the data accuracy;
[0011] S6, sliding average value: for the data adjusted by adaptive weights, sliding average is performed every 12 consecutive months as one year to smooth the impact of short-term climate change on wind speed and obtain the representative annual value;
[0012] S7, data verification and optimization: verify the revised representative year data and make optimization adjustments based on actual conditions to ensure the applicability and accuracy of the revised method;
[0013] S8, result output: Output the final revised representative year data to provide data support for wind resource assessment and planning and construction of mountain wind farms.
[0014] Optionally, the S1 specifically includes:
[0015] S11, historical observation data collection of wind towers: select wind towers installed at the wind farm location, ensure that the wind towers are equipped with wind speed, wind direction, temperature, air pressure and humidity sensors, extract at least 3 years of historical observation data from the data recording system of the wind towers, these data are usually stored in the data logger or can be downloaded through the remote monitoring system, check the time interval of data recording (for example, once every hour), ensure that the number of records per year is close to the expected (for example, the amount of data recorded per hour should be 8760), store the data in a file with a standardized format (for example, CSV or Excel), including fields such as timestamp, wind speed, wind direction, temperature and air pressure;
[0016] S12, data collection of reference stations: select reference stations with similar or comparable meteorological conditions to the wind farm, ensure that the reference stations have long-term (at least 20 years) historical observation data, obtain historical observation data of the reference stations from the meteorological department or relevant data providers, these data can be obtained through data sharing agreements or public databases, ensure that the data integrity rate of the reference stations is not less than 90%, the calculation method is the same as that of the wind tower data, store the data in a standardized format, ensure that it contains fields such as timestamp, wind speed, wind direction, temperature and air pressure;
[0017] S13, format review: check the structure of the data file, including the column header and order of each field, to ensure that it is consistent with expectations, and check the timestamp format;
[0018] S14, Completeness Review: Check the data records each year and calculate the completeness rate.
[0019] Optionally, the S2 specifically includes:
[0020] S21, data preparation: Use the historical observation data of the wind tower and the historical observation data of the reference station collected in S1 to ensure data integrity. The data includes parameters such as timestamp, wind speed, wind direction, temperature and air pressure;
[0021] S22, determining the wind direction quadrant: dividing the wind direction into 16 quadrants, each quadrant corresponding to 22.5 degrees, for example, 0°-22.5° is the first quadrant, 22.5°-45° is the second quadrant, and so on;
[0022] S23, data classification: classify the wind speed data of the wind tower and the reference station into corresponding quadrants according to the wind direction. For example, when the wind direction is between 0° and 22.5°, the wind speed data is classified into the first quadrant.
[0023] S24, calculating the wind speed correlation curve: for each wind direction quadrant, draw the wind speed correlation curve between the wind tower and the reference station, and establish a rectangular coordinate system, where the horizontal coordinate is the wind speed of the reference station and the vertical coordinate is the wind speed of the wind tower;
[0024] S25, calculating the algebraic difference of wind speed: in each wind direction quadrant, calculating the annual average wind speed of the reference station and the wind tower, and finding the algebraic difference;
[0025] S26, correcting the wind speed of the wind tower: for each time point in each wind direction quadrant, use the algebraic difference to correct the wind speed of the wind tower;
[0026] S27, merging correction results: merging the corrected wind speed data in each wind direction quadrant to obtain a preliminary correction result.
[0027] Optionally, the S3 specifically includes:
[0028] S31, remove outliers: use the Z-Score method to detect outliers and delete them;
[0029] S32, processing missing values: record the location and number of missing data, and use linear interpolation to fill in the missing data;
[0030] S33, data smoothing: Smoothing the data to eliminate short-term abnormal fluctuations and ensure the continuity and rationality of the data.
[0031] Optionally, the S4 specifically includes:
[0032] S41, Merge Data: Use the cleaned and quality-controlled wind tower data and reference station data to create a comprehensive data set, including timestamp, wind tower wind speed, reference station wind speed, wind direction, temperature and air pressure;
[0033] S42, data set division: divide the data set into training set and test set to ensure uniform data distribution;
[0034] S43, model selection: select the random forest regression model;
[0035] S44, model training: Use the training set data to train the random forest regression model. The input features include wind speed, wind direction, temperature and air pressure of the reference station. The output is the wind speed value of the wind tower.
[0036] S45, Model Evaluation: Use the test set data to evaluate the performance of the model.
[0037] Optionally, the S5 specifically includes:
[0038] S51, data analysis and weight factor calculation: group the data according to wind direction and time period. For example, the wind direction can be divided into 16 quadrants (one quadrant for each 22.5 degrees), and the time period can be grouped by season or month. Calculate the statistical characteristics of the wind speed in each quadrant and time period, including the mean and variance. These characteristics are used to evaluate the fluctuation and change trend of the data. Calculate the weight factor based on the stability and representativeness of the wind speed data. For example, use the inverse of the variance of each quadrant and time period as the weight factor. The smaller the variance, the larger the weight, indicating that the data is more stable.
[0039] S52, introducing weight factors in the quadrant correction method: According to the quadrant correction method, the wind speed in each wind direction quadrant is corrected separately to obtain a preliminary correction result. On the basis of the preliminary correction result, the calculated weight factors are introduced to perform weighted averaging on the wind speed correction results in each wind direction quadrant, and the weight factors are used as weights to adjust the correction results.
[0040] Optionally, the S6 specifically includes:
[0041] S 61, arranging the wind speed data after adaptive weight adjustment in chronological order to ensure the continuity and integrity of the data;
[0042] S62, taking each consecutive 12 months as a sliding window, for example, January 2023 to December 2023 is one window, February 2023 to January 2024 is the next window, and so on, averaging the data in each sliding window to obtain the representative annual wind speed value of each window.
[0043] Optionally, the S7 specifically includes:
[0044] S71, data verification: part of the merged data set is reserved as a verification set. The verification set data does not participate in the model training and correction process, and is used to evaluate the accuracy of the correction results. The corrected representative year data is compared with the verification set data, and the evaluation indicators are calculated. The evaluation indicators include mean square error, root mean square error, mean absolute error and determination coefficient. The calculated evaluation indicators are compared with the preset indicators to evaluate the accuracy of the correction results.
[0045] S72, Analyze the source of error: Through error analysis, determine the data points and time periods with large errors, find out the specific reasons, such as insufficient models, data quality issues, etc., check whether the calculation of weight factors for different wind directions and time periods is reasonable, and whether they need to be readjusted;
[0046] S73, re-adjusting the weight factor: recalculating the weight factors of different wind directions and time periods according to the error analysis results, and adjusting the calculation method of the weight factor, for example, introducing more statistical features (such as median, quartile, etc.) to improve the representativeness of the weight factor;
[0047] S74, data reprocessing: reprocess the historical observation data collected from the wind tower and the reference station, such as re-data cleaning, outlier detection and missing value filling, etc., and re-correct the wind speed data and perform sliding average processing based on the new weight factor.
[0048] Optionally, the S8 specifically includes:
[0049] S81, Data preparation and collation: Ensure that the final revised representative year data is verified and optimized, arranged in chronological order, and the data is complete and accurate;
[0050] S82, Data storage and formatting: Store the final revised representative year data in a standardized format to facilitate subsequent analysis and use;
[0051] S83, data report generation: generating a data report.
[0052] Beneficial effects of the present invention:
[0053] The present invention significantly improves the accuracy of wind speed data by introducing quadrant correction method, adaptive weight adjustment and sliding average processing. The quadrant correction method makes preliminary corrections according to the changes in different wind directions, and the adaptive weight adjustment calculates the weight factor according to the changes in different wind directions and time periods, and further accurately corrects the wind speed data. The sliding average processing can smooth the impact of short-term climate change, highlight long-term trends, and ensure the stability and reliability of the final representative annual data. Through the comprehensive application of these methods, the present invention greatly reduces the random error and systematic error of the data and improves the accuracy of wind resource assessment.
[0054] The present invention provides a systematic and full-process data processing method, which ensures the integrity and continuity of the data from data collection, cleaning, correction, verification to final output. In the data collection stage, ensure that the data integrity rate is not less than 90%; in the data cleaning and quality control stage, eliminate outliers and fill in missing values; in the data correction stage, improve the accuracy of the data through a variety of methods; in the verification and optimization adjustment stage, ensure the reliability of the data. The representative annual data finally output has undergone rigorous processing and verification, which ensures the integrity and continuity of the data and provides solid data support for the wind resource assessment and planning and construction of mountain wind farms. Through such a full-process processing method, the present invention not only improves the quality of the data, but also provides a reliable reference basis for the subsequent design and operation of wind farms. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0056] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;
[0057] Figure 2 Schematic diagram of S2 process of an embodiment of the present invention. DETAILED DESCRIPTION
[0058] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0059] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0060] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0061] like Figure 1-Figure 2 As shown in FIG. 1 , the wind resource representative year accurate correction method of the mountain wind farm includes the following steps:
[0062] S1, data collection and preprocessing: collect historical (at least 3 years) observation data from the wind tower, including timestamp, wind speed, wind direction, temperature and air pressure, obtain historical observation data from the reference station, including timestamp, wind speed, wind direction, temperature and air pressure, and conduct preliminary format review and integrity review;
[0063] S2, preliminary quadrant correction: using the quadrant correction method, the wind speed of each wind direction quadrant is corrected separately to obtain preliminary correction results;
[0064] S3, data cleaning and quality control: data cleaning and quality control are performed on the data after preliminary correction, outliers and missing values are eliminated, and missing data are filled;
[0065] S4, machine learning model application: the machine learning model is trained using cleaned and quality-controlled historical observation data, with input features including wind speed, wind direction, temperature and air pressure, and the output is the further corrected wind speed value;
[0066] S5, adaptive weight adjustment: according to the changes in different wind directions and time periods, the weight factor is calculated, the weight factor is introduced into the quadrant correction method, and correction is performed according to the weighted average method to further improve the data accuracy;
[0067] S6, sliding average value: for the data adjusted by adaptive weights, sliding average is performed every 12 consecutive months as one year to smooth the impact of short-term climate change on wind speed and obtain the representative annual value;
[0068] S7, data verification and optimization: verify the revised representative year data and make optimization adjustments based on actual conditions to ensure the applicability and accuracy of the revised method;
[0069] S8, result output: Output the final revised representative year data to provide data support for wind resource assessment and planning and construction of mountain wind farms.
[0070] Optionally, the S1 specifically includes:
[0071] S11, historical observation data collection of wind towers: select wind towers installed at the wind farm location, ensure that the wind towers are equipped with wind speed, wind direction, temperature, air pressure and humidity sensors, extract at least 3 years of historical observation data from the data recording system of the wind towers, these data are usually stored in the data logger or can be downloaded through the remote monitoring system, check the time interval of data recording (for example, once every hour), ensure that the number of records per year is close to the expected (for example, the amount of data recorded per hour should be 8760), store the data in a file with a standardized format (for example, CSV or Excel), including fields such as timestamp, wind speed, wind direction, temperature and air pressure;
[0072] S12, data collection of reference stations: select reference stations with similar or comparable meteorological conditions to the wind farm, ensure that the reference stations have long-term (at least 20 years) historical observation data, obtain historical observation data of the reference stations from the meteorological department or relevant data providers, these data can be obtained through data sharing agreements or public databases, ensure that the data integrity rate of the reference stations is not less than 90%, the calculation method is the same as that of the wind tower data, store the data in a standardized format, ensure that it contains fields such as timestamp, wind speed, wind direction, temperature and air pressure;
[0073] S13, format review: check the structure of the data file, including the column header and order of each field, to ensure that it is consistent with expectations, check the timestamp format, ensure a unified date and time format (e.g., YYYY-MM-DDHH:MM), verify that the data type of each field is correct, for example, wind speed should be a floating point number, wind direction should be an integer from 0° to 360°, temperature and pressure should be floating point numbers, check that the column headers of all files are consistent, and ensure that the field names and order in different files are consistent;
[0074] S14, completeness review: Check the data records each year and calculate the completeness rate, completeness rate = (actual number of records / theoretical number of records) × 100%, to ensure that the data completeness rate is not less than 90% each year.
[0075] Optionally, the S2 specifically includes:
[0076] S21, data preparation: Use the historical observation data of the wind tower and the historical observation data of the reference station collected in S1 to ensure data integrity. The data includes parameters such as timestamp, wind speed, wind direction, temperature and air pressure;
[0077] S22, determining the wind direction quadrant: dividing the wind direction into 16 quadrants, each quadrant corresponding to 22.5 degrees, for example, 0°-22.5° is the first quadrant, 22.5°-45° is the second quadrant, and so on;
[0078] S23, data classification: classify the wind speed data of the wind tower and the reference station into corresponding quadrants according to the wind direction. For example, when the wind direction is between 0° and 22.5°, the wind speed data is classified into the first quadrant.
[0079] S24, calculating the wind speed correlation curve: for each wind direction quadrant, draw the wind speed correlation curve between the wind tower and the reference station, and establish a rectangular coordinate system, where the horizontal coordinate is the wind speed of the reference station and the vertical coordinate is the wind speed of the wind tower;
[0080] S25, calculate the algebraic difference of wind speed: in each wind direction quadrant, calculate the annual average wind speed of the reference station and the wind tower, and find the algebraic difference. The algebraic difference is calculated as:
[0081] ΔV=V 参证站 -V 测风塔 ;
[0082] Among them, V 参证站 and V 测风塔 are the annual average wind speeds at the reference station and the wind tower, respectively;
[0083] S26, Correction of wind speed at the wind tower: For each time point in each wind direction quadrant, the wind speed at the wind tower is corrected using algebraic difference. The corrected wind speed is calculated as:
[0084] V′=V+ΔV;
[0085] Where V is the original wind speed of the wind tower, ΔV is the algebraic difference, and V′ is the corrected wind speed;
[0086] S27, merging correction results: merging the corrected wind speed data in each wind direction quadrant to obtain a preliminary correction result;
[0087] Example steps: Suppose the following data:
[0088] Wind speed data from wind tower: V 测风塔 = [5.2, 4.8, 5.0, ...];
[0089] Wind direction data from wind tower: θ 测风塔 = [10°, 30°, 15°, ...];
[0090] Wind speed data of reference station: V 参证站 = [5.5, 5.1, 5.3, ...];
[0091] Reference station wind direction data: θ 参证站 = [12°, 35°, 20°, ...];
[0092] Determine wind direction quadrants: Divide wind direction into 16 quadrants and define the range of each quadrant;
[0093] Data classification: According to the wind direction, the wind speed data of the wind tower and the reference station are classified into the corresponding quadrants. For example: The first quadrant (0°-22.5°): wind speed V of the wind tower 测风塔 =[5.2, 5.0, ...], wind speed V at the reference station 参证站 = [5.5, 5.3, ...];
[0094] Calculate the wind speed correlation curve: In the first quadrant, draw the wind speed correlation curve between the wind tower and the reference station;
[0095] Calculate the algebraic difference of wind speed: Calculate the algebraic difference of wind speed in the first quadrant:
[0096] ΔV 第一个象限 = mean([5.5, 5.3]) - mean([5.2, 5.0]);
[0097] ΔV 第一个象限 =5.4-5.1=0.3;
[0098] Correcting Mast Wind Speed: Correcting the Mast Wind Speed using algebraic differences:
[0099] V′ 测风塔 =5.2+0.3=5.5;
[0100] V′ 测风塔 =5.0+0.3=5.3;
[0101] Merge the correction results: Merge the corrected wind speed data in each wind direction quadrant to obtain a preliminary correction result;
[0102] Through the above steps, the quadrant correction method can be systematically used to correct the wind speed in each wind direction quadrant, and preliminary correction results can be obtained, providing a basis for subsequent further correction and analysis.
[0103] Optionally, the S3 specifically includes:
[0104] S31, remove outliers: use the Z-Score method to detect outliers and delete them;
[0105] The Z-Score method detects outliers by calculating the deviation of each data point from the mean and normalizing it to a Z value, calculated as:
[0106] Where Z is the standardized Z value, x is the value of the data point, μ is the mean of the data, and σ is the standard deviation of the data;
[0107] Detection condition: If the Z value of a data point satisfies: |Z|>k, then the data point is considered an outlier. Usually k is 2 or 3.
[0108] S32, processing missing values: record the location and number of missing data, and use linear interpolation to fill in the missing data;
[0109] Linear interpolation is used to estimate missing data points between two known data points and is calculated as:
[0110]
[0111] Where (x1, y1) and (x2, y2) are known data points, x is the horizontal coordinate of the missing data point, and y is the vertical coordinate of the estimated missing data point;
[0112] S33, data smoothing: Smoothing the data to eliminate short-term abnormal fluctuations and ensure the continuity and rationality of the data.
[0113] Optionally, the S4 specifically includes:
[0114] S41, Merge Data: Use the cleaned and quality-controlled wind tower data and reference station data to create a comprehensive data set, including timestamp, wind tower wind speed, reference station wind speed, wind direction, temperature and air pressure;
[0115] S42, data set division: divide the data set into training set and test set to ensure uniform data distribution;
[0116] S43, model selection: select the random forest regression model;
[0117] S44, model training: Use the training set data to train the random forest regression model. The input features include wind speed, wind direction, temperature and air pressure of the reference station. The output is the wind speed value of the wind tower.
[0118] S45, model evaluation: Use the test set data to evaluate the performance of the model, compare the model prediction value with the actual wind speed value of the wind tower, and verify the accuracy of the model;
[0119] By using the random forest regression model, multi-feature data can be effectively processed and the accuracy and reliability of wind speed correction can be improved, thereby providing more reliable data support for wind resource assessment and wind farm planning.
[0120] Optionally, the S5 specifically includes:
[0121] S51, data analysis and weight factor calculation: group the data according to wind direction and time period. For example, the wind direction can be divided into 16 quadrants (one quadrant for each 22.5 degrees), and the time period can be grouped by season or month. Calculate the statistical characteristics of wind speed in each quadrant and time period, including mean and variance. These characteristics are used to evaluate the fluctuation and change trend of the data. Calculate the weight factor based on the stability and representativeness of the wind speed data. For example, use the inverse of the variance of each quadrant and time period as the weight factor. The smaller the variance, the larger the weight, indicating that the data is more stable, which is expressed as:
[0122] Among them, w i,j represents the weight factor of the jth time period in the i-th quadrant, σ i,j represents the wind speed variance in the jth time period of the ith quadrant;
[0123] S52, introducing weight factors into the quadrant correction method: According to the quadrant correction method, the wind speed of each wind direction quadrant is corrected respectively to obtain a preliminary correction result. On the basis of the preliminary correction result, the calculated weight factors are introduced to perform weighted averaging on the wind speed correction results of each wind direction quadrant. The weight factors are used as weights to adjust the correction results, which are expressed as:
[0124] Among them, V i,j represents the weighted average correction result of the jth time period in the i-th quadrant, V i,j,k represents the preliminary corrected wind speed of the kth data point in the jth time period in the i-th quadrant, w i,j represents the weight factor of the jth time period in the i-th quadrant;
[0125] Example steps: Suppose the following data:
[0126] Wind direction data: θ = [10°, 30°, 50°, 70°, ...];
[0127] Time period: grouped by month (e.g., January, February, March, etc.);
[0128] Wind speed data: V 初步订正 = [5.2, 4.8, 5.0, ...];
[0129] (1) Group analysis:
[0130] Group the data by wind direction and time period, for example:
[0131] The first quadrant (0°-22.5°), data from January;
[0132] The second quadrant (22.5°-45°), data from January; ...:
[0134] (2) Calculate the weight factor:
[0135] Calculate the variance of wind speed in each quadrant and time period:
[0136] First quadrant, wind speed variance in January:
[0137] Second quadrant, wind speed variance in January: ...;
[0139] Calculate the weight factor:
[0140] The weight factor for the first quadrant, January:
[0141] Second quadrant, weight factor for January: ...;
[0143] (3) Introduce weight factor for weighted average correction:
[0144] A weighted average correction is made to the wind speed in each wind direction quadrant and time period:
[0145] The first quadrant, the weighted average revised results for January:
[0146] The second quadrant, the weighted average revised results for January: ...;
[0148] By calculating the changes in different wind directions and time periods, determining the weight factors and introducing these weight factors in the quadrant correction method, the wind speed data can be further corrected using the weighted average method. This method can better reflect the stability and representativeness of the data, improve the accuracy of the correction results, and provide more reliable data support for wind resource assessment and wind farm planning.
[0149] Based on the above implementation scheme, regarding the method of obtaining the annual value in S6, the long sequence average method, the sliding average method or the sampling average method can also be used to obtain the annual value. The specific method is as follows:
[0150] Long-sequence average method: The observation period of the wind tower is no less than 3 years. When the average value of the reference station in the same period is close to the multi-year average value, the average value of the wind tower in this period can be used as the representative annual value;
[0151] Sliding average value method: Assume that the wind tower has m consecutive months of observation data, where m>12, and each consecutive 12 months is a year, which can form m-11 years; if the reference station has a year with the same period average value as the wind tower that is close to the multi-year average value, then the wind tower value of that year can be used as the representative year value; if the reference station has multiple such years, or the average value of multiple annual values is close to the multi-year average value, then the average value of multiple years corresponding to the wind tower is taken as the representative year value;
[0152] Sampling value averaging method: suppose the wind tower has m consecutive months of observation data, where m>12, and there are m1 Januarys. If among the m1 January data corresponding to the reference station, there is a January average value in a certain year that is close to the multi-year January average value, then the average value of the wind tower in January of the same year is taken as the average value of January in the representative year; if such values appear in multiple Januarys at the reference station, or the average values of multiple Januarys are close to the multi-year January average value, then the average value of the multiple values corresponding to the wind tower is taken as the average value of January in the representative year, and the representative annual and monthly values of other months are obtained by the same method. If the representative annual and monthly values of 12 months can be obtained, their average is the representative annual value; if there are some months for which no suitable representative annual values can be found, two or more different months can be combined to obtain values.
[0153] Based on the comparison of the annual values of the above-mentioned regions, the observation period of the long series average method is longer. This method is only calculated based on past data, ignoring other factors that may affect the future, resulting in inaccurate prediction results; and if there are outliers in the data, the long series average method may be greatly affected, thereby affecting the accuracy of the prediction; and the sliding average method involves a large number of separate operations, and it is necessary to wait until the required time period has passed before a reliable prediction can be obtained, and as time goes by, the longer the demand data is, the less relevant it becomes to future demand forecasts, and the smaller the impact on it; secondly, the sampling sampling average method has limitations and cannot completely replace the full forecast; its scope of application is limited, and for values taken in short or discrete time periods, the sampling results may not accurately reflect the overall situation.
[0154] Based on the above comparison, optionally, S6 specifically includes:
[0155] S 61, arranging the wind speed data after adaptive weight adjustment in chronological order to ensure the continuity and integrity of the data;
[0156] S62, take each 12 consecutive months as a sliding window, for example, January 2023 to December 2023 is a window, February 2023 to January 2024 is the next window, and so on. Average the data in each sliding window to obtain the representative annual wind speed value of each window. When calculating the sliding average, for windows close to the beginning or end of the data set, there may be less than 12 months of data points. The existing data points can be used to calculate the average value, or windows with incomplete boundaries can be discarded. The sliding average is expressed as:
[0157]
[0158] Where V 平均 (t) is the average wind speed value in the sliding window starting at time t, V(t+i) is the wind speed value at time t+i, and N is the size of the sliding window (i.e., the number of data points in 12 months);
[0159] Through the above steps, the wind speed data after adaptive weight adjustment can be sliding averaged to smooth the impact of short-term climate change and obtain representative annual wind speed values. This method can effectively reduce the impact of short-term fluctuations and highlight long-term trends, thereby providing more stable and reliable data support for wind resource assessment and wind farm planning.
[0160] Optionally, the S7 specifically includes:
[0161] S71, data verification: part of the merged data set is reserved as a verification set. The verification set data does not participate in the model training and correction process, and is used to evaluate the accuracy of the correction results. The corrected representative year data is compared with the verification set data, and the evaluation indicators are calculated. The evaluation indicators include mean square error, root mean square error, mean absolute error and determination coefficient. The calculated evaluation indicators are compared with the preset indicators to evaluate the accuracy of the correction results.
[0162] Mean Square Error: Measures the average square error between the predicted value and the actual value, expressed as:
[0163] where yi is the actual value, is the predicted value, n is the number of samples;
[0164] Root mean square error: The square root of the mean square error more intuitively reflects the size of the error and is expressed as:
[0165]
[0166] Mean absolute error: The average absolute error between the predicted value and the actual value, expressed as:
[0167]
[0168] Coefficient of determination: measures the goodness of fit of the model, with a value range of 0 to 1. The closer it is to 1, the better the model fit is, expressed as:
[0169] in is the mean of the actual values;
[0170] S72, Analyze the source of error: Through error analysis, determine the data points and time periods with large errors, find out the specific reasons, such as insufficient models, data quality issues, etc., check whether the calculation of weight factors for different wind directions and time periods is reasonable, and whether they need to be readjusted;
[0171] S73, re-adjusting the weight factor: recalculating the weight factors of different wind directions and time periods according to the error analysis results, and adjusting the calculation method of the weight factor, for example, introducing more statistical features (such as median, quartile, etc.) to improve the representativeness of the weight factor;
[0172] S74, data reprocessing: reprocess the historical observation data collected from the wind tower and the reference station, such as re-data cleaning, outlier detection and missing value filling, etc., and re-correct the wind speed data and perform sliding average processing based on the new weight factor.
[0173] Optionally, the S8 specifically includes:
[0174] S81, Data preparation and collation: Ensure that the final revised representative year data is verified and optimized, arranged in chronological order, and the data is complete and accurate;
[0175] S82, Data storage and formatting: Store the final revised representative year data in a standardized format to facilitate subsequent analysis and use;
[0176] S83, data report generation: generating a data report;
[0177] Data report examples: Wind resource assessment report;
[0178] Data source and collection method: The data in this report comes from the wind towers and reference stations of mountain wind farms, including wind speed, wind direction, temperature and air pressure. The data collection period is at least 3 years, and the data completeness rate is not less than 90%;
[0179] Data cleaning and quality control process: Conduct preliminary format and integrity review of the collected data, use statistical methods to eliminate outliers, use interpolation to fill missing values, and perform data smoothing;
[0180] Adaptive weight adjustment and sliding average processing: Calculate the weight factor according to the changes in different wind directions and time periods, introduce the weight factor into the quadrant correction method, and make corrections through the weighted average method. The adjusted data is averaged over a period of 12 consecutive months to smooth out the impact of short-term climate change;
[0181] Data verification and optimization adjustment: Use data that has not been corrected for verification, calculate evaluation indicators, and perform error analysis and model optimization.
[0182] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0183] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for accurately correcting the representative year of wind resources for mountainous wind farms, characterized in that: The following steps are involved: S1, data collection and preprocessing: collect historical observation data from wind towers, including timestamp, wind speed, wind direction, temperature and air pressure, obtain historical observation data from reference stations, including timestamp, wind speed, wind direction, temperature and air pressure, and conduct preliminary format review and integrity review; S2, preliminary quadrant correction: using the quadrant correction method, the wind speed of each wind direction quadrant is corrected separately to obtain preliminary correction results; S3, data cleaning and quality control: data cleaning and quality control are performed on the data after preliminary correction, outliers and missing values are eliminated, and missing data are filled; S4, machine learning model application: the machine learning model is trained using cleaned and quality-controlled historical observation data, with input features including wind speed, wind direction, temperature and air pressure, and the output is the further corrected wind speed value; S5, adaptive weight adjustment: according to the changes in different wind directions and time periods, the weight factor is calculated, the weight factor is introduced into the quadrant correction method, and correction is performed according to the weighted average method to further improve the data accuracy; S6, sliding average value: for the data adjusted by adaptive weights, sliding average is performed every 12 consecutive months as one year to smooth the impact of short-term climate change on wind speed and obtain the representative annual value; S7, data verification and optimization: verify the revised representative year data and make optimization adjustments based on actual conditions; S8, result output: Output the final revised representative year data to provide data support for wind resource assessment and planning and construction of mountain wind farms.
2. The method for accurately correcting the representative year of wind resources for a mountain wind farm according to claim 1, characterized in that: The S1 specifically includes: S11, historical observation data collection of wind towers: select wind towers installed at the wind farm location, ensure that the wind towers are equipped with wind speed, wind direction, temperature, air pressure and humidity sensors, extract historical observation data from the data recording system of the wind towers, and store the data in a file in a standardized format; S12, data collection of reference stations: select reference stations with meteorological conditions similar to those of the wind farm, ensure that the reference stations have long-term historical observation data, obtain historical observation data of the reference stations from the meteorological department or relevant data providers, and store the data in files in a standardized format; S13, format review: check the structure of the data file, including the column header and order of each field, to ensure that it is consistent with expectations, and check the timestamp format; S14, Completeness Review: Check the data records each year and calculate the completeness rate.
3. The method for accurately correcting the representative year of wind resources for a mountain wind farm according to claim 2, characterized in that: The S2 specifically includes: S21, data preparation: using the historical observation data of the wind tower and the historical observation data of the reference station collected in S1; S22, determine the wind direction quadrant: divide the wind direction into 16 quadrants, each quadrant corresponds to 22.5 degrees; S23, data classification: classify the wind speed data of the wind tower and the reference station into the corresponding quadrants according to the wind direction; S24, calculating the wind speed correlation curve: for each wind direction quadrant, draw the wind speed correlation curve between the wind tower and the reference station, and establish a rectangular coordinate system, where the horizontal coordinate is the wind speed of the reference station and the vertical coordinate is the wind speed of the wind tower; S25, calculating the algebraic difference of wind speed: in each wind direction quadrant, calculating the annual average wind speed of the reference station and the wind tower, and finding the algebraic difference; S26, correcting the wind speed of the wind tower: for each time point in each wind direction quadrant, the wind speed of the wind tower is corrected using algebraic difference; S27, merging correction results: merging the corrected wind speed data in each wind direction quadrant to obtain a preliminary correction result.
4. The method for accurately correcting the representative year of wind resources for a mountain wind farm according to claim 3 is characterized in that: The S3 specifically includes: S31, remove outliers: use the Z-Score method to detect outliers and delete them; S32, processing missing values: record the location and number of missing data, and use linear interpolation to fill in the missing data; S33, data smoothing: Smoothing the data to eliminate short-term abnormal fluctuations and ensure the continuity and rationality of the data.
5. The method for accurately correcting the representative year of wind resources for a mountain wind farm according to claim 4, characterized in that: The S4 specifically includes: S41, Merge Data: Use the cleaned and quality-controlled wind tower data and reference station data to create a comprehensive data set, including timestamp, wind tower wind speed, reference station wind speed, wind direction, temperature and air pressure; S42, data set division: divide the data set into training set and test set; S43, model selection: select the random forest regression model; S44, model training: Use the training set data to train the random forest regression model. The input features include wind speed, wind direction, temperature and air pressure of the reference station. The output is the wind speed value of the wind tower. S45, Model Evaluation: Use the test set data to evaluate the performance of the model.
6. The method for accurately correcting the wind resource representative year of a mountain wind farm according to claim 5, characterized in that: The S5 specifically includes: S51, data analysis and weight factor calculation: grouping the data according to wind direction and time period, calculating the statistical characteristics of wind speed in each quadrant and time period, including mean and variance, and calculating the weight factor based on the wind speed data; S52, introducing weight factors in the quadrant correction method: According to the quadrant correction method, the wind speed in each wind direction quadrant is corrected separately to obtain a preliminary correction result. On the basis of the preliminary correction result, the calculated weight factors are introduced to perform weighted averaging on the wind speed correction results in each wind direction quadrant, and the weight factors are used as weights to adjust the correction results.
7. The method for accurately correcting the wind resource representative year of a mountain wind farm according to claim 6, characterized in that: The S6 specifically includes: S 61, arranging the wind speed data after adaptive weight adjustment in chronological order; S62, taking each 12 consecutive months as a sliding window, averaging the data in each sliding window to obtain a representative annual wind speed value for each window.
8. The method for accurately correcting the wind resource representative year of a mountain wind farm according to claim 7, characterized in that: The S7 specifically includes: S71, data verification: part of the merged data set is reserved as a verification set. The verification set data does not participate in the model training and correction process, and is used to evaluate the accuracy of the correction results. The corrected representative year data is compared with the verification set data, and the evaluation index is calculated. The calculated evaluation index is compared with the preset index to evaluate the accuracy of the correction results. S72, analyze the source of error: through error analysis, determine the data points and time periods with large errors and find out the specific reasons; S73, re-adjusting weight factors: recalculating weight factors of different wind directions and time periods according to the error analysis results; S74, data reprocessing: reprocessing the collected historical observation data of the wind tower and the historical observation data of the reference station.
9. The method for accurately correcting the wind resource representative year of a mountain wind farm according to claim 8, characterized in that: The S8 specifically includes: S81, Data preparation and arrangement: Ensure that the final revised representative year data are verified and optimized and arranged in chronological order; S82, Data storage and formatting: Store the final revised representative year data in a standardized format to facilitate subsequent analysis and use; S83, data report generation: generating a data report.
Citation Information
Cited By
Wind speed forecast correction method, device and equipment and readable storage medium
CN121188716A
