Large wind turbine power curve testing and analysis system

Abstract

The present application relates to the technical field of industrial cloud platform, and disclose a large wind turbine power curve test and analysis system.The system is deployed on the industrial cloud platform, receives wind speed power data uploaded by the local data acquisition terminal of the wind farm through the data interface of the industrial cloud platform, adds a skewness nonlinear correction module, calculates the wind speed skewness value and the power curve curvature indicator for each Bin, when the absolute value of skewness, the absolute value of curvature indicator and the sample number meet the preset threshold condition at the same time, the power arithmetic mean value of the Bin is corrected and calculated.The present application compensates the arithmetic mean estimation deviation caused by the coupling of wind speed distribution skewness and power curve bending in Bin, and realizes remote access and multi-user collaborative analysis of test data through the deployment of industrial cloud platform, and improves the test accuracy of large wind turbine power curve in the high curvature section near the rated wind speed.

Description

Technical Field

[0001] This invention relates to the field of industrial cloud platform technology, and in particular to a power curve testing and analysis system for large wind turbine units. Background Technology

[0002] In power curve testing of large wind turbines, industry practice typically involves dividing the collected wind speed and power data into multiple wind speed intervals (Bin) at fixed intervals. The power values ​​within each Bin are then arithmetically averaged to generate the measured power curve. This method is computationally simple and easy to implement, and is widely used in existing testing and analysis systems. However, the power curve of large wind turbines near their rated wind speed exhibits significant curvature. Furthermore, the wind speed distribution within a single Bin is often not ideally uniform, but rather skewed, with most wind speed data concentrated at one end of the Bin. When the wind speed distribution within a Bin is skewed and that Bin happens to be located in the curved section of the power curve, the arithmetic mean will produce a systematic bias due to the coupling between the asymmetry of the wind speed distribution and the nonlinearity of the power curve: the power values ​​at the midpoint of the wind speed data are overrepresented, while the power values ​​at the sparse end are underrepresented, causing the average power estimate of that Bin to deviate from the turbine's true steady-state output at that wind speed. This deviation can reach tens of kilowatts in the steep transition section near the rated wind speed, which happens to be the most sensitive range for power curve verification and contract compliance assessment.

[0003] While existing technologies offer various improvements to data cleaning and curve fitting, such as outlier removal based on density clustering and piecewise nonlinear function fitting, these methods primarily address data quality itself and do not address the coupling effect between the wind speed distribution pattern within a bin and the geometric curvature of the power curve. How to effectively compensate for power estimation biases caused by this coupling effect, and improve the testing accuracy of power curves for large wind turbines in high-curvature wind speed ranges, while remaining compatible with existing bin partitioning rules and without adding extra measurement hardware, is a problem that needs to be solved in current power curve testing and analysis technologies. Summary of the Invention

[0004] The present invention aims to address the problem in the prior art that the power values ​​within each wind turbine are arithmetically averaged after dividing the wind speed into multiple wind speed intervals (Bin) at fixed wind speed intervals, resulting in significant deviations in the results. To address this issue, we propose a power curve testing and analysis system for large wind turbine units.

[0005] To achieve the above objectives, this application adopts the following technical solution: a large-scale wind turbine power curve testing and analysis system, deployed on an industrial cloud platform, including: a data import and configuration parsing module, used to receive time-synchronized wind speed and power data files uploaded by local data acquisition terminals of wind farms through the data interface of the industrial cloud platform, and convert the wind speed and power data files into a standard wind speed and power data table in a unified format within the system; a data validity filtering module, used to receive the standard wind speed and power data table, and according to the turbine configuration parameters and preset filtering rules, remove invalid data records that do not meet the test conditions, and output a valid wind speed and power data table; a wind speed bin grouping and statistics module, used to divide each data record in the valid wind speed and power data table into the corresponding wind speed interval bin according to a preset fixed wind speed interval, and calculate the arithmetic mean of wind speed, the arithmetic mean of power, the sample size, and the wind speed variance for each bin; The nonlinear correction module is used to determine whether each Bin meets the preset correction trigger conditions. For Bins that meet the correction trigger conditions, the module uses the Bin's wind speed skewness, power curve curvature indicator, and wind speed variance to perform a correction calculation on the Bin's arithmetic mean power value to obtain the corrected power value. For Bins that do not meet the correction trigger conditions, the Bin's arithmetic mean power value is directly output as the corrected power value. The power curve generation and comparison analysis module is used to generate a measured power curve based on the Bin center wind speed and the corrected power value of each Bin, and compares the measured power curve with the input contract-guaranteed power curve point by point, outputting the comparison analysis results. The report output module is used to summarize the measured power curve, comparison analysis results, and correction information of each Bin to generate a test report, and store the test report on the storage node of the industrial cloud platform, providing authorized remote user terminals with the opportunity to view and download the report.

[0006] Preferably, the data import and configuration parsing module is specifically used for: reading wind speed and power data files through the data interface of the industrial cloud platform, automatically identifying the wind speed and power fields in the file, and mapping the wind speed and power fields to the standard wind speed field name and standard power field name within the system to form a standard wind speed and power data table; reading the unit configuration parameters, which at least include: unit model, hub height, air density, cut-in wind speed, cut-out wind speed, and the start and end angles of invalid wind direction sectors; and storing the unit configuration parameters as system global configuration variables for use by the data validity filtering module and the power curve generation and comparison analysis module.

[0007] Preferably, the data validity filtering module is specifically used to mark and remove data records in the standard wind speed and power data table that meet at least one of the following rejection conditions: data records where the wind turbine is in a stopped state; data records where the wind turbine is in a fault state; data records where the wind turbine is in a limited power operation state; data records where the wind direction value falls within the invalid wind direction sector range; data records where the wind speed value exceeds the range from the cut-in wind speed to the cut-out wind speed; and data records where the wind speed field or power field is empty or negative.

[0008] Preferably, the wind speed Bin grouping and statistics module is specifically used for: dividing the effective wind speed and power data table into multiple wind speed intervals (Bin) at preset fixed wind speed intervals; taking the midpoint of the upper and lower wind speed values ​​of each Bin as the center wind speed; for each Bin, traversing all data records in the effective wind speed and power data table whose wind speed values ​​fall into that Bin, and extracting the corresponding wind speed value sequence and power value sequence; calculating the arithmetic mean of wind speed based on the wind speed value sequence; calculating the arithmetic mean of power based on the power value sequence; counting the number of data records falling into that Bin as the sample size; calculating the wind speed variance based on the wind speed value sequence and the wind speed arithmetic mean; and taking the Bin number, Bin center wind speed, wind speed arithmetic mean, power arithmetic mean, sample size, wind speed variance, and wind speed standard deviation of each Bin as the statistical output record of that Bin.

[0009] Preferably, the power curve generation and comparison analysis module is specifically used for: connecting the data points corresponding to each Bin sequentially with the Bin center wind speed as the abscissa and the corrected power value of each Bin as the ordinate to form a measured power curve; receiving the contract-guaranteed power curve and generating a continuous guaranteed power curve benchmark by interpolating the discrete data points of the contract-guaranteed power curve; calculating the power deviation between the measured power curve and the contract-guaranteed power curve for each wind speed point using the same wind speed coordinate; distinguishing and displaying data points belonging to the Bin that triggered the correction and data points belonging to the Bin that did not trigger the correction using different visual markers in the generated measured power curve; and additionally marking the correction amount for the corresponding record of the Bin that triggered the correction in the output deviation table.

[0010] Preferably, the skewness nonlinear correction module comprises: a skewness calculation unit, used to calculate the skewness value of the wind speed distribution within each Bin, wherein the skewness value characterizes the degree and direction of asymmetry in the wind speed data distribution within the Bin; a curvature calculation unit, used to calculate the curvature indicator of the measured power curve at the center wind speed position of the current Bin based on the arithmetic mean of the power of the current Bin and its adjacent Bins, wherein the curvature indicator characterizes the bending direction and degree of bending of the power curve at the center wind speed position of the Bin; a trigger judgment unit, connected to the outputs of the skewness calculation unit and the curvature calculation unit respectively, used to determine whether the current Bin simultaneously meets the preset skewness threshold condition, the preset curvature threshold condition, and the preset sample number condition based on the skewness value, the curvature indicator, and the sample number, and to determine the Bin that simultaneously meets the three conditions as the Bin for triggering correction; and a correction execution unit, connected to the output of the trigger judgment unit, used to perform correction calculation on the power arithmetic mean using the skewness value, the curvature indicator, and the wind speed variance to obtain the corrected power value.

[0011] Preferably, the correction execution unit calculates the corrected power value by subtracting the correction term from the power arithmetic mean. The correction term is composed of the product of the skewness value, the curvature indicator, and the wind speed variance. The mechanism of the correction term is as follows: when the wind speed distribution within a bin is skewed and the bin is located in the curved section of the power curve, the power arithmetic mean will produce a system deviation due to the coupling between the asymmetry of the wind speed distribution and the nonlinearity of the power curve. In the correction term, the positive or negative skewness value indicates whether the wind speed skewness is concentrated at the low or high wind speed end of the bin, the positive or negative curvature indicator indicates whether the power curve is concave upwards or downwards at that bin, and the wind speed variance indicates the amplification effect of the wind speed dispersion degree within the bin on the skewness effect. By subtracting the correction term from the power arithmetic mean, compensation for the system deviation is achieved.

[0012] Preferably, the curvature calculation unit calculates the curvature indicator in the following ways: For non-first and last bins, the curvature indicator is calculated using a central difference method based on the arithmetic mean of the power of the current bin, the arithmetic mean of the power of the previous bin, and the arithmetic mean of the power of the next bin, with the denominator being the square of the fixed wind speed interval; for the first and last bins, the curvature indicator is calculated using a one-sided difference method of forward difference or backward difference; or, a local curve is fitted based on the arithmetic mean of the power of multiple adjacent bins, and the curvature indicator is extracted from the fitted curve; or, the first and last bins are directly marked as not triggering correction.

[0013] Preferably, the trigger judgment unit determines whether the current Bin meets the correction trigger condition in the following ways: comparing the absolute value of the skewness value with a preset skewness threshold; comparing the absolute value of the curvature indicator with a preset curvature threshold; comparing the number of samples with a preset minimum number of samples; when the absolute value of the skewness value is greater than the preset skewness threshold, the absolute value of the curvature indicator is greater than the preset curvature threshold, and the number of samples is greater than the preset minimum number of samples, the current Bin is determined to meet the correction trigger condition, and the correction calculation needs to be performed by the correction execution unit; when any of the above three conditions is not met, the current Bin is determined not to meet the correction trigger condition, and the arithmetic mean of the power of the current Bin is directly output as the corrected power value.

[0014] Preferably, the correction execution unit is further configured to output correction marker information for each Bin that triggers correction. The correction marker information includes: correction trigger flag, correction amount, skewness value, and curvature indicator; wherein, the correction amount is equal to the product of the skewness value, curvature indicator, and wind speed variance; the correction marker information serves as an extended field of the statistical output record of the Bin and is output to the power curve generation and comparison analysis module along with the corrected power value of the Bin, so as to provide traceable marking when the measured power curve generation and comparison analysis results are output.

[0015] The technical effects and advantages of this invention are as follows: By embedding a skewed nonlinear correction module between traditional bin grouping statistics and power curve generation, this invention can automatically identify the arithmetic mean estimation deviation caused by the coupling between skewed wind speed distribution within a bin and local curvature of the power curve, and perform directional correction on bins that meet the triggering conditions. It requires no additional measurement hardware, does not change existing data acquisition methods or bin partitioning rules, and can effectively improve the accuracy of power estimation in high curvature sections near rated wind speed through pure data processing, making the measured power curve more realistically reflect the steady-state power output characteristics of large wind turbines. The system uses three threshold triggers—skewness, curvature, and sample size—to initiate correction only on reliable bins with sufficient data and significant coupling effects, avoiding overprocessing in small sample or low-skewness scenarios. Attached Figure Description

[0016] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts:

[0017] Figure 1 This is a schematic diagram of the module composition of the present invention; Figure 2 This is a flowchart of the internal processing of the skewness nonlinearity correction module of the present invention; Figure 3 This is a schematic diagram of the data processing timing of the present invention. Detailed Implementation

[0018] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0019] Reference Figures 1-3 As shown, this invention provides a power curve testing and analysis system for large wind turbine units, which is deployed as a whole on an industrial cloud platform. Each module of the system runs as a cloud service on the cloud computing nodes of the industrial cloud platform. Through the data interface of the industrial cloud platform, it receives wind speed and power data files that have been synchronized with the time uploaded by the local data acquisition terminal of the wind farm. The local data acquisition terminal of the wind farm is connected to the wind measurement equipment (such as lidar or wind measurement tower) and the wind turbine SCADA system respectively. After aligning and merging the collected wind speed data and power data with timestamps locally, it is uploaded to the industrial cloud platform through the network.

[0020] This system comprises six modules arranged sequentially in the data processing flow: a data import and configuration parsing module, a data validity screening module, a wind speed binning and statistics module, a skewness nonlinearity correction module, a power curve generation and comparative analysis module, and a report output module. Each module runs as a cloud service software program on the cloud computing nodes of the industrial cloud platform. The output of one module serves as the input for the next, forming a complete data processing pipeline. The report output module stores the generated test reports on the storage nodes of the industrial cloud platform and provides report viewing and download services to authorized remote user terminals.

[0021] The following example, using a real-world test case, details the specific implementation methods of each module in the system and their internal processing details.

[0022] I. Data Import and Configuration Parsing Module; Reading and Standardization of Wind Speed ​​and Power Data Files: In this implementation, a large wind turbine with a rated power of 6250kW and a rotor diameter of 200 meters is used as an example. Wind speed and power data have been collected and synchronized at the test site. The local data acquisition terminal at the wind farm uploads the merged wind speed and power data file to the industrial cloud platform via the network. The data file contains tens of thousands of ten-minute average data records.

[0023] The data import and configuration parsing module receives wind speed and power data files through the data interface of the industrial cloud platform. The module identifies the file format through the file reading interface. If it is an Excel file, it scans each table to locate the worksheet containing the data; if it is a CSV file, it identifies the field delimiters and encoding format.

[0024] After successful data retrieval, the module analyzes the header row of the data file and extracts the field names for each column. Because different wind farms and testing units use different SCADA systems and wind measurement equipment brands, the field naming conventions vary. For example, the wind speed column might be named "Wind Speed," "WindSpeed," etc. " The power column may be named "Power", "Power", etc. The module includes fields such as "ActivePower" and "ActivePower". It has a built-in field name mapping rule library, which pre-includes the field naming conventions of common SCADA systems and wind measurement equipment. Using a fuzzy string matching algorithm, the module maps the identified wind speed and power fields to the system's unified standard wind speed field name "WindSpeed" and standard power field name "Power", respectively, thus standardizing the fields and creating a standard wind speed and power data table.

[0025] The standard wind speed and power data table is stored in memory as a structured data table, with each row corresponding to a ten-minute data record and each column corresponding to a data field.

[0026] Reading and storing unit configuration parameters: In addition to the wind speed and power data file, users also need to upload the unit configuration parameter file through the user interface or configuration interface of the industrial cloud platform. Configuration parameters include, but are not limited to: unit model, such as EN200-6.25; hub height, such as 170 meters; air density, such as 1.225 kg / m³; cut-in wind speed, such as 3 m / s; cut-out wind speed, such as 25 m / s; invalid wind direction sector list: each invalid sector group includes a start angle and an end angle, and multiple groups can be set.

[0027] The module reads the configuration parameter file, extracts and stores the above parameter values ​​as system global configuration variables, and stores them in a global configuration object in memory for subsequent data validity screening module, skewness nonlinearity correction module and power curve generation and comparison analysis module to call as needed.

[0028] II. Data Validity Filtering Module: The data validity filtering module receives the standard wind speed and power data table and global configuration variables, performs validity checks on each data record in the data table row by row, and marks and removes data records that do not meet the test conditions as invalid.

[0029] The filtering logic of this module covers the following six rejection conditions. Each data record is rejected if it meets at least one of them: (1) Rejection of shutdown status; The data record carries a status signal field. When the status signal value is 0, it means that the wind turbine is in a shutdown state during the ten-minute period and is not connected to the grid for power generation. This record is not included in the power curve analysis and is rejected. (2) Rejection of fault status; The data record carries a fault signal field. When the fault signal value is 1, it means that the wind turbine has a fault during the ten-minute period. Its power output cannot represent the normal power generation performance of the unit and is rejected. (3) Rejection of power-limited operation status; The data record carries a power-limited signal field. When the power-limited signal value is 1, it means that the wind turbine is limited by the grid dispatch or wind farm control system during the ten-minute period. The output power is artificially suppressed and cannot reflect the true power characteristics of the unit and is rejected. (4) Rejection of invalid wind direction sectors; According to the invalid wind direction sector list set in the configuration parameters, the start angle and end angle of invalid sectors are read one by one. For each data record, its wind direction value is compared with the angle range of each invalid sector. If the wind direction value falls within any invalid sector range, it indicates that the wind measurement data may be affected by the wake of a nearby unit or other obstacles, and the record is removed. (5) Wind speed range exceeding limits is removed; according to the cut-in wind speed and cut-out wind speed in the configuration parameters, data records with wind speed values ​​less than the cut-in wind speed or greater than the cut-out wind speed are removed because the wind speed range exceeds the normal operating range of the unit. (6) Null or negative values ​​are removed; invalid data records with null values ​​or negative values ​​in the wind speed field or power field are directly removed. After all data records are filtered, the remaining valid data records constitute the valid wind speed and power data table and are output to the next module.

[0030] III. Wind Speed ​​Bin Grouping and Statistics Module: Bin Division; The wind speed bin grouping and statistics module receives the valid wind speed power data table and divides the entire wind speed range into multiple continuous wind speed intervals (Bin) according to a preset fixed wind speed interval. In this embodiment, the fixed wind speed interval is 0.5 m / s, consistent with industry practice.

[0031] The specific division method is as follows: using a step size of 0.5 m / s, starting from the cut-in wind speed and ending at the cut-out wind speed, multiple consecutive wind speed intervals are formed. Each bin has a lower limit and an upper limit wind speed, with the difference between the upper and lower limits being 0.5 m / s. The center wind speed of each bin is taken as the arithmetic mean of the lower and upper limit wind speed values ​​of that bin, i.e., the lower limit wind speed plus 0.25 m / s. For example, the center wind speed of the bin corresponding to the wind speed interval of 3.0 m / s to 3.5 m / s is 3.25 m / s; the center wind speed of the bin corresponding to the wind speed interval of 3.5 m / s to 4.0 m / s is 3.75 m / s, and so on.

[0032] Statistical calculation for each Bin: For each Bin obtained from the above division, the module iterates through all data records in the effective wind speed and power data table, filtering out records whose wind speed values ​​fall within the wind speed range of that Bin. Specifically, if a data record's wind speed value is greater than or equal to the lower limit of the wind speed for that Bin and less than the upper limit of the wind speed for that Bin (or uses a left-closed-right-open or left-open-right-closed interval assignment rule, maintaining consistency across the entire wind speed range), then that record belongs to that Bin.

[0033] After extracting all data records belonging to the same bin, a wind speed value sequence and a power value sequence for that bin are formed. Based on these two sequences, the module calculates the following statistics in sequence: (1) Wind speed arithmetic mean: The wind speed arithmetic mean is obtained by summing all wind speed values ​​in the bin and dividing by the number of samples, reflecting the central position of the wind speed data in the bin. (2) Power arithmetic mean: The power arithmetic mean is obtained by summing all power values ​​in the bin and dividing by the number of samples, which is the value directly used by existing methods to generate measured power curves. (3) Number of samples: The total number of data records belonging to the bin is counted, reflecting the sufficiency of the data in the bin. (4) Wind speed variance: The calculation method is as follows: For each wind speed value in the bin, the square of the difference between it and the wind speed arithmetic mean is calculated, and all square values ​​are summed and divided by the number of samples to obtain the wind speed variance, reflecting the dispersion of the wind speed data in the bin. (5) Wind speed standard deviation: The arithmetic square root of the wind speed variance is taken.

[0034] For each Bin, a statistical output record is constructed, containing the following fields: Bin number, Bin center wind speed, arithmetic mean of wind speed, arithmetic mean of power, sample size, wind speed variance, and wind speed standard deviation. The statistical output records of all Bins across the entire wind speed range constitute the Bin statistical output table, which is then output to the skewness nonlinearity correction module.

[0035] IV. Skewness Nonlinearity Correction Module: The skewness nonlinearity correction module is the core improvement module that distinguishes this invention from existing technologies. This module is internally divided into four functional units, in the following order: skewness calculation unit, curvature calculation unit, trigger judgment unit, and correction execution unit. The specific descriptions of each unit are as follows.

[0036] 4.1 Skewness Calculation Unit: The skewness calculation unit receives the wind speed value sequence, arithmetic mean, and standard deviation of wind speed for each Bin in the Bin statistical output table, and calculates the skewness value of the wind speed distribution within that Bin.

[0037] Skewness is a standard statistical measure used to measure the asymmetry of a distribution, belonging to the third-order standardized moments. The specific calculation process of the skewness calculation unit is as follows: for each wind speed value in the Bin, calculate the difference between it and the arithmetic mean of the wind speed, and then raise the cube of the difference; sum all the cubed values ​​and divide by the sample size to obtain the third-order central moment; then divide the third-order central moment by the cube of the standard deviation of the wind speed to obtain the skewness value.

[0038] The skewness value ranges from all real numbers. When the skewness value is positive, it indicates that the wind speed distribution within the Bin is positively skewed (right-skewed), meaning that the peak value of the wind speed data distribution is biased towards the low wind speed end of the Bin, with most wind speed data concentrated in the low wind speed range of the Bin, and a few data extending towards the high wind speed end to form a long tail. When the skewness value is negative, it indicates that the wind speed distribution within the Bin is negatively skewed (left-skewed), with most wind speed data concentrated at the high wind speed end of the Bin. The closer the skewness value is to zero, the more symmetrical the wind speed distribution within the Bin tends to be.

[0039] 4.2 Curvature Calculation Unit: The curvature calculation unit calculates the curvature indicator of the measured power curve at the center wind speed location of the current Bin based on the arithmetic mean of the power of the current Bin and its adjacent Bins. The curvature indicator is a numerical index that characterizes the local bending direction and degree of bending of the power curve.

[0040] For non-first and non-last bins in the middle of the sequence (i.e., cases where there is a preceding and following bin), the curvature indicator is calculated using the central difference method in numerical analysis. The specific process is as follows: obtain the arithmetic mean of the power of the current bin, the arithmetic mean of the power of the preceding adjacent bin, and the arithmetic mean of the power of the following adjacent bin; subtract twice the arithmetic mean of the power of the current bin from the arithmetic mean of the power of the following bin, add the arithmetic mean of the power of the preceding bin, and divide the result by the square of a fixed wind speed interval to obtain the curvature indicator. The mathematical essence of this method is a discrete approximation of the second derivative of the power curve at the wind speed at the center of the bin, and its accuracy is sufficient to meet the requirements of this application scenario.

[0041] For the first bin, since there is no preceding bin, the aforementioned center difference method cannot be used. The first bin can be processed using a forward difference method: obtain the arithmetic mean of the power of the current bin, the next bin, and the next two bins, and use the forward difference formula to approximately calculate the second derivative as a curvature indicator.

[0042] For the last Bin, since there is no next Bin, it can be processed symmetrically using the backward difference method: obtain the arithmetic mean of the power of the current Bin, the previous Bin, and the two previous Bins, and then use the backward difference formula to approximate the calculation.

[0043] Another alternative is to fit a quadratic or cubic polynomial curve in a local area based on the power arithmetic mean data points of the current Bin and its multiple neighboring Bins, and solve for the second derivative of the fitted curve at the wind speed at the center of the Bin as a curvature indicator.

[0044] Another simplified approach is to directly mark the first and last bins as not triggering corrections. Since the nonlinear abrupt change in the power curve of large wind turbines typically occurs near the rated wind speed, and the first and last bins are located in the low-wind-speed linear segment and the high-wind-speed saturation segment respectively, the curve curvature is relatively small, and the deviation caused by skewness and nonlinear coupling is negligible in engineering practice. Furthermore, the sample size of the first and last bins is often small, making it difficult to meet the conditions for triggering corrections.

[0045] The physical meaning of the curvature indicator is as follows: When the curvature indicator is greater than zero, it indicates that the power curve is concave and upward curved at the current Bin center wind speed position. At this time, within the Bin wind speed range, the power value corresponding to the low wind speed end is relatively low, and the power value corresponding to the high wind speed end is relatively high. When the curvature indicator is less than zero, it indicates that the power curve is concave and downward curved. At this time, the power value corresponding to the low wind speed end is relatively high, and the power value corresponding to the high wind speed end is relatively low. The larger the absolute value of the curvature indicator, the more severe the curvature of the curve at that point.

[0046] 4.3 Trigger Judgment Unit: The trigger judgment unit receives the skewness value output by the skewness calculation unit, the curvature indication output by the curvature calculation unit, and the number of samples for the current Bin. It then performs a joint conditional judgment based on three preset thresholds to determine whether the current Bin needs to be triggered for correction. The recommended values ​​for the three preset thresholds in this embodiment are: preset skewness threshold of 0.3, preset curvature threshold of 100, and preset minimum sample number of 30. The specific values ​​of the thresholds can be adaptively adjusted according to the model of the unit under test and the test conditions. For example, for models with a steeper power curve transition section, the curvature threshold can be appropriately reduced.

[0047] The specific judgment process for triggering the judgment unit is as follows: First, calculate the absolute value of the skewness value and compare it with a preset skewness threshold. If the absolute value of the skewness value is less than or equal to the preset skewness threshold, it indicates that the wind speed distribution within the Bin is relatively symmetrical. Even if the curve is curved, the arithmetic mean will not produce a significant estimation deviation, therefore no correction is needed, and the arithmetic mean of the power of this Bin is directly output as the corrected power value. Second, calculate the absolute value of the curvature indicator and compare it with a preset curvature threshold. If the absolute value of the curvature indicator is less than or equal to the preset curvature threshold, it indicates that the current Bin is in an approximately linear segment of the power curve, the curve curvature is slight, and the impact of wind speed asymmetry on the mean estimation is negligible, requiring no correction. Third, compare the sample size with a preset minimum sample size. If the sample size is less than or equal to the preset minimum sample size, it indicates that the data volume of this Bin is insufficient, the statistical estimation of the skewness value itself is not reliable enough, and correction should not be performed based on this.

[0048] The triggering judgment unit determines that the Bin meets the correction triggering conditions if and only if the above three conditions are met simultaneously: the absolute value of the skewness value is greater than the preset skewness threshold, the absolute value of the curvature indicator is greater than the preset curvature threshold, and the number of samples is greater than the preset minimum number of samples. Then, the correction execution unit performs the correction calculation.

[0049] If any of the three conditions is not met, the Bin is determined not to meet the correction trigger condition, and the power arithmetic mean is directly output as the corrected power value without generating a correction amount or correction flag.

[0050] 4.4 Correction Execution Unit and its Core Correction Algorithm: The correction execution unit is the core algorithm execution unit of this invention, used to perform correction calculations on the arithmetic mean power of Bins that meet the correction trigger conditions. The core correction algorithm used by the correction execution unit is: subtracting a correction term from the current Bin's arithmetic mean power to obtain the corrected power value. The correction term consists of the product of the skewness value, the curvature indicator, and the wind speed variance.

[0051] The core correction algorithm expression is as follows: In the formula: The power value is the corrected value, in kilowatts; This is the arithmetic mean of the current Bin's power, in kilowatts. The skewness value output by the skewness calculation unit is dimensionless. The curvature indicator output by the curvature calculation unit is expressed in kilowatts per meter per second squared. This represents the current wind speed variance in Bin, expressed in meters per second squared. (Correction term) The construction of the correction term is not a simple combination of existing statistical formulas or numerical analysis formulas, but a compensation quantity specially constructed based on an in-depth analysis of the physical mechanism of the technical problem. The working principle and construction logic of the correction term are explained in detail below.

[0052] In the existing Bin grouping statistical process, the arithmetic mean of power is directly taken as the output for each Bin. The implicit premise of this operation is that within the Bin wind speed range, there is a linear relationship between the power value corresponding to different wind speed values ​​and the frequency of the occurrence of that wind speed value within the range, that is, the wind speed distribution is uniform and the power curve is flat. However, in actual testing, this premise does not hold true for the section near the rated wind speed of large wind turbine units.

[0053] When the wind speed distribution within a bin is positively skewed (skewness value greater than zero), most wind speed data are concentrated at the low-speed end of that bin. If the bin happens to be located in the concave downward curve of the power curve (curvature indicator less than zero), then the low-speed end corresponds to a higher power value. The large concentration of wind speed data at the high-power low-speed end causes the high-power data at the low-speed end to be overrepresented in the arithmetic mean calculation, while the low-power data at the high-speed end is underrepresented, resulting in a systematic overestimation of the arithmetic mean. In this case, the skewness value is positive, the curvature indicator is negative, and their product is negative. The wind speed variance is positive, and the correction term is negative overall. Subtracting this negative value from the arithmetic mean is equivalent to adding the correction amount, resulting in an appropriate downward adjustment of the overestimated average value, reverting it to the true steady-state value.

[0054] Conversely, when the skewness value is negative and the curvature indicator is positive, the correction term can still provide compensation that is correct in direction and reasonable in magnitude.

[0055] The role of wind speed variance in the correction term is to adjust the correction magnitude. The larger the variance of wind speed values ​​within a bin, the higher the dispersion of the data, and the more obvious the distortion effect of the uneven distribution on the mean estimation, thus requiring a larger correction magnitude. Including the variance in the product allows the correction amount to adaptively adjust according to the degree of data dispersion.

[0056] In summary, the correction items By organically combining three statistical quantities of different orders—skewness (a standardized quantity of the third-order central moment), curvature (a discrete approximation of the second-order derivative), and variance (the second-order central moment)—in a product form, this method achieves, for the first time, targeted and quantitative compensation for the coupling deviation between wind speed distribution skewness and curve curvature in the specific application scenario of power curve testing. The formula is concise, involving only one multiplication and one subtraction operation of four variables, resulting in extremely low computational overhead and making it perfectly suitable for engineering implementation.

[0057] 4.5 Output of Correction Flag Information; While outputting the corrected power value, the correction execution unit also generates correction flag information for each Bin that triggers the correction. The correction flag information includes the following four fields: (1) Correction Trigger Flag: a Boolean flag that indicates whether the Bin has triggered the correction calculation; (2) Correction Amount: i.e., the correction item The specific value is recorded, and the magnitude of the correction is recorded; (3) Skewness value: the calculated skewness value of the Bin, which records the distribution skewness state when the correction is triggered; (4) Curvature indicator: the calculated curvature indicator value of the Bin, which records the curve bending state when the correction is triggered. The correction mark information is added as an extended field to the statistical output record of the Bin, and is output to the power curve generation and comparison analysis module along with the corrected power value, so as to realize the data traceability of the entire correction process.

[0058] V. Power Curve Generation and Comparison Analysis Module: The power curve generation and comparison analysis module receives the complete Bin statistical output table from the skewed nonlinear correction module, and generates the measured power curve based on the Bin center wind speed and corrected power value of each Bin in the table.

[0059] The module plots the data points corresponding to each Bin in a Cartesian coordinate system, with the wind speed at the center of the Bin as the horizontal axis variable and the corrected power value as the vertical axis variable. The data points are connected by line segments in order of increasing wind speed at the center of the Bin to form a continuous broken line, which is the measured power curve.

[0060] Meanwhile, the module receives contract-guaranteed power curve data imported by the user. This curve is typically provided by the wind turbine manufacturer in the form of a discrete data table, containing multiple wind speed points and their corresponding guaranteed power values. The intervals between these wind speed points may vary across different wind speed ranges. The module performs interpolation processing on the discrete data points of the contract-guaranteed power curve, using either piecewise linear interpolation or conformal interpolation to ensure that the interpolation results do not exceed or oscillate within the power plateau region (rated power range) of the guaranteed power curve, generating a continuous baseline for the guaranteed power curve.

[0061] The module plots the measured power curve and the baseline of the guaranteed power curve on the same coordinate system. Using the same horizontal coordinate grid, it reads the power value at each wind speed point on the two curves and calculates the difference to obtain the power deviation value at each wind speed point.

[0062] In terms of visualization, the module uses different visual markers to distinguish between data points of Bins that trigger correction and those that do not. For example, Bin data points that do not trigger correction are drawn with solid circles, while Bin data points that trigger correction are drawn with solid triangles, and the meaning of the two is explained in the legend. At the same time, the module organizes information such as the deviation value and correction amount of the Bin into a deviation comparison table, and highlights or annotates the rows of Bins that have triggered correction in the table.

[0063] VI. Report Output Module: The report output module summarizes the processing results of all the above modules and integrates them to form a complete power curve test analysis report.

[0064] The report includes, but is not limited to, the following parts: (1) Unit parameter overview table: listing the basic parameters of the tested unit, such as model, rated power, impeller diameter, hub height, cut-in wind speed, and cut-out wind speed; (2) Data validity overview: displaying the total number of original data records, the number of valid data records, the number of data records removed, and the percentage distribution of various reasons for removal; (3) Bin detailed statistics table: listing the complete information of each Bin in tabular form, including Bin serial number, Bin center wind speed, wind speed arithmetic mean, power arithmetic mean, corrected power value, sample size, skewness value, curvature indication, correction trigger flag, correction amount, and deviation from guaranteed power. The table presents the intermediate and final data from uncorrected to corrected, meeting the transparency and traceability requirements of the test report review; (4) Power curve comparison chart: The measured power curve and the contract guaranteed power curve are plotted simultaneously in the same coordinate system, and the corrected Bin data points and the uncorrected Bin data points are marked; (5) Wind energy utilization coefficient curve chart; (6) Annual power generation comparison table.

[0065] The report output module stores the generated test reports on the storage nodes of the industrial cloud platform and provides report viewing and download services to authorized remote user terminals. Remote users can log in to the industrial cloud platform through a web browser or client software to view or download the test reports in PDF format, or export the data tables separately as Excel files for further analysis.

[0066] The system of this invention, while being fully compatible with the bin division method in existing test specifications and without adding any additional measurement hardware, solves the problem of arithmetic mean estimation deviation caused by the coupling between the skewed wind speed distribution within the bin and the bending of the power curve in high-curvature wind speed sections of large wind turbines by embedding a skewed nonlinear correction module in the data processing pipeline.

[0067] The correction process employs a three-threshold joint triggering mechanism to ensure that corrections are performed only on reliable bins with high skewness, high curvature, and large sample sizes, avoiding over-correction on low-reliability data. The correction algorithm uses a concise product construct to achieve coordinated compensation of skewness, curvature, and variance, with clear physical meaning and minimal computational overhead.

[0068] The entire correction process is completely transparent. Whether each bin triggers a correction, the magnitude of the correction, and the original values ​​of skewness and curvature are all recorded and traceable. The application of this system effectively improves the data accuracy of the power curve of large wind turbine units in the critical contract verification range near rated wind speed, enhancing the scientific rigor, objectivity, and probative value of the test conclusions in commercial evaluations.

[0069] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A power curve testing and analysis system for large wind turbine units, characterized in that, The system, deployed on an industrial cloud platform, includes: a data import and configuration parsing module, used to receive time-synchronized wind speed and power data files uploaded by local data acquisition terminals of wind farms through the data interface of the industrial cloud platform, and convert the wind speed and power data files into a standard wind speed and power data table in a unified format within the system; a data validity filtering module, used to receive the standard wind speed and power data table, and according to the unit configuration parameters and preset filtering rules, remove invalid data records that do not meet the test conditions, and output a valid wind speed and power data table; a wind speed bin grouping and statistics module, used to divide each data record in the valid wind speed and power data table into a corresponding wind speed interval bin according to a preset fixed wind speed interval, and calculate the arithmetic mean of wind speed, the arithmetic mean of power, the sample size, and the wind speed variance for each bin; and a skewness nonlinearity correction module, used to determine whether each bin meets the preset correction trigger conditions, and to correct the nonlinearity ... For a given Bin that triggers the correction condition, the arithmetic mean of its power is calculated using the Bin's wind speed skewness, power curve curvature indicator, and wind speed variance to obtain a corrected power value. The curvature indicator represents the bending direction and degree of the power curve at the center wind speed position of the Bin. For Bins that do not meet the correction trigger condition, the arithmetic mean of their power is directly output as the corrected power value. The power curve generation and comparison analysis module generates a measured power curve based on the Bin's center wind speed and the corrected power value, and performs a point-by-point comparison analysis between the measured power curve and the input contract-guaranteed power curve, outputting the comparison analysis results. The report output module summarizes the measured power curve, the comparison analysis results, and the correction information of each Bin to generate a test report, stores the test report on the storage node of the industrial cloud platform, and provides authorized remote user terminals with the opportunity to view and download the report.

2. The large-scale wind turbine power curve testing and analysis system according to claim 1, characterized in that, The data import and configuration parsing module is specifically used for: reading the wind speed and power data file through the data interface of the industrial cloud platform, automatically identifying the wind speed and power fields in the file, and mapping the wind speed and power fields to the system's internal standard wind speed field name and standard power field name to form the standard wind speed and power data table; reading the unit configuration parameters, which at least include: unit model, hub height, air density, cut-in wind speed, cut-out wind speed, and the start and end angles of invalid wind direction sectors; and storing the unit configuration parameters as system global configuration variables for use by the data validity filtering module and the power curve generation and comparison analysis module.

3. The large-scale wind turbine power curve testing and analysis system according to claim 2, characterized in that, The data validity filtering module is specifically used to mark and remove data records in the standard wind speed and power data table that meet at least one of the following rejection conditions: data records where the wind turbine is in a shutdown state; data records where the wind turbine is in a fault state; data records where the wind turbine is in a power-limited operation state. Data records where the wind direction value falls within the invalid wind direction sector; Data records where the wind speed value exceeds the range from the cut-in wind speed to the cut-out wind speed; Data records where the wind speed or power field is empty or negative.

4. The large-scale wind turbine power curve testing and analysis system according to claim 1, characterized in that, The wind speed Bin grouping and statistics module is specifically used to: divide the effective wind speed power data table into multiple wind speed intervals Bin at a preset fixed wind speed interval, and the center wind speed of each Bin is the midpoint between the upper limit and lower limit of the wind speed in that Bin. For each Bin, iterate through all data records in the effective wind speed and power data table where the wind speed value falls into that Bin, and extract the corresponding wind speed value sequence and power value sequence; calculate the arithmetic mean of the wind speed based on the wind speed value sequence; calculate the arithmetic mean of the power based on the power value sequence; count the number of data records falling into that Bin, and use this as the sample size. The wind speed variance is calculated based on the wind speed value sequence and the wind speed arithmetic mean. The Bin number, the wind speed at the center of the Bin, the arithmetic mean of the wind speed, the arithmetic mean of the power, the number of samples, the variance of the wind speed, and the standard deviation of the wind speed are recorded as the statistical output records for each Bin.

5. The large-scale wind turbine power curve testing and analysis system according to claim 1, characterized in that, The power curve generation and comparison analysis module is specifically used for: connecting the data points corresponding to each Bin sequentially with the center wind speed of each Bin as the abscissa and the corrected power value of each Bin as the ordinate to form the measured power curve; receiving the contract-guaranteed power curve and generating a continuous guaranteed power curve benchmark by interpolating the discrete data points of the contract-guaranteed power curve; calculating the power deviation between the measured power curve and the contract-guaranteed power curve point by point using the same wind speed coordinate; and distinguishing and displaying data points belonging to the Bin that triggered the correction and data points belonging to the Bin that did not trigger the correction using different visual markers in the generated measured power curve. In the output deviation table, the correction amount is additionally marked for the corresponding record of the Bin that triggered the correction.

6. The large-scale wind turbine power curve testing and analysis system according to any one of claims 1-5, characterized in that, The skewness nonlinear correction module includes: a skewness calculation unit, used to calculate the skewness value of the wind speed distribution within each bin, wherein the skewness value characterizes the degree and direction of asymmetry in the wind speed data distribution within the bin; a curvature calculation unit, used to calculate the curvature indication of the measured power curve at the center wind speed position of the current bin based on the arithmetic mean of the power of the current bin and its adjacent bins; a trigger judgment unit, connected to the outputs of the skewness calculation unit and the curvature calculation unit respectively, used to determine whether the current bin simultaneously meets the preset skewness threshold condition, the preset curvature threshold condition, and the preset sample number condition based on the skewness value, the curvature indication, and the sample number, and to determine the bin that simultaneously meets all three conditions as the bin for triggering correction; and a correction execution unit, connected to the output of the trigger judgment unit, used to perform correction calculation on the arithmetic mean of the power using the skewness value, the curvature indication, and the wind speed variance to obtain the corrected power value.

7. The large-scale wind turbine power curve testing and analysis system according to claim 6, characterized in that, The correction execution unit calculates the corrected power value by subtracting the correction term from the power arithmetic mean. The correction term is composed of the skewness value, the curvature indicator, and the wind speed variance. The mechanism of the correction term is as follows: when the wind speed distribution within a bin is skewed and the bin is located in the curved section of the power curve, the power arithmetic mean will experience a systemic deviation due to the coupling between the asymmetric wind speed distribution and the nonlinearity of the power curve. In the correction term, the positive or negative skewness value indicates whether the wind speed skewness is concentrated at the low or high wind speed end of the bin; the positive or negative curvature indicator indicates whether the power curve is concave upwards or downwards at that bin; and the wind speed variance indicates the amplification effect of the wind speed dispersion within the bin on the skewness effect. Compensation for the systemic deviation is achieved by subtracting the correction term from the power arithmetic mean.

8. The large-scale wind turbine power curve testing and analysis system according to claim 7, characterized in that, The curvature calculation unit calculates the curvature indicator in the following way: for non-first and last Bins, the curvature indicator is calculated using the central difference method based on the power arithmetic mean of the current Bin, the power arithmetic mean of the previous Bin, and the power arithmetic mean of the next Bin. The denominator of the calculation is the square of the fixed wind speed interval. For the first and last bins, the curvature indicator is calculated using a one-sided difference method of forward difference or backward difference; or, a local curve is fitted based on the arithmetic mean of the power of multiple adjacent bins, and the curvature indicator is extracted from the fitted curve. Alternatively, you can directly mark the first and last Bins as not triggering the fix.

9. The large-scale wind turbine power curve testing and analysis system according to claim 6, characterized in that, The specific method by which the triggering judgment unit determines whether the current Bin meets the correction triggering condition is to compare the absolute value of the skewness value with a preset skewness threshold. The absolute value of the curvature indicator is compared with a preset curvature threshold; the number of samples is compared with a preset minimum number of samples; when the absolute value of the skewness value is greater than the preset skewness threshold, the absolute value of the curvature indicator is greater than the preset curvature threshold, and the number of samples is greater than the preset minimum number of samples, it is determined that the current Bin meets the correction trigger condition, and the correction calculation needs to be performed by the correction execution unit; when any of the above three conditions is not met, it is determined that the current Bin does not meet the correction trigger condition, and the arithmetic mean of the power of the current Bin is directly used as the corrected power value for output.

10. The large-scale wind turbine power curve testing and analysis system according to claim 6, characterized in that, The correction execution unit is further configured to output correction marker information for each Bin that triggers correction. The correction marker information includes: a correction trigger flag, a correction amount, the skewness value, and the curvature indicator. The correction amount is equal to the product of the skewness value, the curvature indicator, and the wind speed variance. The correction marker information serves as an extended field of the statistical output record of the Bin and is output to the power curve generation and comparison analysis module along with the corrected power value of the Bin, so as to provide traceable marking during the generation of the measured power curve and the output of the comparison analysis results.

Images