Abnormal data processing method and abnormal data processing device for wind power plant

Abstract

The invention provides an abnormal data processing method and an abnormal data processing device for a wind power plant. The abnormal data processing method comprises the steps that wind speed-power data of a wind power plant are acquired, and the wind speed-power data comprise wind speed and power corresponding to the wind speed; constructing a data set for fitting a curve based on the acquired wind speed-power data; based on the data set, fitting a preset fitting function to determine parameters of the fitting function, and generating the fitting curve corresponding to the fitting function with the parameters; identifying the wind speed-power data within the credible data range of the fitting curve as normal data, and identifying the wind speed-power data outside the credible data range of the fitting curve as abnormal data; in response to identifying the anomalous data, the data set is updated by rejecting the anomalous data, and the step of generating the fitted curve is iteratively performed until no anomalous data is identified.

Description

Technical Field

[0001] The present disclosure relates to the field of wind power technology, and more specifically, to an abnormal data processing method and abnormal data processing device for a wind farm. Background Art

[0002] At present, the scale of new energy is constantly expanding, and power regulation requires the power prediction level to be continuously improved. Among them, wind power generation is limited by geographical conditions and environmental factors, and there are many challenges. The data quality of wind farms is one of them. Due to factors such as wind turbine shutdown maintenance, incorrect data collection, and transmission anomalies, there are a large number of abnormal values ​​in the measured power data collected by wind turbines. These data seriously affect the modeling and prediction effect of the wind farm power prediction model. Therefore, how to effectively deal with abnormal data in wind farms is an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present disclosure is to provide an abnormal data processing method and abnormal data processing device for a wind farm.

[0004] One aspect of the present disclosure provides a method for processing abnormal data of a wind farm, the method comprising: acquiring wind speed-power data of the wind farm, wherein the wind speed-power data comprises wind speed and power corresponding to the wind speed; constructing a data set for fitting a curve based on the acquired wind speed-power data; fitting a preset fitting function based on the data set to determine parameters of the fitting function, and generating the fitting curve corresponding to the fitting function with the parameters; identifying the wind speed-power data within a credible data range of the fitting curve as normal data, and identifying the wind speed-power data outside the credible data range of the fitting curve as abnormal data; in response to identifying the abnormal data, updating the data set by eliminating the abnormal data, and iteratively executing the step of generating the fitting curve until no abnormal data is identified.

[0005] Optionally, the data set includes a first data set corresponding to a low wind speed segment, a second data set corresponding to an intermediate wind speed segment, and a third data set corresponding to a full-power wind speed segment, wherein the low wind speed segment corresponds to a wind speed segment that the wind turbine cannot cut in, and the power in the first data set is set to 0, wherein the intermediate wind speed segment corresponds to a wind speed segment that the wind turbine can cut in and has not yet reached full power; wherein the full-power wind speed segment corresponds to a wind speed segment where the wind turbine is fully powered, and the power in the third data set is set to maximum power.

[0006] Optionally, a predetermined number of wind speed-power data are collected at predetermined wind speed intervals in an intermediate wind speed segment as a second data set, wherein if the number of wind speed-power data collected at predetermined wind speed intervals in the intermediate wind speed segment is less than the predetermined number, the predetermined number of wind speed-power data is obtained by upsampling the collected wind speed-power data.

[0007] Optionally, the size of the credible data range of the fitting curve corresponding to the first data set, the size of the credible data range of the fitting curve corresponding to the second data set, and the size of the credible data range of the fitting curve corresponding to the third data set increase in sequence.

[0008] Optionally, the credible data range of the fitting curve is determined based on the installed capacity of the wind farm.

[0009] Optionally, when the wind speed is less than the first wind speed, the lower limit of the credible data range of the fitting curve is 0, and the upper limit of the credible data range of the fitting curve is the sum of the first value and the fitting value of the fitting curve, wherein the first value corresponds to the product of the installed capacity and the second value; wherein, when the wind speed is greater than or equal to the first wind speed and less than the second wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the third value and the fourth value, and the upper limit of the credible data range of the fitting curve is the smaller value between the installed capacity and the fifth value, wherein the third value corresponds to the product of the installed capacity and the sixth value. , the fourth value corresponds to the difference between the fitting value of the fitting curve and the product of the installed capacity and the seventh value, and the fifth value corresponds to the sum of the fitting value of the fitting curve and the product of the installed capacity and the eighth value; wherein, when the wind speed is greater than or equal to the second wind speed and less than the third wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the ninth value and the tenth value, and the upper limit of the credible data range of the fitting curve is the installed capacity, wherein the ninth value corresponds to the product of the installed capacity and the eleventh value, and the tenth value corresponds to the difference between the fitting value of the fitting curve and the product of the installed capacity and the twelfth value.

[0010] Optionally, the abnormal data processing method further includes: acquiring wind speed in the abnormal data; acquiring new power by substituting the wind speed in the abnormal data into the fitting curve; and using the wind speed in the abnormal data and the acquired new power as normal data.

[0011] Optionally, the wind speed-power data includes wind speed-power data during a preset time period before and after the wind turbine installation is changed or wind speed-power data in the current season.

[0012] One aspect of the present disclosure provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, any of the abnormal data processing methods described above is implemented.

[0013] One aspect of the present disclosure provides an abnormal data processing device for a wind farm, the abnormal data processing device comprising: a processor; a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements any abnormal data processing method as described above.

[0014] According to the abnormal data processing method of various exemplary embodiments, a generalized logic function with a more flexible S-curve is applied to the fitting of the wind farm-based data set, thereby achieving better fitting.

[0015] According to the abnormal data processing method of various example embodiments, taking into account that in a wind farm, as the wind speed increases, the degree of power fluctuation will generally increase accordingly, therefore, by dynamically adjusting the size of the credible data range of the fitting curve (for example, increasing the size of the credible data range as the wind speed increases), normal data and abnormal data can be better distinguished.

[0016] According to the abnormal data processing method of various exemplary embodiments, normal data and abnormal data can be more accurately identified by setting the upper and lower bounds of the credible data range of the fitting curve according to the magnitude of the wind speed and the installed capacity of the wind farm.

[0017] According to the abnormal data processing method of various exemplary embodiments, the upper and lower bounds of the credible data range of the fitting curve may be determined by adjusting the parameters of the generalized logic function, so that normal data and abnormal data may be distinguished more accurately.

[0018] According to the abnormal data processing method of various exemplary embodiments, considering the interference caused by abnormal data in the fitting process, the fitting curve may be biased. Therefore, after the outliers are eliminated, you can perform the sum by continuous iteration until no new outliers are judged to be finished, and the critical outliers can be better eliminated through multiple iterations.

[0019] In order to avoid or reduce the elimination of useful data, the abnormal data processing method of the wind farm according to the exemplary embodiment of the present disclosure may also include data correction to achieve data enhancement, which is conducive to improving the accuracy of wind power modeling. This is very effective when facing the following problems: some electric fields have insufficient historical data preservation and too much abnormal data. If only normal data can be used for wind power modeling, the model may have poor stability due to too little modeling data. Correcting abnormal data to normal data and using it together with other data for modeling can improve the stability of the model and improve the prediction accuracy.

[0020] The data before abnormal data and installation changes cannot be used in modeling. If all data are deleted, a large amount of data will be lost. According to the abnormal data processing method of the exemplary embodiment of the present disclosure, the abnormal data is effectively corrected to normal data, which is conducive to modeling and improves prediction accuracy. The present disclosure proposes that the wind speed of the abnormal data point can be corrected to the normal value by fitting the generalized logic function.

[0021] According to the abnormal data processing method of the exemplary embodiment of the present disclosure, by adopting the latest data in segments and combining the data construction of low wind and high wind, the latest wind speed-power curve after the installation change can be fitted.

[0022] According to the exemplary embodiment of the present disclosure, the abnormal data processing method can automatically fit the latest data curve according to the latest data, so as to obtain data that is more suitable for the current season and meet the requirements of automatically adjusting the abnormal data processing method according to the season. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and other objects and features of the present disclosure will become more apparent through the following description in conjunction with the accompanying drawings which exemplarily illustrate an example, in which:

[0024] Figure 1A A flow chart showing a method for processing abnormal data of a wind farm according to an exemplary embodiment of the present disclosure;

[0025] Figure 1B A schematic diagram showing acquired data according to an exemplary embodiment of the present disclosure;

[0026] Figure 2 A flow chart showing a construction data set of a wind farm according to an exemplary embodiment of the present disclosure;

[0027] Figure 3 A flowchart showing a method of constructing a trusted data range according to an exemplary embodiment of the present disclosure is shown;

[0028] Figure 4 A flow chart showing a method for processing abnormal data of a wind farm according to an exemplary embodiment of the present disclosure;

[0029] Figure 5 A scatter plot showing measured wind speed and measured power of a wind farm according to an exemplary embodiment of the present disclosure;

[0030] Figure 6 A data set for fitting a curve of a wind farm according to an exemplary embodiment of the present disclosure is shown;

[0031] Figure 7 A fitting curve corresponding to a fitted generalized logic function according to an exemplary embodiment of the present disclosure is shown;

[0032] Figure 8 A schematic diagram showing abnormal data identification according to an exemplary embodiment of the present disclosure;

[0033] Fig. 9 A schematic diagram showing abnormal data correction according to an exemplary embodiment of the present disclosure;

[0034] Fig.10 This is a data overlay diagram of a wind farm from January to February and from July to August according to an exemplary embodiment of the present disclosure;

[0035] Fig.11 The filtering effect on the wind farm data from January to February according to an exemplary embodiment of the present disclosure;

[0036] Fig.12 This is the filtering effect on the wind farm data from July to August according to the exemplary embodiment of the present disclosure;

[0037] Fig.13 A flowchart showing a method for processing abnormal data according to an example embodiment of the present disclosure;

[0038] Fig.14 A block diagram showing an abnormal data processing apparatus according to an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0039] Hereinafter, various example embodiments will be described in detail with reference to the accompanying drawings. These example embodiments are provided to specifically describe the technical ideas of the disclosed concept, but are not intended to restrict the scope of the disclosed concept. It should be understood that all modifications that can be easily obtained by those skilled in the art from this specification and these example embodiments fall within the scope of the disclosed concept.

[0040] It will be understood that when the term “comprising” is used in this specification, it specifies the presence of stated features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0041] It will also be understood that although the terms "first", "second", "third", etc. may be used herein to describe various elements, components, regions, layers and / or parts, these elements, components, regions, layers and / or parts should not be limited by these terms. These terms are only used to distinguish one element, component, region, layer or part from another element, component, region, layer or part. As used herein, the term "and / or" includes any and all combinations of one or more related listed items. When a statement such as "at least one of..." follows a column of elements, the statement modifies the entire column of elements, rather than modifying a single element of the column.

[0042] In the present disclosure, abnormal data of a wind farm may refer to data of wind speed-power mismatch in wind speed-power data, wherein the wind speed-power data includes a wind speed and a power corresponding to the wind speed.

[0043] Figure 1A A flow chart showing a method for processing abnormal data of a wind farm according to an exemplary embodiment of the present disclosure is shown.

[0044] Reference Figure 1A In step S110, wind speed-power data of the wind farm may be acquired, wherein the wind speed-power data includes the wind speed and the power corresponding to the wind speed.

[0045] In one example, the wind speed-power data of the wind farm may be the measured data of the wind farm. In another example, the acquired data may also include the time corresponding to the measured wind speed and / or the measured power. In other words, the measured data may optionally include time, measured wind speed, and measured power.

[0046] In addition, optionally, Figure 1B As shown, both basic wind farm information and measured data can also be obtained.

[0047] In one embodiment, the wind speed-power data includes wind speed-power data during a preset time period before and after the wind turbine installed capacity is changed, or wind speed-power data in the current season. For example, the current season may be a season in which the power generation pattern changes due to seasonal changes. The installed capacity change may indicate that the wind farm is expanded or connected to the grid, and its installed capacity has changed. Figures 5 to 12 The wind speed-power data during a preset time period before and after the wind turbine installation change or the wind speed-power data of the current season is described in more detail.

[0048] In step S120, a data set for fitting a curve may be constructed based on the acquired wind speed-power data.

[0049] The data set for fitting the curve can be constructed in various ways. For example, all the acquired wind speed-power data can be used to construct the data set for fitting the curve.

[0050] In one embodiment, the data set used for fitting the curve may be constructed to include a first data set corresponding to a low wind speed segment, a second data set corresponding to an intermediate wind speed segment, and a third data set corresponding to a full-power wind speed segment. Here, the low wind speed segment corresponds to a wind speed segment that the wind turbine cannot cut in, and the power in the first data set is set to 0. The intermediate wind speed segment corresponds to a wind speed segment that the wind turbine can cut in and has not yet reached full power. The full-power wind speed segment corresponds to a wind speed segment where the wind turbine is fully powered, and the power in the third data set is set to the maximum power.

[0051] In this embodiment, the power of the wind speed section that the wind turbine cannot cut in can be set to 0 and the power of the wind speed section that the wind turbine is fully powered can be set to the maximum power to reduce the influence of the abnormal data of the wind speed section that the wind turbine cannot cut in and the abnormal data of the wind speed section that the wind turbine is fully powered on the fitting curve. Therefore, the abnormal data in the wind speed-power data can be processed more effectively. Figure 2 This embodiment is described as an example.

[0052] In one example, a predetermined amount of wind speed-power data is collected as the second data set at a predetermined wind speed interval in the intermediate wind speed segment. If the amount of wind speed-power data collected at the predetermined wind speed interval in the intermediate wind speed segment is less than the predetermined amount, the predetermined amount of wind speed-power data is obtained by upsampling the collected wind speed-power data.

[0053] In step S130 , based on the data set, a preset fitting function is fitted to determine the parameters of the fitting function, and a fitting curve corresponding to the fitting function with the parameters is generated.

[0054] In one embodiment, the fitting function may include a generalized logic function. As shown in equation (1) below, the generalized logic function is an extension of the logic function, has a more flexible S-curve, and has the following form:

[0055]

[0056] Among them, x is the independent variable, Y is the dependent variable, v is the hyperparameter, and A, B, C, K, and Q are parameters. More specifically, A is used to control the lower (left) asymptote, B is used to control the growth rate of the curve, C is used to control the upper (right) asymptote, K is used to control the opening angle of the curve, and Q is used to control the left and right displacement.

[0057] Considering the data characteristics of the wind farm (for example, the power value must be positive and the lower asymptote is 0), A can be set to 0; and v can be set to 1.

[0058] Accordingly, the generalized logic function according to the embodiment of the present disclosure can be simplified to the following equation (2):

[0059]

[0060] The generalized logic function according to the embodiment of the present disclosure has a more flexible S-curve, and is more suitable for the application scenario of wind farms when used for fitting a data set based on wind farms, and therefore can achieve better fitting.

[0061] However, the preset fitting function is not limited to the generalized logical function. For example, the preset fitting function may include but is not limited to an exponential function, a logarithmic function, a power function, and the like.

[0062] In one example, the parameters of the fitting function can be determined using the least squares method. The least squares method is a numerical optimization method. It finds the best function match for the data by minimizing the sum of the squares of the errors (the difference between the true target object and the fitted target object). The specific implementation can utilize the optimize.curve_fit function in the Python scientific computing package scipy. Scipy is an advanced scientific computing library that includes functions such as optimization, linear algebra, integration, interpolation, fitting, special functions, fast Fourier transform, signal processing and image processing, ordinary differential equation solving, and other commonly used calculations in science and engineering. However, the method for determining the parameters of the fitting function in this disclosure is not limited to the least squares method and can also be any other existing fitting method.

[0063] In step S140, the wind speed-power data within the credible data range of the fitting curve is identified as normal data, and the wind speed-power data outside the credible data range of the fitting curve is identified as abnormal data.

[0064] That is to say, the credible data range can be the basis for distinguishing whether the wind speed-power data is normal data or abnormal data. The credible data range can be determined in various ways. In one example, the preset range corresponding to the fitting curve can be set as the credible data range. For example, as a non-limiting example, assuming the fitting curve is y = x, the credible data range of the fitting curve can be x - 1 < y < x + 1. In another example, the credible data range can be determined according to a fixed value or a variable value z determined based on the historical data of the wind farm. For example, as a non-limiting example, assuming the fitting curve is y = x, the credible data range of the fitting curve can be x - z < y < x + z.

[0065] In one embodiment, when the data set includes a first data set corresponding to a low wind speed section, a second data set corresponding to an intermediate wind speed section, and a third data set corresponding to a full load wind speed section, the size of the credible data range of the fitting curve corresponding to the first data set, the size of the credible data range of the fitting curve corresponding to the second data set, and the size of the credible data range of the fitting curve corresponding to the third data set increase in sequence.

[0066] In this embodiment, considering that in a wind farm, as the wind speed increases, the degree of power fluctuation usually also increases accordingly. Therefore, by dynamically adjusting the size of the credible data range of the fitting curve (for example, increasing the size of the credible data range as the wind speed increases), normal data and abnormal data can be better distinguished.

[0067] That is to say, according to the abnormal data processing method of the exemplary embodiment of the present disclosure, different wind speeds can correspond to credible data ranges. In the wind speed interval between wind turbine cut-in and full power, the average wind speed is getting higher and higher, but not all wind turbine head wind speeds can reach higher wind speeds. Therefore, the higher the wind speed, the wider the corresponding power credible range should be set. When the wind speed exceeds the wind speed of the wind turbine cut-out, there is no need to identify abnormal values ​​for the wind speed, and the power attenuation at this moment is normal.

[0068] In one embodiment, the credible data range of the fitting curve is determined based on the installed capacity of the wind farm.

[0069] In this embodiment, considering that in a wind farm, as the installed capacity changes, the degree of power fluctuation usually changes accordingly, therefore, by determining the credible data range of the fitting curve based on the installed capacity of the wind farm, normal data and abnormal data can be better distinguished.

[0070] In an example embodiment, when the wind speed is less than the first wind speed, the lower limit of the credible data range of the fitting curve is 0, and the upper limit of the credible data range of the fitting curve is the sum of the first value and the fitting value of the fitting curve, wherein the first value corresponds to the product of the installed capacity and the second value. The second value can be predefined in various ways. For example, the second value can be predefined by referring to the historical wind speed-power data of the wind farm.

[0071] In addition, when the wind speed is greater than or equal to the first wind speed and less than the second wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the third value and the fourth value, and the upper limit of the credible data range of the fitting curve is the smaller value between the installed capacity and the fifth value, wherein the third value corresponds to the product of the installed capacity and the sixth value, the fourth value corresponds to the difference between the fitted value of the fitting curve and the product of the installed capacity and the seventh value, and the fifth value corresponds to the sum of the fitted value of the fitting curve and the product of the installed capacity and the eighth value. The sixth value, the seventh value, and the eighth value can be predefined in various ways. For example, the sixth value, the seventh value, and the eighth value can be predefined by referring to the historical wind speed-power data of the wind farm.

[0072] In addition, when the wind speed is greater than or equal to the second wind speed and less than the third wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the ninth value and the tenth value, and the upper limit of the credible data range of the fitting curve is the installed capacity, wherein the ninth value corresponds to the product of the installed capacity and the eleventh value, and the tenth value corresponds to the difference between the fitting value of the fitting curve and the product of the installed capacity and the twelfth value. The eleventh value and the twelfth value can be predefined in various ways. For example, the eleventh value and the twelfth value can be predefined by referring to the historical wind speed-power data of the wind farm.

[0073] In this exemplary embodiment, by setting the upper and lower bounds of the credible data range of the fitting curve according to the magnitude of the wind speed and the installed capacity of the wind farm, normal data and abnormal data can be identified more accurately.

[0074] This exemplary embodiment will be described in more detail below with reference to the drawings.

[0075] In yet another embodiment, the upper and lower bounds of the credible data range of the fitted curve are determined by adjusting parameters of the generalized logistic function.

[0076] In this embodiment, the upper and lower bounds of the credible data range of the fitting curve are determined by adjusting the parameters of the generalized logistic function, so that normal data and abnormal data can be distinguished more accurately.

[0077] The various embodiments described above may be separated and / or combined.

[0078] In step S150 , in response to identifying abnormal data, the data set is updated by removing the abnormal data, and the step of generating a fitting curve is iteratively performed until no abnormal data is identified.

[0079] That is, when abnormal data is identified from the wind speed-power data, the data set is updated by removing the abnormal data so that the data set only includes the currently determined normal data. Thereafter, the updated data set can be applied to steps S130, S140, and S150. When no abnormal data is identified in step S150, the process can end.

[0080] That is to say, considering the interference caused by abnormal data in the fitting process, the fitting curve may be biased. Therefore, in the present disclosure, after the outliers are eliminated, the above steps S130, S140 and S150 can be iterated continuously until no new outliers are judged to be finished. Critical outliers can be better eliminated through multiple iterations.

[0081] Figure 2 A flow chart showing a construction data set of a wind farm according to an exemplary embodiment of the present disclosure.

[0082] Figure 2 The specific numerical values ​​shown in are merely exemplary, and the various values ​​of the parameters used in the present disclosure are not limited to the specifically shown values.

[0083] Reference Figure 2, the constructed data set should include three sections. The first is the low wind speed section where the wind turbine cannot be cut in, and the power is directly set to 0; the second is the full-power and high-wind stage, and its power is set to the maximum power in the data set; the third is the intermediate wind speed section, and the power takes the latest 30 data at every 0.25m / s interval. If the data volume cannot be met at each interval, it will be continuously upsampled until it meets the requirements; the three sections of data are combined to form a complete data set. In an example, the low wind speed where the wind turbine cannot be cut in can be set to 0-2m / s; according to the maximum wind speed corresponding to the maximum power (if the wind speed is greater than 12, it is set to 12), the full-power and high-wind stage is set to the maximum wind speed ~ maximum wind speed + 4m / s.

[0084] Figure 3 A flowchart of constructing a trusted data range according to an exemplary embodiment of the present disclosure is shown.

[0085] Figure 3 The specific numerical values ​​shown in are merely exemplary, and the various values ​​of the parameters used in the present disclosure are not limited to the specifically shown values.

[0086] Reference Figure 3 , the fitting curve can be used to construct a credible data range and exclude abnormal data. Each wind speed corresponds to an upper and lower bound of the credible data range. The wind speed between the limits of the credible data range is judged as normal, otherwise it is abnormal data.

[0087] Further references Figure 3 If the wind speed is within 3m / s, the wind turbine cannot be cut in normally, and the power is generally 0. At this time, the lower limit of the power is set to 0. At the same time, considering that the wind speed of a small number of wind turbine heads is higher than the wind speed of the entire field, it may be slightly higher than 0 at this time. The upper limit is set to the fitting value of the curve + installed capacity × 0.05.

[0088] If the wind speed is between 3m / s and 12m / s, that is, it is in the wind speed range between wind turbine cut-in and full power; the average wind speed is getting higher and higher, but not all wind turbine head wind speeds can reach higher wind speeds, so the higher the wind speed, the wider the corresponding power credible range should be set; here the upper limit is set to the fitting curve plus a certain growth space, and the lower limit is set to the fitting curve minus a certain growth space.

[0089] If the wind speed is between 12m / s and 22m / s, the wind turbine is normally at full power and the power is generally equal to the installed capacity. In this case, the upper limit of the power is set to the installed capacity. At the same time, considering that the wind speed at the head of a small number of wind turbines is lower than the wind speed of the entire field, it may be slightly lower than the installed capacity at this time. Therefore, the lower limit is set to the installed capacity × 0.8.

[0090] If the wind speed is above 22m / s, the wind speed is too high and normally the wind turbine will be switched off. In this case, the power will no longer be filtered and the upper and lower limits will be set to 0 and the installed capacity.

[0091] Figure 4 A flow chart showing a method for processing abnormal data of a wind farm according to an exemplary embodiment of the present disclosure is shown.

[0092] As mentioned above, normal data can be gradually obtained by excluding abnormal data. However, the excluded abnormal data may not only include real abnormal data due to wind turbine abnormalities, data collection process abnormalities, etc., but may also include data before some installation changes. If these data are completely eliminated, too much data will be lost, which is not conducive to modeling. Therefore, in the present disclosure, in order to avoid or reduce the elimination of useful data, the abnormal data processing method of the wind farm according to the exemplary embodiment of the present disclosure may also include data correction to achieve data enhancement, which is conducive to improving the accuracy of wind power modeling. This is very effective when facing the following problems: some electric fields have too much abnormal data due to insufficient historical data preservation. If only normal data can be used for wind power modeling, the model may have too little modeling data and the model stability may be poor. Correcting the abnormal data to normal data and using it together with other data for modeling can improve the stability of the model and improve the prediction accuracy.

[0093] More specifically, refer to Figure 4 , in step S410, the wind speed in the abnormal data is obtained.

[0094] In one example, the wind speed in all abnormal data may be obtained. However, the present disclosure is not limited thereto, and the wind speed in part of the abnormal data may also be obtained.

[0095] In step S420, new power is obtained by substituting the wind speed in the abnormal data into the fitting curve.

[0096] Here, the fitting curve can be a reference Figure 1A The final fitting curve was obtained by the abnormal data processing method of Figure 1A , the fitting curve may correspond to the fitting function whose parameters have been determined. Therefore, obtaining the new power by substituting the wind speed in the abnormal data into the fitting curve may be understood as obtaining the new power by substituting the wind speed in the abnormal data into the fitting function whose parameters have been determined.

[0097] In step S430, the wind speed in the abnormal data and the obtained new power are taken as normal data.

[0098] That is, the wind speed in the abnormal data and the obtained new power can be used as the corrected values.

[0099] In addition, optionally, abnormal data correction can be made up not only through curve fitting processing, but also by building machine learning models, mathematical formulas, etc.

[0100] Therefore, the data before abnormal data and installation changes cannot be used in modeling. If all data are deleted, a large amount of data will be lost. According to the abnormal data processing method of the exemplary embodiment of the present disclosure, the abnormal data is effectively corrected to normal data, which is conducive to modeling and improves the prediction accuracy. The present disclosure proposes that the wind speed of the abnormal data point can be corrected to the normal value by fitting the generalized logic function.

[0101] Figure 5 A scatter plot showing measured wind speed and measured power of a wind farm according to an exemplary embodiment of the present disclosure is shown.

[0102] As an example only, Table 1 below shows a table of measured data.

[0103] time Measured wind speed Measured power 2022 / 7 / 1 0:00 13.08 4673.67 2022 / 7 / 1 0:15 12.4 4685.17 2022 / 7 / 1 0:30 12.1 4687.67 2022 / 7 / 1 0:45 12.32 4668.5 2022 / 7 / 1 1:00 11.35 4673.5 2022 / 7 / 1 1:15 11.83 4676 2022 / 7 / 1 1:30 11.4 4686.67

[0104] Table 1 Measured data table

[0105] The scatter plot of measured wind speed and measured power is as follows: Figure 5 As shown. Figure 5 In the example, the horizontal axis may represent the measured wind speed, and the vertical axis may represent the measured power. The electric field has undergone multiple installation changes during this period, and the historical data rules cannot be applied to the latest modeling. The abnormal data processing method according to the exemplary embodiment of the present disclosure can identify and correct abnormal data and data before installation change.

[0106] Figure 6 A data set used for fitting a curve of a wind farm according to an exemplary embodiment of the present disclosure is shown.

[0107] Reference Figure 6 , shows a data set for constructing a fitting curve based on the data of the latest period of time, where the gray points are all data points and the black points are the constructed data set for fitting the curve.

[0108] According to the abnormal data processing method of the exemplary embodiment of the present disclosure, by adopting the latest data in segments and combining the data construction of low wind and high wind, the latest wind speed-power curve after the installation change can be fitted.

[0109] Figure 7 A fitting curve corresponding to the fitted generalized logic function according to an exemplary embodiment of the present disclosure is shown.

[0110] Reference Figure 7 , the black dashed line shows the Figure 6 The fitted curve corresponding to the generalized logistic function is fitted to the data set.

[0111] Figure 8 A schematic diagram illustrating abnormal data processing according to an exemplary embodiment of the present disclosure.

[0112] Reference Figure 8 , according to the fitted generalized logic function, construct the credible data range. According to the corresponding upper and lower bounds in the credible data range, judge whether the power corresponding to each wind speed is abnormal data, and the data outside the upper and lower bounds are abnormal data. Figure 8 In the figure, the normal data is black and the abnormal data is gray. It can be clearly seen that the data before the installation change is effectively identified.

[0113] That is to say, according to the abnormal data processing method disclosed in the present invention, even if the electric field has an installed capacity change, there is no need to manually adjust the abnormal data processing method. According to the abnormal data processing method disclosed in the present invention, the normal credible range is automatically adjusted according to the latest data to distinguish before and after the installed capacity change. The same data will be judged as different types of data in different time periods due to the installed capacity change.

[0114] As wind farms continue to increase in wind power installed capacity and the industry's technological innovation capabilities continue to improve, wind turbines continue to develop towards large-scale, power farms continue to expand and gradually connect to the grid, and power farm installed capacity continues to change in a short period of time, and the wind speed-power curve continues to change. This leads to the need for constant adjustment of abnormal data identification methods to distinguish before and after the installation change, and also leads to the fact that the measured power data before the installation change cannot be used for actual modeling. Therefore, adaptive installation changes and processing of these abnormal data can better detect abnormal data.

[0115] Fig. 9 A schematic diagram showing abnormal data correction according to an exemplary embodiment of the present disclosure.

[0116] Reference Fig. 9 , the wind speed at the abnormal point can be substituted into the fitting curve to calculate the power value as the corrected value. Fig. 9 In the figure, the corrected data is black and the normal data is gray.

[0117] Fig.10 1 is a data overlay diagram of a wind farm from January to February (black) and from July to August (gray) according to an exemplary embodiment of the present disclosure. Fig.11 : This is the filtering effect on the data of the wind farm from January to February according to an exemplary embodiment of the present disclosure (black represents normal data, and gray represents abnormal data). Fig.12 This is the filtering effect on the wind farm data from July to August according to the exemplary embodiment of the present disclosure (black is normal data, and gray is abnormal data).

[0118] Reference Fig.10 , we can see that there is a clear difference between the wind farm data from January to February (black) and the data from July to August (gray). Fig.11 and Fig.12According to the abnormal data processing method of the exemplary embodiment of the present disclosure, the wind farm can automatically adapt to the power generation law of the corresponding season in different seasons and effectively filter abnormal data.

[0119] The abnormal data processing method according to the exemplary embodiment of the present disclosure is adaptive to seasonal changes. The wind speed-power model will change regularly due to seasonal changes; the abnormal data processing method according to the exemplary embodiment of the present disclosure can automatically fit the latest data curve based on the latest data, thereby obtaining data that is more suitable for the current season and automatically adjusting the abnormal data processing method according to the season.

[0120] That is to say, when the seasons change, the wind speed-power model will also change regularly. According to the abnormal data processing method of the exemplary embodiment of the present disclosure, the measured data that conforms to the current rules can be screened according to the latest seasonal data characteristics, and the constructed model can be more in line with the latest situation, and the prediction accuracy can be higher.

[0121] Fig.13 A flowchart of a method for processing abnormal data according to an exemplary embodiment of the present disclosure is shown.

[0122] Reference Fig.13 According to the abnormal data processing method of the example embodiment of the present disclosure, the measured wind speed-measured power data and the electric field basic information table can be first combined to construct a data set of fitting curves, fit a generalized logic function to the data set, and construct the credible upper and lower bounds of the data through the fitted curve to identify the abnormal data; if a new abnormal value is identified this time, it is necessary to return to the above-mentioned data set construction stage and iterate repeatedly until no new abnormal value can be identified, then correct the abnormal value and output all the normal data after processing.

[0123] Combined with the measured wind speed-measured power data and the basic electric field information table, a data set of the fitting curve is constructed, and a generalized logic function is fitted to the data set. The credible upper and lower bounds of the data are constructed through the fitted curve to identify abnormal data. If new outliers are identified this time, it is necessary to return to the above-mentioned data set construction stage and iterate repeatedly until no new outliers can be identified. Then, the outliers are corrected and all normal data after processing are output.

[0124] Fig.14 A block diagram showing an abnormal data processing apparatus according to an exemplary embodiment of the present disclosure is shown.

[0125] Reference Fig.14 According to an exemplary embodiment of the present disclosure, an abnormal data processing device 1400 may include a processor 1410 and a memory 1420. Here, the memory 1420 stores a computer program, wherein the computer program implements reference when executed by the processor 1410. Figures 1A to 13 For simplicity, the reference data processing method executed by the processor 1410 is not described again here. Figures 1A to 13 Any abnormal data processing method described in at least one of the above.

[0126] In addition, the method according to the exemplary embodiment of the present disclosure can be implemented as a computer program in a computer-readable recording medium. Those skilled in the art can implement the computer program according to the description of the above method. When the computer program is executed in a computer, the above method of the present disclosure is implemented.

[0127] In addition, it should be understood that the various units in the device according to the exemplary embodiments of the present disclosure can be implemented as hardware components and / or software components. Those skilled in the art can, for example, use a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC) to implement the various units according to the processing performed by the defined various units.

[0128] Although the present disclosure includes specific examples, it will be clear to those of ordinary skill in the art that various changes in form and detail may be made in these examples without departing from the spirit and scope of the claims and their equivalents. The examples described herein should be considered to be descriptive only and not for limiting purposes. The description of features or aspects in each example should be considered to be applicable to similar features or aspects in other examples. If the described techniques are performed in a different order, and / or if the components in the described systems, architectures, devices, or circuits are combined in different ways, and / or replaced or supplemented by other components or their equivalents, suitable results can be achieved. Therefore, the scope of the disclosure is not limited by specific embodiments, but by the claims and their equivalents, and all changes within the scope of the claims and their equivalents should be interpreted as included in the disclosure.

Claims

1. A method for processing abnormal data of a wind farm, characterized in that: The abnormal data processing method comprises: Acquire wind speed-power data of the wind farm, wherein the wind speed-power data includes a wind speed and a power corresponding to the wind speed; constructing a data set for fitting a curve based on the acquired wind speed-power data; Based on the data set, fitting a preset fitting function to determine the parameters of the fitting function, and generating the fitting curve corresponding to the fitting function with the parameters; Identify the wind speed-power data within the credible data range of the fitting curve as normal data, and identify the wind speed-power data outside the credible data range of the fitting curve as abnormal data; In response to identifying the abnormal data, the data set is updated by removing the abnormal data, and the step of generating the fitting curve is iteratively performed until no abnormal data is identified.

2. The abnormal data processing method according to claim 1, characterized in that: The data set includes a first data set corresponding to a low wind speed segment, a second data set corresponding to an intermediate wind speed segment, and a third data set corresponding to a full wind speed segment. The low wind speed section corresponds to the wind speed section that the wind turbine cannot cut in, and the power in the first data set is set to 0. The intermediate wind speed section corresponds to the wind speed section where the wind turbine can be cut in and has not yet reached full power; The full-power wind speed section corresponds to the wind speed section at which the wind turbine is fully powered, and the power in the third data set is set to the maximum power.

3. The abnormal data processing method according to claim 2, characterized in that: In the middle wind speed section, a predetermined amount of wind speed-power data is collected at a predetermined wind speed interval as a second data set. If the number of wind speed-power data collected at predetermined wind speed intervals in the intermediate wind speed segment is less than the predetermined number, the predetermined number of wind speed-power data is obtained by upsampling the collected wind speed-power data.

4. The abnormal data processing method according to claim 2, characterized in that: The size of the credible data range of the fitting curve corresponding to the first data set, the size of the credible data range of the fitting curve corresponding to the second data set, and the size of the credible data range of the fitting curve corresponding to the third data set increase in sequence.

5. The abnormal data processing method according to claim 1, characterized in that: The credible data range of the fitting curve is determined based on the installed capacity of the wind farm.

6. The abnormal data processing method according to claim 5, characterized in that: When the wind speed is less than the first wind speed, the lower limit of the credible data range of the fitting curve is 0, and the upper limit of the credible data range of the fitting curve is the sum of the first value and the fitting value of the fitting curve, wherein the first value corresponds to the product of the installed capacity and the second value; Wherein, when the wind speed is greater than or equal to the first wind speed and less than the second wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the third value and the fourth value, and the upper limit of the credible data range of the fitting curve is the smaller value between the installed capacity and the fifth value, wherein the third value corresponds to the product of the installed capacity and the sixth value, the fourth value corresponds to the difference between the fitting value of the fitting curve and the product of the installed capacity and the seventh value, and the fifth value corresponds to the sum of the fitting value of the fitting curve and the product of the installed capacity and the eighth value; When the wind speed is greater than or equal to the second wind speed and less than the third wind speed, the lower limit of the credible data range of the fitting curve is the smaller value between the ninth value and the tenth value, and the upper limit of the credible data range of the fitting curve is the installed capacity, wherein the ninth value corresponds to the product of the installed capacity and the eleventh value, and the tenth value corresponds to the difference between the fitting value of the fitting curve and the product of the installed capacity and the twelfth value.

7. The abnormal data processing method according to claim 1, characterized in that: The abnormal data processing method further includes: Obtaining wind speed in the abnormal data; Obtaining new power by substituting the wind speed in the abnormal data into the fitting curve; The wind speed in the abnormal data and the obtained new power are taken as normal data.

8. The abnormal data processing method according to claim 1, characterized in that: The wind speed-power data includes wind speed-power data during a preset time period before and after the wind turbine installation is changed or wind speed-power data in the current season.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the abnormal data processing method according to any one of claims 1 to 8 is implemented.

10. An abnormal data processing device for a wind farm, characterized in that: The abnormal data processing device comprises: processor; Memory, The memory stores a computer program, and when the computer program is executed by the processor, the abnormal data processing method according to any one of claims 1 to 8 is implemented.

Images