Wind turbine generator yaw static error calculation method and system based on data analysis

Abstract

The invention relates to a wind turbine generator yaw static error calculation method and system based on data analysis, and the method comprises the steps: obtaining historical data of parameters, such as the wind speed, the wind direction and the active power, of a target generator, carrying out the effective screening of the data, removing the abnormal data of the active power, and finally obtaining an effective data set; taking the weighted average value of the yaw errors of all the wind speed intervals as a first yaw error value of the target unit; the yaw error average value in the wind direction interval serves as a second yaw error value of the target unit according to the maximum generating capacity principle; and if the calculation condition verification is passed, taking the average value of the first yaw error value and the second yaw error value as the final yaw static error value Y of the target unit. According to the method, calculation is carried out from the two dimensions of wind speed and wind direction based on operation data analysis, the static yaw error of the wind turbine can be accurately recognized, the error value serves as compensation input of a yaw control system, and the power generation performance of the wind turbine generator system can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power generation, and in particular relates to a method and system for calculating the yaw static error of a wind turbine generator set based on data analysis. Background Art

[0002] As the global fossil energy shortage and environmental pollution become increasingly prominent, wind energy, as a clean and renewable energy source, is becoming more and more popular. As a key area and main layout point for China's future development, the wind power industry has entered the fast lane, and the future development environment will continue to improve.

[0003] The yaw system is an important part of the wind turbine control system. Its main function is to quickly and smoothly align the wind direction when the wind direction changes so that the wind rotor can capture the maximum wind energy. However, due to meteorological, mechanical installation, sensor errors and other reasons, there is an angle deviation between the nacelle position and the actual wind direction, which is the yaw static error. The angle relationship is detailed in the attached Figure 1 This error will lead to a decrease in wind energy utilization and power generation, while increasing the load on the unit and reducing the structural life of the unit. How to accurately identify the yaw static error of a wind turbine is crucial to improving the power generation efficiency and operating stability of the wind turbine.

[0004] Currently, there are two common methods for calculating yaw static error: one is the actual measurement method, which uses laser radar anemometer data to compare and analyze the wind speed and direction instrument data of the wind turbine, and calculates the yaw error with the radar wind measurement data as a reference. This method has high calculation accuracy, but has problems such as long test cycle, high cost, and great influence of light; the other is the data analysis method, which analyzes a large amount of data such as wind speed, wind direction, power, etc. of the wind turbine SCADA system, and calculates the yaw static error with "maximum power generation" as the core. This method is easy to implement, low cost, and highly scalable, but the data quality and algorithm have a great influence on its calculation accuracy. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for calculating the static yaw error of a wind turbine based on data analysis. The method obtains data such as wind speed, wind direction and power through the SCADA system of the wind turbine, screens the data within an effective range and eliminates abnormalities, and comprehensively calculates the static yaw error from two dimensions of wind speed and wind direction to improve the calculation accuracy of the static yaw error and solve the problem of low calculation accuracy of the data analysis method.

[0006] The present invention provides a method for calculating the static yaw error of a wind turbine generator set based on data analysis, comprising the following steps:

[0007] Step 1: Obtain historical operation data of the wind turbine through the wind turbine SCADA system to obtain an original data sample O; the historical operation data includes wind speed, wind direction, and active power of the target unit;

[0008] Step 2: Perform data cleaning on the original data sample O obtained in step 1 to obtain a cleaned valid data set N;

[0009] Step 3: Analyze the data set N obtained in step 2 from the wind speed dimension, divide the data of the target model within the rated wind speed range into wind speed bin intervals according to the wind speed step of 0.5 m / s, and obtain the data set {Ni}, where i is the i-th wind speed bin interval;

[0010] Step 4: For the data set {Ni} obtained in step 3, divide each wind speed interval into wind direction bin intervals according to the wind direction step of 1° to obtain the data set {Ni_j}, where j is the jth wind direction bin interval;

[0011] Step 5: Calculate the data set {Pi_j} of the average active power in each wind direction interval in the data set {Ni_j} in step 4;

[0012] Step 6: For the data set {Pi_j} obtained in step 5, according to the maximum power principle, calculate the yaw error data set {D a i_m}, and the yaw error weighting factor dataset {D n i_m}, m is the wind direction interval where the average active power is the largest;

[0013] Step 7: The yaw error value dataset {D a i_m} and yaw error weighting factor dataset {D n i_m}, calculate the weighted average of the yaw errors in all wind speed intervals as the first yaw error value Y_fs of the target unit;

[0014] Step 8: Analyze the data set N obtained in step 2 from the wind direction dimension, divide all data into wind direction bin intervals according to the wind direction step of 1°, and obtain the data set {Nj}, where j is the jth wind direction bin interval;

[0015] Step 9: For the data set {Nj} obtained in step 8, divide each wind direction interval into wind speed bin intervals according to the wind speed step size of 0.5 m / s to obtain the data set {Nj_k}, where k is the kth wind speed bin interval;

[0016] Step 10: For the data set {Nj_k} obtained in step 9, calculate the average wind speed and active power in each wind speed bin interval to obtain the data set {VPj_k} corresponding to wind speed and power, where j is the jth wind direction bin interval and k is the kth wind speed bin interval;

[0017] Step 11: According to the data set {VPj_k} obtained in Step 10, perform curve fitting on the wind speed and power data for each wind direction interval to obtain the power curve set {PC_j} corresponding to all wind direction intervals;

[0018] Step 12: According to the power curve set {PC_j} obtained in Step 11, calculate the integral value of each power curve in the wind speed interval [V 切入 , V 额定 to obtain the data set {P integral _j} of the integral values of the power curves for all wind direction intervals;

[0019] Step 13: According to the principle of maximum power generation, take the average value of the wind directions in the wind direction interval where the maximum integral value is located in the data set {P integral _j} obtained in Step 12 as the second yaw error value Y_fx of the target unit;

[0020] Step 14: Perform calculation condition verification on the first yaw error value and the second yaw error value obtained in Step 7 and Step 13. If the verification fails, increase the sample size and continue to execute Steps 1 to 14. If the verification passes, directly execute the subsequent steps;

[0021] Step 15: After the verification conditions in Step 14 are met, take the average value of the first yaw error value Y_fs obtained in Step 7 and the second yaw error value Y_fx obtained in Step 13 as the yaw static error value Y of the target unit.

[0022] Furthermore, the original data variables in Step 1 include the fan status word S, wind direction D, wind speed V, active power P, pitch angle θ, yaw operation mode M, data accuracy of 1 minute level, and the data volume of at least 1 year or more. Denote the original data sample as O = {S, D, V, P, θ, M};

[0023] The data cleaning in Step 2 includes screening the data valid range and removing abnormal data. The data valid range screening conditions are as follows:

[0024] 1) Status word S = normal power generation operation;

[0025] 2) -180° ≤ D ≤ 180°;

[0026] 3) V 切入 - 0.5 ≤ V ≤ V 切出 ;

[0027] 4) 0 < P ≤ 1.1 * rated power of the unit;

[0028] 5) 0° ≤ θ < 89°;

[0029] 6) Yaw operation mode M = non-yaw operation state;

[0030] After effectiveness screening, we get the data set X = {D, V, P, θ};

[0031] Abnormal data removal uses the standardized distance in the K nearest neighbor algorithm to remove abnormal active power data in the data set X, including:

[0032] Divide the data set X into wind speed bin intervals according to the wind speed step size of 0.5m / s to obtain the data set {X_a}, where a is the number of wind speed bin intervals;

[0033] Calculate the average data set {P_a} and standard deviation data set {S_a} of active power in each wind speed interval;

[0034] According to the standard distance formula: Calculate the standardized distance data set {d_a} from all active power data P to the active power average value P_a in each wind speed interval;

[0035] According to the elimination condition d_a>b*P_a, b is the threshold, and the data points whose normalized distance of active power in each wind speed interval is greater than b times the average active power are eliminated; after eliminating abnormal data, the data set N={D,V,P,θ} is obtained.

[0036] Furthermore, the specific division method of the wind direction bin interval in each wind speed interval in step 4 is:

[0037] Determine the absolute wind direction angle Ф for yaw start according to the target aircraft model, and calculate Then the number of wind direction bin intervals in this wind direction segment is e * =2*e;

[0038] Taking -e as the starting point and the step length of 1°, the wind direction bin interval is divided into [-e, -e+1), [-e+1, -e+2), [-e+2, -e+3), ..., [-e+j-1, -e+j], where j is the jth wind direction bin interval;

[0039] Get the data set {Ni_j} divided by wind direction bin interval.

[0040] Furthermore, in step 6, according to the maximum power principle, the data set {Pi_j} obtained in step 5 is processed as follows:

[0041] Filter out the data set of the wind direction interval with the maximum average active power in each wind speed interval {P max i_m}, where i is the i-th wind speed interval, and m is the wind direction interval where the average active power is the largest;

[0042] Calculate the average wind direction in the m wind direction interval as the yaw error value in the wind speed interval, and obtain the data set {D a i_m};

[0043] Calculate the number of wind directions in the m wind direction interval as the weighting factor of the yaw error in the wind speed interval, and obtain the yaw error weighting factor data set {D n i_m}.

[0044] Furthermore, the calculation method of the first yaw error value Y_fs of the target unit in step 7 is: Y_fs=D a i_m*D n i_m, where i is the i-th wind speed interval;

[0045] The specific division method of the N wind direction bin intervals of the data set in step 8 is:

[0046] Determine the absolute wind direction angle Ф for yaw start according to the target aircraft model, and calculate Then the number of wind direction bin intervals in this wind direction segment is e * =2*e;

[0047] Taking -e as the starting point and the step length of 1°, the wind direction bin interval is divided into [-e, -e+1), [-e+1, -e+2), [-e+2, -e+3), ..., [-e+j-1, -e+j], where j is the jth wind direction bin interval;

[0048] Get the data set {Nj} divided by wind direction bin interval;

[0049] The specific method for dividing the wind speed bin intervals in each wind direction interval in step 9 is:

[0050] Determine the cut-in wind speed and cut-out wind speed according to the target model, and calculate The number of wind speed bins in this wind speed segment is

[0051] V 切入 -0.25 is the starting point, and the step length is 0.5m / s to divide the wind speed bin interval [V 切入 -0.25,V 切入 +0.25), [V 切入 +0.25,V 切入 +3*0.25),...,[V 切入 +(2(k-1)-1)*0.25,V 切入 +(2(k-1)+1)*0.25), where k is the k-th wind speed bin interval;

[0052] Get the data set {Nj_k} divided by wind speed bin interval;

[0053] In step 10, the data set {Nj_k} obtained in step 9 is processed as follows:

[0054] Calculate the average wind speed data set in each wind speed bin interval in all wind direction intervals {V a j_k};

[0055] Calculate the average active power data set in each wind speed bin interval in all wind direction intervals {P a j_k};

[0056] The dataset {V a j_k} and {P a j_k} are merged into the data set corresponding to wind speed and power {VPj_k}, where j is the jth wind direction bin interval and k is the kth wind speed bin interval.

[0057] Furthermore, in step 11, curve fitting is performed on the wind speed and power data of each wind direction interval. Specifically, the least squares method is used for polynomial fitting. The polynomial order and fitting coefficient are selected according to the RMSE, SSE, and R-square parameters to obtain the power curve set {PC_j} corresponding to all wind direction intervals.

[0058] Furthermore, the logic of verifying the calculation conditions of the first yaw error value Y_fs and the second yaw error value Y_fx in step 14 is that one of the following conditions is met:

[0059] 1) The numerical signs of Y_fs and Y_fx are the same, and |Y_fs-Y_fx|≤g, where g is the threshold of the absolute value of the error;

[0060] 2) If the calculation condition 1) fails to pass the verification, the number of samples is increased by the amount of data for each 2-month period and recalculated. When the number of sample increases = h, h is the threshold for the number of sample increases.

[0061] The present invention also provides a wind turbine yaw static error calculation system based on data analysis, comprising a calculation module, wherein the calculation module executes the wind turbine yaw static error calculation method based on data analysis.

[0062] The present invention also provides a non-transitory computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, the method for calculating the static yaw error of a wind turbine set based on data analysis is implemented.

[0063] The present invention also provides an electronic device, comprising:

[0064] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for calculating the static yaw error of a wind turbine based on data analysis by executing the computer instructions.

[0065] Through the above scheme, the wind turbine yaw static error calculation method and system based on data analysis have the following technical effects:

[0066] 1) Based on the standardized distance screening method in the K-Nearest Neighbors algorithm, the abnormal active power data is eliminated, which improves the quality of sample data and provides effective data support for the accurate calculation of the unit's yaw static error.

[0067] 2) Based on the analysis of operating data, the maximum power calculation method in the wind speed range and the maximum power generation calculation method in the wind direction range are used to comprehensively calculate the yaw static error from the two dimensions of wind speed and wind direction, respectively. This method has the characteristics of easy implementation, low cost and high accuracy.

[0068] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic diagram of the relationship between the static yaw error angles of a wind turbine generator set;

[0070] Figure 2 It is a flow chart of a method for calculating static yaw error of a wind turbine generator set based on data analysis of the present invention;

[0071] Figure 3 It is a wind speed-power scatter point comparison diagram before and after the data cleaning obtained in step 2 of the implementation case of the present invention;

[0072] Figure 4 It is a comparison diagram of the power curve before and after data cleaning and the static power curve obtained in step 2 of the implementation case of the present invention;

[0073] Figure 5 is a wind direction probability distribution histogram after data cleaning obtained in step 2 of the implementation case of the present invention;

[0074] Figure 6 is a polynomial fitting process diagram of the wind speed-power curve obtained in step 11 in the implementation case of the present invention;

[0075] Figure 7is a fitting diagram of 30 wind speed-power curves divided by wind direction intervals obtained in step 11 of the implementation case of the present invention;

[0076] Figure 8 It is a schematic diagram of the structure of the electronic equipment of the present invention. DETAILED DESCRIPTION

[0077] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0078] Ginseng Figures 2 to 7 As shown, this embodiment provides a method for calculating the static yaw error of a wind turbine generator set based on operation data analysis. The method uses the wind turbine generator set data acquisition and monitoring control (SCADA) system as the data source. First, historical data of parameters such as wind speed, wind direction, and active power of the target unit are obtained, and the data is screened for validity according to the parameter range of the target model. Then, the abnormal active power data is eliminated by using the standardized distance in the K-nearest neighbors algorithm according to the wind speed step of 0.5 m / s, and finally a valid data set N is obtained; secondly, the data set N is first divided into wind speed bin intervals with a step of 0.5 m / s within the rated wind speed, and each wind speed interval is then divided into bin intervals according to the wind direction, and the average value of the active power in each wind direction interval is calculated. The yaw error value and the corresponding wind direction number of each wind speed interval are calculated according to the maximum active power (maximum active power average value) principle, and finally the weighted average value of the yaw errors of all wind speed intervals is taken as the first yaw error value Y_fs of the target unit. ; Then first divide the data set N into wind direction bin intervals with a step size of 1° within the effective wind direction, and then divide each wind direction interval into bin intervals according to the wind speed, obtain the actual power curve of each wind direction interval, and calculate the integral value of the actual power curve within the rated wind speed interval of each wind direction interval. According to the principle of maximum power generation (maximum integral value of power curve), the average yaw error value within the wind direction interval is used as the second yaw error value Y_fx of the target unit; finally, if the calculation conditions are verified, the average value of the first yaw error value and the second yaw error value is used as the final yaw static error value Y of the target unit.

[0079] The present invention performs calculations from two dimensions of wind speed and wind direction based on operating data analysis, can accurately identify the static yaw error of the wind turbine, and uses the error value as the compensation input of the yaw control system, which can effectively improve the power generation performance of the wind turbine generator set and has strong application value.

[0080] The implementation object is a 3MW doubly-fed wind turbine generator set in an onshore wind farm. The wind speed and direction acquisition device uses a mechanical anemometer. The data source used is the wind turbine SCADA system. The algorithm implementation part is implemented by MATLAB software. The detailed implementation steps are as follows:

[0081] Step 1: Obtain the original data variables required by the present invention through the wind turbine SCADA system, including wind turbine status word, wind direction, wind speed, active power, pitch angle, yaw operation mode, and the data variable information is shown in the table below.

[0082]

[0083] The data accuracy is 1 minute, and the time period is selected from 2022.06.28 to 2023.06.27, with a total of about 518,000 data. The specific sample data is shown in the table below.

[0084]

[0085] The original data samples obtained above are denoted as O = {S, D, V, P, θ, M}.

[0086] Step 2: Perform data cleaning on the original data sample O obtained in step 1, including data valid range screening and abnormal data removal.

[0087] Data valid range screening:

[0088] 1) S = 1, normal power generation operation;

[0089] 2) -180°≤D≤180°;

[0090] 3) V 切入 -0.5≤V≤V 切出 , in this case V 切入 =3m / s, V 切出 =23m / s;

[0091] 4)0 <P≤1.1*P 额定 , in this case P 额定 =3000kW;

[0092] 5) 0°≤θ<89°;

[0093] 6) M = 0, non-yaw operation state.

[0094] After effectiveness screening, variables related to yaw error calculation are selected to obtain the data set X = {D, V, P, θ}.

[0095] Abnormal data removal:

[0096] In order to solve the problem of abnormal fluctuation of active power data, the standardized distance in the K-Nearest Neighbors algorithm is used to remove abnormal active power data in data set X:

[0097] First, divide the data set X into wind speed bin intervals according to the wind speed step size of 0.5m / s to obtain the data set {X_a}, where a is the number of wind speed bin intervals; secondly, calculate the average data set {P_a} and standard deviation data set {S_a} of active power in each wind speed interval;

[0098] Then according to the standard distance formula: Calculate the standardized distance data set {d_a} from all active power data P to the active power average value P_a in each wind speed interval;

[0099] Finally, according to the elimination condition d_a>b*P_a, b is the threshold, and referring to the "3σ principle", b=3 in this example, the data points whose normalized distance of active power in each wind speed interval is greater than 3 times the average value of active power are eliminated; after eliminating abnormal data, the data set N={D,V,P,θ} is obtained;

[0100] As attached Figure 3 The following is a comparison of wind speed and power scatter points before and after data cleaning. Figure 4 The following is a comparison of the power curves before and after data cleaning and the static power curve. Figure 5 Shown is the wind direction probability distribution histogram after data cleaning.

[0101] Step 3: Analyze the data set N obtained in step 2 from the wind speed dimension, and divide the data of the instance unit within the rated wind speed range into wind speed bin intervals with a wind speed step of 0.5 m / s. The specific division method is as follows:

[0102] First, determine V according to the example unit 切入 =3m / s and V 额定 =11m / s, calculate Then the number of wind speed bins in this wind speed segment is c * =16;

[0103] Then V 切入 -0.25 is the starting point, and the step length is 0.5m / s to divide the wind speed bin interval into [2.75,3.25), [3.25,3.75), ..., [V 切入 +(2(i-1)-1)*0.25,V 切入 +(2(i-1)+1)*0.25),...,[10.25,10.75], where i is the i-th wind speed bin interval;

[0104] Finally, the data set {Ni} divided by wind speed bin interval is obtained, where i is the i-th wind speed bin interval.

[0105] Step 4: For the data set {Ni} obtained in step 3, divide each wind speed interval into wind direction bin intervals according to the wind direction step of 1°. The specific division method is as follows:

[0106] First, according to the example unit, the absolute wind direction angle for yaw start is determined to be Ф = 15°, then e = 15, and the number of wind direction bin intervals in this wind direction segment is e * =30;

[0107] Then, starting from -e, the wind direction bin interval is divided into [-15,-14), [-14,-13), [-1,0), ..., [-15+j-1,-15+j), ..., [14,15] with a step size of 1°, where j is the jth wind direction bin interval;

[0108] Finally, we get the data set {Ni_j} divided by wind direction bin intervals.

[0109] Step 5: Calculate the data set {Pi_j} of the average active power in each wind direction interval in the data set {Ni_j} in step 4.

[0110] Step 6: According to the maximum power principle, the data set {Pi_j} obtained in step 5 is processed as follows:

[0111] First, filter out the data set {P max i_m}, where i is the i-th wind speed interval, and m is the wind direction interval where the average active power is the largest;

[0112] Then the average wind direction in the m wind direction interval is calculated as the yaw error value in the wind speed interval, and the data set {D a i_m};

[0113] Finally, the number of wind directions in the m wind direction interval is calculated as the weighting factor of the yaw error in the wind speed interval, and the yaw error weighting factor data set {D n i_m}.

[0114] Step 7: The yaw error value dataset {D a i_m} and yaw error weighting factor dataset {D n i_m}, calculate the weighted average of the yaw errors in all wind speed intervals as the first yaw error value Y_fs of the target unit;

[0115] The specific calculation formula of the first yaw error value Y_fs is: Y_fs = D ai_m*D n i_m, where i is the i-th wind speed interval;

[0116] Through calculation, the first yaw error value of the unit in this example is Y_fs=-2.8235.

[0117] Step 8: Analyze the data set N obtained in step 2 from the wind direction dimension, and divide all data into wind direction bin intervals according to the wind direction step of 1°. The specific division method is as follows:

[0118] First, according to the example unit, the absolute wind direction angle for yaw start is determined to be Ф = 15°, then e = 15, and the number of wind direction bin intervals in this wind direction segment is e * =30;

[0119] Then, starting from -e, the wind direction bin interval is divided into [-15,-14), [-14,-13), [-1,0), ..., [-15+j-1,-15+j), ..., [14,15] with a step size of 1°, where j is the jth wind direction bin interval;

[0120] Finally, we get the data set {Nj} divided by wind direction bin intervals.

[0121] Step 9: For the data set {Nj} obtained in step 8, divide the data in each wind direction interval into wind speed bin intervals with a wind speed step of 0.5 m / s. The specific division method is as follows:

[0122] First, determine V according to the example unit 切入 =3m / s and V 切出 =23m / s, calculate f=40, then the number of wind speed bins in this wind speed segment is f * =40;

[0123] Then V 切入 -0.25 is the starting point, and the step length is 0.5m / s to divide the wind speed bin interval into [2.75,3.25), [3.25,3.75), ..., [V 切入 +(2(i-1)-1)*0.25,V 切入 +(2(i-1)+1)*0.25), ..., [22.25, 22.75], where k is the k-th wind speed bin interval;

[0124] Finally, we get the data set {Nj_k} divided by wind speed bin intervals.

[0125] Step 10: Process the data set {Nj_k} obtained in step 9 as follows:

[0126] First, calculate the average wind speed data set {V a j_k};

[0127] Then calculate the average active power data set {P a j_k};

[0128] Finally, the dataset {V a j_k} and {P a j_k} are merged into a data set {VPj_k} with one-to-one correspondence between wind speed and power, where j is the jth wind direction bin interval and k is the kth wind speed bin interval;

[0129] According to the example unit, the data set {VPj_k} has 30 wind direction bin intervals, each wind direction interval has 40 wind speed bin intervals, so each wind direction interval has 40 sets of wind speed and power corresponding point sets.

[0130] Step 11: According to the data set {VPj_k} obtained in step 10 and the power curve characteristics, a polynomial fitting method is used to perform curve fitting on the wind speed and power data in each wind direction interval. The best fitting curve is determined according to the fitting effect. An optimal power curve fitting curve PC_j can be obtained for each wind direction interval. The specific method is as follows:

[0131] Firstly, the data in the data set {VPj_k} are imported into the Curve Fitting toolbox in the MATLAB software; secondly, wind speed is selected as the independent variable x of the fitting, power is selected as the dependent variable y of the fitting, and the polynomial fitting method is selected. Different x orders are set to perform curve fitting on 40 wind speed-power point sets. In this example, polynomials of 2 to 9 orders are fitted;

[0132] Then, based on the fitting results of 2nd to 9th order polynomials, the performance parameters such as RMSE, SSE, and R-square in each order fitting are statistically calculated. According to the principle that the smaller the RMSE and SSE, the better, and the closer the R-square is to 1, the better, the polynomial with the best fitting effect is selected, as shown in the attached figure. Figure 6 Shown is a diagram of the polynomial fitting process of the wind speed-power curve;

[0133] Finally, the above curve fitting is performed for each wind direction interval to obtain the set of power curves {PC_j} for all wind direction intervals. In this example, a total of 30 power curve fitting curves are obtained, as shown in the attached figure. Figure 7 Shown is a fitting diagram of 30 wind speed-power curves divided by wind direction range.

[0134] Step 12: Based on the power curve set {PC_j} obtained in step 11, calculate the power curves of 30 power curves in [V 切入, V 额定 ] The integral value in the wind speed interval is obtained to obtain the data set of the integral value of the power curve in each wind direction interval {P integral _j}.

[0135] Step 13: According to the maximum power generation principle, the average wind direction value of the wind direction interval where the integral value of the data set {P_j} obtained in step 12 is the largest is used as the second yaw error value Y_fx of the target unit;

[0136] Through calculation, the second yaw error value of the unit in this example is Y_fx=-2.5286.

[0137] Step 14: Verify the calculation conditions of the first yaw error value and the second yaw error value obtained in step 7 and step 13. If the verification fails, increase the number of samples and continue to execute steps 1 to 14. If the verification passes, directly execute the subsequent steps. The logic of the calculation condition verification is as follows (verification passes if any condition is met):

[0138] 1) The numerical signs of Y_fs and Y_fx are the same, and |Y_fs-Y_fx|≤g, where g is the threshold of the absolute value of the error, and in this example g=6;

[0139] 2) If the calculation condition 1) is not verified, the sample size is increased by the amount of data for each 2-month period and recalculated. When the number of sample increases = h, h is the threshold for the number of sample increases, in this example h = 6.

[0140] Step 15: After the verification condition of step 14 is satisfied, the average value of the first yaw error value Y_fs and the second yaw error value Y_fx obtained in step 7 and step 13 is used as the yaw static error value Y of the example unit;

[0141] Through calculation, the yaw static error value of the unit in this example is Y=-2.676.

[0142] This embodiment further provides a wind turbine yaw static error calculation system based on data analysis, including a calculation module, wherein the calculation module executes the wind turbine yaw static error calculation method based on data analysis.

[0143] This embodiment further provides a non-transitory computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, the method for calculating the static yaw error of a wind turbine generator system based on data analysis is implemented.

[0144] Ginseng Figure 8 As shown, this embodiment also provides an electronic device, including:

[0145] A memory 201 and a processor 202, wherein the memory 201 and the processor 202 are communicatively connected to each other, the memory 201 stores computer instructions, and the processor 202 executes the method for calculating the static yaw error of a wind turbine based on data analysis by executing the computer instructions.

[0146] The wind turbine yaw static error calculation method and system based on data analysis have the following technical effects:

[0147] 1) Based on the standardized distance screening method in the K-Nearest Neighbors algorithm, the abnormal active power data is eliminated, which improves the quality of sample data and provides effective data support for the accurate calculation of the unit's yaw static error.

[0148] 2) Based on the analysis of operating data, the maximum power calculation method in the wind speed range and the maximum power generation calculation method in the wind direction range are used to comprehensively calculate the yaw static error from the two dimensions of wind speed and wind direction, respectively. This method has the characteristics of easy implementation, low cost and high accuracy.

[0149] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for calculating the static yaw error of a wind turbine based on data analysis, characterized in that: It includes the following steps: Step 1: Obtain the historical operation data of the wind turbine through the wind turbine SCADA system to obtain the original data sample O; the historical operation data includes the wind speed, wind direction, and active power of the target unit; Step 2: Perform data cleaning on the original data sample O obtained in Step 1 to obtain the cleaned effective data set N; Step 3: Analyze the data set N obtained in Step 2 from the wind speed dimension, divide the data of the target model within the rated wind speed range into wind speed bin intervals at a step size of 0.5 m / s of wind speed to obtain the data set {Ni}, where i is the i-th wind speed bin interval; Step 4: For the data set {Ni} obtained in Step 3, divide each wind speed interval into wind direction bin intervals at a step size of 1° of wind direction to obtain the data set {Ni_j}, where j is the j-th wind direction bin interval; Step 5: Calculate the data set {Pi_j} of the average active power in each wind direction interval in the data set {Ni_j} in Step 4; Step 6: For the data set {Pi_j} obtained in step 5, according to the maximum power principle, calculate the yaw error data set {D a i_m}, and the yaw error weighting factor dataset {D n i_m}, m is the wind direction interval where the average active power is the largest; Step 7: The yaw error value dataset {D a i_m} and yaw error weighting factor dataset {D n i_m}, calculate the weighted average of the yaw errors in all wind speed intervals as the first yaw error value Y_fs of the target unit; Step 8: Analyze the data set N obtained in Step 2 from the wind direction dimension, divide all the data into wind direction bin intervals at a step size of 1° of wind direction to obtain the data set {Nj}, where j is the j-th wind direction bin interval; Step 9: For the data set {Nj} obtained in Step 8, divide each wind direction interval into wind speed bin intervals at a step size of 0.5 m / s of wind speed to obtain the data set {Nj_k}, where k is the k-th wind speed bin interval; Step 10: For the data set {Nj_k} obtained in Step 9, calculate the average wind speed and active power in each wind speed bin interval to obtain the data set {VPj_k} corresponding to wind speed and power, where j is the j-th wind direction bin interval and k is the k-th wind speed bin interval; Step 11: According to the data set {VPj_k} obtained in Step 10, perform curve fitting on the wind speed and power data of each wind direction interval to obtain the power curve set {PC_j} corresponding to all wind direction intervals; Step 12: Based on the power curve set {PC_j} obtained in step 11, calculate the power curve of each power curve in [V 切入 , V 额定 ] The integral value in the wind speed interval is used to obtain the data set of the integral value of the power curve in all wind direction intervals {P integral _j}; Step 13: According to the principle of maximum power generation, the data set {P integral The average wind direction value of the wind direction interval where the integral value in _j} is the largest is taken as the second yaw error value Y_fx of the target unit; Step 14: Perform calculation condition verification on the first yaw error value and the second yaw error value obtained in Step 7 and Step 13. If the verification fails, increase the sample size and continue to execute Steps 1 to 14. If the verification passes, directly execute the subsequent steps; Step 15: After the verification condition in Step 14 is satisfied, take the average value of the first yaw error value Y_fs obtained in Step 7 and the second yaw error value Y_fx obtained in Step 13 as the yaw static error value Y of the target unit.

2. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 1, characterized in that: The original data variables in Step 1 include the wind turbine status word S, wind direction D, wind speed V, active power P, pitch angle θ, yaw operation mode M, data accuracy at the 1-min level, and the data volume with a duration of at least 1 year or more. Denote the original data sample as O = {S, D, V, P, θ, M}; The data cleaning in Step 2 includes screening the data valid range and removing abnormal data. The data valid range screening conditions are as follows: 1) Status word S = normal power generation operation; 2)-180°≤D≤180°； 3)In 切入 -0.5≤V≤V 切出 ; 4) 0 < P ≤ 1.1 * rated power of the unit; 5) 0° ≤ θ < 89°; 6) Yaw operation mode M = non-yaw operation state; After the validity screening, obtain the data set X = {D, V, P, θ}; Abnormal data removal uses the standardized distance in the K nearest neighbor algorithm to remove abnormal active power data in the data set X, including: Divide the data set X into wind speed bin intervals according to the wind speed step size of 0.5m / s to obtain the data set {X_a}, where a is the number of wind speed bin intervals; Calculate the average data set {P_a} and standard deviation data set {S_a} of active power in each wind speed interval; According to the standard distance formula: Calculate the standardized distance data set {d_a} from all active power data P to the active power average value P_a in each wind speed interval; According to the elimination condition d_a>b*P_a, b is the threshold, and the data points whose normalized distance of active power in each wind speed interval is greater than b times the average active power are eliminated; after eliminating abnormal data, the data set N={D,V,P,θ} is obtained.

3. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 2, characterized in that: The specific division method of the wind direction bin interval in each wind speed interval in step 4 is: Determine the absolute wind direction angle Ф for yaw start according to the target aircraft model, and calculate Then the number of wind direction bin intervals in this wind direction segment is e * =2*e; Taking -e as the starting point and the step length of 1°, the wind direction bin interval is divided into [-e, -e+1), [-e+1, -e+2), [-e+2, -e+3), ..., [-e+j-1, -e+j], where j is the jth wind direction bin interval; Get the data set {Ni_j} divided by wind direction bin interval.

4. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 3, characterized in that: In step 6, according to the maximum power principle, the data set {Pi_j} obtained in step 5 is processed as follows: Filter out the data set of the wind direction interval with the maximum average active power in each wind speed interval {P max i_m}, where i is the i-th wind speed interval, and m is the wind direction interval where the average active power is the largest; Calculate the average wind direction in the m wind direction interval as the yaw error value in the wind speed interval, and obtain the data set {D a i_m}; Calculate the number of wind directions in the m wind direction interval as the weighting factor of the yaw error in the wind speed interval, and obtain the yaw error weighting factor data set {D n i_m}.

5. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 4, characterized in that: The calculation method of the first yaw error value Y_fs of the target unit in step 7 is: Y_fs = D a i_m*D n i_m, where i is the i-th wind speed interval; The specific division method of the N wind direction bin intervals of the data set in step 8 is: Determine the absolute wind direction angle Ф for yaw start according to the target aircraft model, and calculate Then the number of wind direction bin intervals in this wind direction segment is e * =2*e; Taking -e as the starting point and the step length of 1°, the wind direction bin interval is divided into [-e, -e+1), [-e+1, -e+2), [-e+2, -e+3), ..., [-e+j-1, -e+j], where j is the jth wind direction bin interval; Get the data set {Nj} divided by wind direction bin interval; The specific method for dividing the wind speed bin intervals in each wind direction interval in step 9 is: Determine the cut-in wind speed and cut-out wind speed according to the target model, and calculate The number of wind speed bins in this wind speed segment is V 切入 -0.25 is the starting point, and the step length is 0.5m / s to divide the wind speed bin interval [V 切入 -0.25,V 切入 +0.25), [V 切入 +0.25,V 切入 +3*0.25),...,[V 切入 +(2(k-1)-1)*0.25,V 切入 +(2(k-1)+1)*0.25), where k is the k-th wind speed bin interval; Get the data set {Nj_k} divided by wind speed bin interval; In step 10, the data set {Nj_k} obtained in step 9 is processed as follows: Calculate the average wind speed data set in each wind speed bin interval in all wind direction intervals {V a j_k}; Calculate the average active power data set in each wind speed bin interval in all wind direction intervals {P a j_k}; The dataset {V a j_k} and {P a j_k} are merged into the data set corresponding to wind speed and power {VPj_k}, where j is the jth wind direction bin interval and k is the kth wind speed bin interval.

6. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 5, characterized in that: In step 11, the wind speed and power data of each wind direction interval are curve fitted. Specifically, the least squares method is used for polynomial fitting. The polynomial order and fitting coefficient are selected according to the RMSE, SSE, and R-square parameters to obtain the power curve set {PC_j} corresponding to all wind direction intervals.

7. The method for calculating the static yaw error of a wind turbine generator system based on data analysis according to claim 6, characterized in that: The logic of verifying the calculation conditions of the first yaw error value Y_fs and the second yaw error value Y_fx in step 14 is that one of the following conditions is met: 1) The numerical signs of Y_fs and Y_fx are the same, and |Y_fs-Y_fx|≤g, where g is the threshold of the absolute value of the error; 2) If the calculation condition 1) fails to pass the verification, the number of samples is increased by the amount of data for each 2-month period and recalculated. When the number of sample increases = h, h is the threshold for the number of sample increases.

8. A wind turbine yaw static error calculation system based on data analysis, characterized in that: It comprises a calculation module, which executes the method for calculating the static yaw error of a wind turbine based on data analysis as described in any one of claims 1 to 7.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, a method for calculating a static yaw error of a wind turbine based on data analysis as described in any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for calculating the static yaw error of a wind turbine based on data analysis as described in any one of claims 1 to 7 by executing the computer instructions.

Images