A method for fitting wind turbine power curve using parameter model based on differential evolution

Abstract

The present invention discloses a method for fitting a wind turbine power curve using a parameter model based on differential evolution. The method cleans data collected by a wind turbine data acquisition and monitoring system based on wind speed information. A five-parameter logistic model is established to fit the cleaned power curve dataset. The logistic parameter model has highly similar characteristics to a standard wind turbine power curve, ensuring the accuracy of the curve fitting. A differential evolution algorithm with adaptive control parameter changes is used to identify the parameters of the five-parameter logistic model, ensuring the smoothness of the fitting curve while improving the fitting speed, ultimately obtaining a five-parameter logistic model that fits the wind turbine power curve. The method of the present invention is a data-based parameter modeling method with a complete data cleaning step. The selected parameter model takes into account the characteristics of both actual operating data and the characteristics of the standard wind turbine power curve, and has certain theoretical value and practical engineering significance.

Description

Technical Field

[0001] The present invention relates to an algorithm for fitting a wind turbine power curve, and in particular to a method for fitting a wind turbine power curve using a parameter model based on differential evolution. Background Art

[0002] Due to the limited reserves and non-renewable nature of fossil energy, as well as its significant environmental pollution, renewable energy is gaining popularity. Wind energy, a renewable energy source with minimal environmental pollution and enormous reserves, is considered a high-quality renewable energy source. In recent years, global installed wind power capacity has grown annually, as has wind power penetration. my country's installed onshore wind turbine capacity ranks first in the world, and offshore wind power is also booming. The wind power industry has become a highly sought-after new energy sector both domestically and internationally.

[0003] However, while the wind power industry is being vigorously developed, the continuous degradation of wind turbines has led to a series of problems in wind turbine operation and maintenance. During actual operation, due to the intermittent and uncertain nature of natural wind speeds, the actual wind speed measurement is inaccurate, resulting in a certain delay in wind turbine operation and maintenance personnel's assessment of the wind turbine's operating status, causing misjudgment of wind turbine control and fault diagnosis. Accurately assessing and diagnosing the operating status and health of wind turbines facilitates the rational planning of wind turbine operation and maintenance, and is also of practical significance for extending the service life of wind turbines.

[0004] During wind turbine operation, much operating status information and fault status information will be intuitively reflected in the wind turbine's power-wind speed curve. Therefore, accurately fitting the wind turbine power curve is the primary basis for studying the operating status and health status of wind turbines. For the data sets generated during actual wind turbine operation, how to clean abnormal data, select appropriate parameter models for fitting the data sets, identify the parameters involved in the parameter models based on the data information, and further analyze the actual physical meaning of the wind turbine's true power curve are the current research focuses in the wind power industry. Currently, wind turbine power curve fitting methods face the following challenges:

[0005] (1) The wind turbine power curve fitting method requires the original operating data of the wind turbine to be cleaned in a reasonable and standardized manner;

[0006] (2) The wind turbine power curve fitting method needs to consider the characteristics of the standard power curve and the characteristics of the actual operating data;

[0007] (3) The wind turbine power curve fitting method needs to ensure both fitting accuracy and fitting speed in order to realize online wind turbine status monitoring based on the wind turbine power curve;

[0008] Therefore, based on the challenges faced by the current wind turbine power curve fitting method, it is necessary to design a standardized wind turbine power curve fitting method to improve the curve fitting process. Summary of the Invention

[0009] This invention aims to improve and standardize existing research and technology, proposing a method for fitting wind turbine power curves using a parameter model based on differential evolution. This method designs a standardized wind turbine power curve fitting method. By cleaning the initial data, the accuracy and reliability of the power curve fitting can be improved. By establishing a parameter model based on operational data and considering both the characteristics of the standard power curve model and the operational data, the model results possess both theoretical value and practical application significance.

[0010] The object of the present invention is achieved by the following technical solution: a method for fitting a wind turbine power curve using a parameter model based on differential evolution, the method comprising the following steps:

[0011] 1) According to the total n pieces of wind turbine operation data collected by the SCADA system of the wind turbine to be evaluated in the corresponding demand period, the wind speed {v i}、Active power {P i} and pitch angle {β i The information constitutes the power curve information data set that needs to be preprocessed, which is recorded as Where i = 1, 2, 3, ..., n;

[0012] 2) The information data set in step 1) is analyzed according to the normal value range of active power and pitch angle corresponding to different wind speed conditions during the operation of the wind turbine. Clean the original data set, remove the obvious outliers and get the remaining n norm Operation data are recorded as the fan standard operation data set where i=1,2,3,…,n norm ;

[0013] 3) The fan standard operation data set after cleaning in step 2) Wind speed in and active power The five-parameter logistic model is established, and the power curve model of the wind turbine is recorded as P 5logistic (v,θ), where wind speed is the independent variable of the power curve model, active power is the dependent variable of the power curve model, θ is the set of five parameters of the five-parameter logistic model (A, B, C, D, G), where A represents the minimum power of the wind turbine, B represents the efficiency of the wind turbine in capturing the maximum wind energy, C represents the median of the maximum power generated by the wind turbine, D represents the maximum power of the wind turbine, and G represents the symmetry of the power curve;

[0014] 4) Setting the initial values θ0 of the five parameter sets θ = (A, B, C, D, G) of the five-parameter logistic model in step 3) by combining the meaning of each parameter of the five-parameter logistic model with the physical meaning of the wind turbine power curve to set the initial values of the parameter set, where:

[0015] The initial value of A is set to the active power in the standard operation data set of the wind turbine The minimum value of

[0016] The initial value of B is set to the slope of the line between the two data points with the largest Euclidean distance in the wind turbine standard operation data set;

[0017] The initial value of C is set to the active power in the standard operation data set of the wind turbine the mean of the maximum and minimum values;

[0018] The initial value of D is set to the active power in the fan standard operation data set The maximum value of

[0019] The initial value of G is set to 1, which means that the initial power curve is symmetrical;

[0020] 5) Based on the initial values θ0 of the five parameter set θ of the five-parameter logistic model in step 4), the wind turbine power curve model P in step 3) is modified using the differential evolution method. 5logistic The five parameters (A, B, C, D, G) in (v, θ) are used for parameter identification, and the standard operation data set of the fan is recorded. Each data point in is an individual, and the whole is a population. The five parameter sets corresponding to each individual are used as chromosome vectors. The donor vector is obtained by randomly extracting three original chromosome vectors and performing differential calculation. The binomial crossover operator is used for crossover operation to generate a test vector. The chromosome vector closest to the best chromosome vector in the two original chromosome vectors and the test vector is selected as the individual of the new generation population. Repeat the iteration until the set upper limit is reached to obtain the optimal five parameter set θ. best =(A best ,B best ,C best ,D best ,G best );

[0021] 6) Obtain the optimal wind turbine power curve model based on the optimal five parameter sets obtained in step 5)

[0022] Furthermore, in step 2), different wind speed ranges are divided, and abnormal point detection is performed on the wind turbine power curve according to the abnormal characteristics of the wind turbine operating state; the specific criteria for data cleaning based on wind speed information are as follows:

[0023] a) When the fan starts, if the information data set Wind speed v i Greater than the cut-in wind speed v cut_in , active power P i Below the active power threshold P thres Or if the value is negative, it is determined to be an abnormal data point. The judgment formula is listed as follows:

[0024]

[0025] b) The fan is running under low wind speed conditions. If the information data set Wind speed v i Greater than the cut-in wind speed v cut_in And less than the rated wind speed v of the fan rated , pitch angle β i Greater than the pitch angle threshold β thres , it is determined to be an abnormal data point. The judgment formula is listed as follows:

[0026]

[0027] c) The fan is running at high wind speed. If the information data set Wind speed v i Greater than the rated wind speed v rated and less than the cut-out wind speed v cut_off , active power P i Greater than the active power threshold P thres And it is less than the rated active power of the wind turbine and the active power threshold P thres The difference between , is determined to be an abnormal data point. The judgment formula is listed as follows:

[0028]

[0029] Furthermore, in step 3), the wind turbine standard operation data set The established wind turbine power curve model P 5logistic (v,θ) is described as follows:

[0030]

[0031] Furthermore, in step 5), the wind turbine power curve model P 5logistic When performing parameter identification on the five parameters (A, B, C, D, G) in (v, θ), the optimization problem of parameter θ = (A, B, C, D, G) is transformed into a nonlinear programming problem. The objective function is the squared error function, and the constraints are interval constraints on the five parameters (A, B, C, D, G) based on expert experience and factory specifications of the fan. The formula of the objective function is as follows:

[0032]

[0033] Among them, P e 5logistic (v i ) is the active power estimated using the wind turbine power curve model, P a (v i ) is the actual active power.

[0034] Furthermore, in step 5), the wind turbine power curve model P is modified using the differential evolution method. 5logistic The specific steps for parameter identification of the five parameters (A, B, C, D, G) in (v, θ) are as follows:

[0035] a) The wind turbine standard operation data set Medium wind speed Each data point is recorded as an individual v i , the total n norm Individuals form an initial population, that is, the data point population when the number of iterations is 0, recorded as

[0036]

[0037] in, is an individual in the initial population, and each individual has five parameters of the five-parameter logistic model as its chromosome vector, denoted as

[0038] b) According to the population size n norm Set the maximum number of iterations N iter , initial scaling factor F 0 , initial crossover probability and the probability τ of the crossover probability adjustment.

[0039] c) When the number of iterations is N, the Nth generation mutation process is described as follows: N Randomly extract two parameter vectors, namely chromosome vectors and At the same time, a parameter vector is randomly selected as the reference chromosome vector And satisfy k, i and j are mutually different integers; let the individual mutation of the Nth generation population be determined by the scaling factor F N Control the degree of mutation and obtain the (N+1)th generation donor vector by differential mutation Record

[0040]

[0041] Wherein, the scaling factor F N It is an adaptive fine-tuning value. The scaling factor is adjusted according to the number of iterations N to control the degree of variation. The scaling factor is large in the initial iteration process to retain the diversity of individuals in the population. When approaching the optimal solution, the scaling factor is small to improve global convergence. The scaling factor F N The specific adjustment strategy is recorded as

[0042]

[0043] d) To improve the diversity of individuals in the population, a binomial crossover operator is used to perform a crossover operation to generate the (N+1)th generation test vector Each element in the vector It can be obtained by the following formula

[0044]

[0045] Among them, rnbr i is a random integer in [1, D], where D is the dimension of the chromosome vector, rand j is a random number in [0, 1], As the crossover probability, it is an adaptive random value that maintains the diversity of the iterative population. The crossover probability The specific value strategy of

[0046]

[0047] e) After the crossover operation, the (N+1) generation population is selected and the value of the objective function to be optimized is calculated. and the target vector Select the best among them to form a new generation of population

[0048]

[0049] f) Repeat steps c) to e) until the number of iterations N reaches the maximum number of iterations N iter ; Finally, the optimal parameters of the five wind turbine power curve fitting models are obtained

[0050] Furthermore, the optimal wind turbine power curve model obtained in step 6) according to the optimal five parameter sets is It can be used as a reference for subsequent wind turbine status monitoring and fault detection. By using real-time wind turbine SCADA data to draw a real-time wind turbine power fitting curve, it helps wind turbine operation and maintenance personnel judge the wind turbine operating status. Based on the fitting power curve, various fault detection index values can be calculated, and real-time monitoring of fault indicators can predict wind turbine failures in advance.

[0051] Compared with the prior art, the present invention has the following innovative advantages and significant effects:

[0052] 1) In order to meet the accuracy requirements of wind turbine power curve fitting, a five-parameter logistic model that is consistent with the characteristics of the standard wind turbine power curve is selected. Physical meanings are given to the model parameters and initial values are selected. This takes into account the characteristics of the standard wind turbine power curve and the actual operating data characteristics, thereby improving the accuracy of wind turbine power curve fitting.

[0053] 2) In response to the requirement of rapid power curve fitting of wind turbines, the differential evolution method with adaptive control parameter changes is selected for parameter identification of the five-parameter logistic model. While maintaining the diversity of optimization parameters, the global convergence is improved, the efficiency of parameter optimization is enhanced, and the rapid power curve fitting of wind turbines is ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of a method for fitting a wind turbine power curve using a parameter model based on differential evolution according to the present invention;

[0055] Figure 2 is a wind speed-power scatter plot of the original data set in an embodiment of the present invention;

[0056] Figure 3 This is the result diagram after data cleaning in step 2 of the embodiment of the present invention;

[0057] Figure 4 Schematic diagram of the five-parameter logistic power curve of the wind turbine obtained by fitting in steps 3 and 4 of the embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0059] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0060] Example

[0061] This example builds a parameter power curve fitting model based on one month of operating data collected by the SCADA system of a wind turbine at a wind farm between May and June 2018. The SCADA system's data sampling interval is 5 minutes, and the data spans 27 days, from 10:55:00 AM on May 16, 2018, to 3:05 PM on June 11, 2018. The specific data variables and related data information included in the dataset are shown in Tables 1 and 2:

[0062] Table 1 Partial data of the SCADA system data set of a wind turbine in a wind farm

[0063] Data sequence number Collection time Actual value of pitch angle Average wind speed 10 minutes Average active power 10min 1 2018 / 5 / 16 10:55:00 -0.029 7.235 522 2 2018 / 5 / 16 11:00:00 -0.029 7.546 561 … … … … … 3178 2018 / 5 / 27 12:05:00 8.69 11.636 1483 3179 2018 / 5 / 27 12:10:00 3.29 11.452 1471 3180 2018 / 5 / 27 12:15:00 9.98 11.5 1509 … … … … …

[0064] Table 2 Variable information of the SCADA system data set of a wind turbine in a wind farm

[0065] variable name Variable meaning Variable units Collection time Real-time data collection time point Year / Month / Day Hour:Minute:Second <![CDATA[Actual value of blade pitch angle (β i )]]> Current wind turbine real-time pitch angle degree <![CDATA[Average wind speed for 10 min (v i )]]> Current wind turbine 10-minute average wind speed m / s <![CDATA[Average active power 10min (p i )]]> Current 10-minute average active power of the fan kW

[0066] The default implementation data set of the power curve data acquisition method in this embodiment is the 27-day operation data of the above-mentioned wind turbine. The method result is the wind turbine power curve fitting model obtained. The detailed implementation steps are as follows:

[0067] 1) According to the total n pieces of wind turbine operation data collected by the Supervisory Control and Data Acquisition (SCADA) system of the wind turbine to be evaluated in the corresponding demand period, the average wind speed {v i}、Average active power {P i} and pitch angle {β i} information constitutes the power curve information data set that needs to be preprocessed, and the information data set is recorded as Where i = 1, 2, 3, ..., n. According to the data information provided in Table 1 and Table 2, it includes all the necessary information required in this step (acquisition time, pitch angle, average wind speed 10 minutes, average active power 10 minutes), as shown in Figure 1. Figure 2 Shown is a scatter plot of the original data of the wind turbine wind speed-power curve in this step.

[0068] 2) The information data set in step 1) is analyzed according to the normal value range of active power and pitch angle corresponding to different wind speed conditions during the operation of the wind turbine. Clean the original data set, remove the obvious outliers and get the remaining n norm Operation data are recorded as the fan standard operation data set where i=1,2,3,…,nnorm The specific criteria for data cleaning based on wind speed information are as follows:

[0069] a) When the fan starts, if the information data set Wind speed v i Greater than the cut-in wind speed v cut_in , active power P i Below the active power threshold P thres Or if the value is negative, it is determined to be an abnormal data point. The judgment formula is listed as follows:

[0070]

[0071] b) The fan is running under low wind speed conditions. If the information data set Wind speed v i Greater than the cut-in wind speed v cut_in And less than the rated wind speed v of the fan rated , pitch angle β i Greater than the pitch angle threshold β thres , it is determined to be an abnormal data point. The judgment formula is listed as follows:

[0072]

[0073] c) The fan is running at high wind speed. If the information data set Wind speed v i Greater than the rated wind speed v rated and less than the cut-out wind speed v cut_off , active power P i Greater than the active power threshold P thres And it is less than the rated active power of the wind turbine and the active power threshold P thres The difference between , is determined to be an abnormal data point. The judgment formula is listed as follows:

[0074]

[0075] In this embodiment, P rated 1560kW, v cut_in is 3m / s, v rated is 12m / s, v cut_off is 28m / s, P thres is 10kW, β thres It is 4°. Figure 3 The result of data cleaning in this embodiment after this step is shown. It can be seen from the figure that the data cleaning effect is obvious, and most of the abnormal points are detected and eliminated.

[0076] 3) The standard operation data set of the fan obtained after cleaning the data in step 2) Perform five-parameter logistic parameter power curve fitting model P 5logistic Modeling of (v,θ):

[0077]

[0078] The wind speed is the independent variable of the power curve model, active power is the dependent variable of the power curve model, θ is the set of five parameters of the five-parameter logistic model (A, B, C, D, G), where A represents the minimum power of the wind turbine, B represents the efficiency of the wind turbine in capturing wind energy to the maximum extent, C represents the median of the maximum power generation of the wind turbine, D represents the maximum power of the wind turbine, and G represents the symmetry of the power curve.

[0079] 4) For the wind turbine power curve model P in step 3) 5logistic When performing parameter identification on the five parameters (A, B, C, D, G) in (v, θ), the identification and optimization problem of parameter θ = (A, B, C, D, G) is transformed into a nonlinear programming problem. The objective function is the square error function, and the constraints are interval constraints on the five parameters (A, B, C, D, G) based on expert experience and factory specifications of the fan. The formula of the objective function is as follows:

[0080]

[0081] Among them, P e 5logistic (v i ) is the active power estimated using the wind turbine power curve model, P a (v i ) is the actual active power.

[0082] 5) When setting the initial value θ0 of the five parameter set θ=(A, B, C, D, G) of the five-parameter logistic model in step 3), the initial value of the parameter set is set in combination with the meaning of each parameter of the five-parameter logistic model and the physical meaning of the wind turbine power curve, where:

[0083] The initial value of A is set to the active power in the standard operation data set of the wind turbine For this embodiment, parameter A0=0;

[0084] The initial value of B is set to the slope of the line between the two data points with the largest Euclidean distance in the fan standard operation data set. For this embodiment, the parameter

[0085] The initial value of C is set to the active power in the standard operation data set of the wind turbine The average of the maximum and minimum values. For this embodiment, the parameter

[0086] The initial value of D is set to the active power in the fan standard operation data set For this embodiment, the parameter D0=1560;

[0087] The initial value of G is set to 1, indicating that the initial power curve is symmetrical, that is, for this embodiment, the parameter G0=1.

[0088] 6) Based on the nonlinear programming problem in step 4) and the initial value θ0 of the five parameter set θ of the five-parameter logistic model in step 5), the wind turbine power curve model P in step 3) is transformed using the differential evolution method. 5logistic The five parameters (A, B, C, D, G) in (v, θ) are identified. The specific implementation steps are as follows:

[0089] a) The wind turbine standard operation data set Medium wind speed Each data point is recorded as an individual v i , the total n norm Individuals form an initial population, that is, the data point population when the number of iterations is 0, recorded as

[0090]

[0091] in, is an individual in the initial population, and each individual has five parameters of the five-parameter logistic model as its chromosome vector, denoted as In this embodiment, the initial chromosome vector can be recorded as

[0092] b) According to the population size n norm Set the maximum number of iterations N iter , initial scaling factor F 0 , initial crossover probability And the probability τ of crossover probability adjustment, in this embodiment, the maximum number of iterations N is selected iter =100, initial scaling factor F 0 =0.6, initial crossover probability and the probability of crossover probability adjustment τ = 0.2.

[0093] c) When the number of iterations is N, the Nth generation mutation process is described as follows: N Randomly extract two parameter vectors, namely chromosome vectors and At the same time, a parameter vector is randomly selected as the reference chromosome vector And satisfy k, i and j are mutually different integers. Let the individual mutation of the Nth generation population be determined by the scaling factor F N Control the degree of mutation and obtain the (N+1)th generation donor vector by differential mutation Record

[0094]

[0095] Wherein, the scaling factor F N It is an adaptive fine-tuning value. The scaling factor is adjusted according to the number of iterations N to control the degree of variation. The scaling factor is large in the initial iteration process to retain the diversity of individuals in the population. When approaching the optimal solution, the scaling factor is small to improve global convergence. The scaling factor F N The specific adjustment strategy is recorded as

[0096]

[0097] d) To improve the diversity of individuals in the population, a binomial crossover operator is used to perform a crossover operation to generate the (N+1)th generation test vector Each element in the vector It can be obtained by the following formula

[0098]

[0099] Among them, rnbr i is a random integer in [1, D], where D is the dimension of the chromosome vector, rand j is a random number in [0, 1], As the crossover probability, it is an adaptive random value that maintains the diversity of the iterative population. The crossover probability The specific value strategy of

[0100]

[0101] e) After the crossover operation, the (N+1) generation population is selected and the value of the objective function to be optimized is calculated. and the target vector Select the best among them to form a new generation of population

[0102]

[0103] f) Repeat steps c) to e) until the number of iterations N reaches the maximum number of iterations N iter Finally, the optimal parameters of the five wind turbine power curve fitting models are obtained. For this embodiment, the parameters of the five optimal wind turbine power curve fitting models are Get the final optimal wind turbine power curve fitting model writing:

[0104]

[0105] Figure 4 The power curve obtained by fitting the data set corresponding to this embodiment after this step is shown. The fitted power curve is a thick solid line, and the data points are marked with ".".

[0106] The present invention proposes a method for fitting a wind turbine power curve using a parameter model based on differential evolution. The method mainly consists of cleaning the initial data set, establishing a five-parameter logistic power curve fitting model for wind speed and power data, and using the differential evolution method to perform parameter identification on the parameter model. Figure 1 The present invention is a specific process for the implementation and application of a method for fitting a wind turbine power curve based on a parameter model of differential evolution. Figure 1 According to the process shown in , the wind turbine power curve dataset is preprocessed and finally fitted to obtain the parameter model power curve. Figure 2-Figure 4 The present invention uses a differential evolution-based parameter model to fit a wind turbine power curve. The results of each step in wind turbine power curve fitting modeling are presented. By establishing a parameter model and optimizing its parameters using differential evolution, the characteristics of the standard wind turbine power curve model and parameter identification based on operating data are considered. Therefore, this method standardizes the wind turbine power curve fitting process and can produce a more accurate and realistic wind turbine power curve. This invention has guiding significance for subsequent research on wind turbine operating status, fault detection, and early warning.

[0107] The above description is only a preferred embodiment of the present invention. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can use the above disclosed methods and technical contents to make many possible changes and modifications to the technical solution of the present invention without departing from the scope of the technical solution of the present invention, or modify it into an equivalent embodiment with equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of protection of the technical solution of the present invention.

Claims

1. A method for fitting a wind turbine power curve using a parameter model based on differential evolution, characterized in that: The following steps are involved: 1) According to the total n pieces of wind turbine operation data information collected by the data acquisition and monitoring system of the wind turbine to be evaluated in the corresponding demand cycle, the wind speed {v i }、Active power {P i } and pitch angle {β i The information constitutes the power curve information data set that needs to be preprocessed, which is recorded as Where i = 1, 2, 3, ..., n; 2) The information data set in step 1) is analyzed according to the normal value range of active power and pitch angle corresponding to different wind speed conditions during the operation of the wind turbine. Clean the original data set, remove the obvious outliers and get the remaining n norm Operation data are recorded as the fan standard operation data set where i=1,2,3,…,n norm ; 3) The fan standard operation data set after cleaning in step 2) Wind speed in and active power The five-parameter logistic model is established, and the power curve model of the wind turbine is recorded as P 5logistic (v,θ), where wind speed is the independent variable of the power curve model, active power is the dependent variable of the power curve model, θ is the set of five parameters of the five-parameter logistic model (A, B, C, D, G), where A represents the minimum power of the wind turbine, B represents the efficiency of the wind turbine in capturing wind energy to the maximum extent, C represents the median of the maximum power generation of the wind turbine, D represents the maximum power of the wind turbine, and G represents the symmetry of the power curve; Wind turbine power curve model P 5logistic (v,θ) is described as follows: 4) Setting the initial values θ0 of the five parameter sets θ = (A, B, C, D, G) of the five-parameter logistic model in step 3) by combining the meaning of each parameter of the five-parameter logistic model with the physical meaning of the wind turbine power curve to set the initial values of the parameter set, where: The initial value of A is set to the active power in the standard operation data set of the wind turbine The minimum value of The initial value of B is set to the slope of the line between the two data points with the largest Euclidean distance in the wind turbine standard operation data set; The initial value of C is set to the active power in the standard operation data set of the wind turbine the mean of the maximum and minimum values; The initial value of D is set to the active power in the fan standard operation data set The maximum value of The initial value of G is set to 1, which means that the initial power curve is symmetrical; 5) Based on the initial values θ0 of the five parameter set θ of the five-parameter logistic model in step 4), the wind turbine power curve model P in step 3) is modified using the differential evolution method. 5logistic The five parameters (A, B, C, D, G) in (v, θ) are used for parameter identification, and the standard operation data set of the fan is recorded. Each data point in is an individual, and the whole is a population. The five parameter sets corresponding to each individual are used as chromosome vectors. The donor vector is obtained by randomly extracting three original chromosome vectors and performing differential calculation. The binomial crossover operator is used for crossover operation to generate a test vector. The chromosome vector closest to the best chromosome vector in the two original chromosome vectors and the test vector is selected as the individual of the new generation population. Repeat the iteration until the set upper limit is reached to obtain the optimal five parameter set θ best =(A best ,B best ,C best ,D best ,G best ); 6) Obtain the optimal wind turbine power curve model based on the optimal five parameter sets obtained in step 5) 2. The method for fitting a wind turbine power curve using a parameter model based on differential evolution according to claim 1, characterized in that: In step 2), different wind speed ranges are divided, and abnormal point detection is performed on the wind turbine power curve based on the abnormal characteristics of the wind turbine operating state; the specific criteria for data cleaning based on wind speed information are as follows: a) When the fan starts, if the information data set If the wind speed in the data point is greater than the cut-in wind speed and the active power is lower than the active power threshold or is negative, it is determined to be an abnormal data point. b) The wind turbine operates under low wind speed conditions, that is, the information data set The wind speed in is greater than the cut-in wind speed and less than the rated wind speed of the wind turbine. If the pitch angle is greater than the pitch angle threshold, it is determined to be an abnormal data point. c) The wind turbine is operating under high wind speed conditions, that is, the information data set The wind speed in is greater than the rated wind speed of the wind turbine and less than the cut-out wind speed. If the active power is greater than the active power threshold and less than the difference between the rated active power of the wind turbine and the active power threshold, it is determined to be an abnormal data point.

3. The method for fitting a wind turbine power curve using a parameter model based on differential evolution according to claim 1, characterized in that: In the step 5), the wind turbine power curve model P 5logistic When performing parameter identification on the five parameters (A, B, C, D, G) in (v, θ), the optimization problem of parameter θ = (A, B, C, D, G) is transformed into a nonlinear programming problem. The objective function is the squared error function, and the constraints are interval constraints on the five parameters (A, B, C, D, G) based on expert experience and factory specifications of the fan. The formula of the objective function is as follows: Among them, P e 5logistic (v i ) is the active power estimated using the wind turbine power curve model, P a (v i ) is the actual active power.

4. The method for fitting a wind turbine power curve using a parameter model based on differential evolution according to claim 1, characterized in that: In step 5), the wind turbine power curve model P is analyzed using differential evolution method. 5logistic The specific steps for parameter identification of the five parameters (A, B, C, D, G) in (v, θ) are as follows: a) The wind turbine standard operation data set Medium wind speed Each data point is recorded as an individual v i , the total n norm Individuals form an initial population, that is, the data point population when the number of iterations is 0, recorded as in, is an individual in the initial population, and each individual has five parameters of the five-parameter logistic model as its chromosome vector, denoted as b) According to the population size n norm Set the maximum number of iterations N iter , initial scaling factor F 0 , initial crossover probability and the probability τ adjusted by the crossover probability; c) When the number of iterations is N, the Nth generation mutation process is described as follows: N Randomly extract two parameter vectors, namely chromosome vectors and At the same time, a parameter vector is randomly selected as the reference chromosome vector And satisfy k, i and j are mutually different integers; let the individual mutation of the Nth generation population be determined by the scaling factor F N Control the degree of mutation and obtain the (N+1)th generation donor vector by differential mutation Record Wherein, the scaling factor F N It is an adaptive fine-tuning value that adjusts the scaling factor according to the number of iterations N to control the degree of variation. The scaling factor F N The specific adjustment strategy is recorded as d) To improve the diversity of individuals in the population, a binomial crossover operator is used to perform a crossover operation to generate the (N+1)th generation test vector Each element in the vector It can be obtained by the following formula Among them, rnbr i is a random integer in [1, D], where D is the dimension of the chromosome vector, rand j is a random number in [0, 1], As the crossover probability, it is an adaptive random value that maintains the diversity of the iterative population. The crossover probability The specific value strategy of e) After the crossover operation, the (N+1) generation population is selected and the value of the objective function to be optimized is calculated. and the target vector Select the best among them to form a new generation of population f) Repeat steps c) to e) until the number of iterations N reaches the maximum number of iterations N iter ; Finally, the optimal parameters of the five wind turbine power curve fitting models are obtained 5. The method for fitting a wind turbine power curve using a parameter model based on differential evolution according to claim 1, characterized in that: The optimal wind turbine power curve model obtained based on the optimal five parameter sets in step 6) It can be used as a reference for subsequent wind turbine status monitoring and fault detection. By using real-time wind turbine SCADA data to draw a real-time wind turbine power fitting curve, it helps wind turbine operation and maintenance personnel judge the wind turbine operating status. Based on the fitting power curve, various fault detection index values can be calculated, and real-time monitoring of fault indicators can predict wind turbine failures in advance.

Images