A modeling method and apparatus for typical power output characteristics of multiple offshore wind farms.

Abstract

This invention discloses a modeling method and apparatus for typical power output characteristics of multiple offshore wind farms. The method includes: fitting various types of parameter distributions to the first wind speed data of the offshore wind farm to determine the optimal distribution type based on performance evaluation indicators; transforming the second wind speed data to obtain wind farm power output data, and generating edge distribution data of the offshore wind farm power output based on the wind farm power output data; obtaining the joint distribution function of the offshore wind farm based on the correlation functions corresponding to the edge distribution data and different types of vine structures, and determining the optimal correlation function based on the initial optimal correlation function and the optimal vine structure; performing inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint power output scenario data, thus generating joint typical power output scenario data of the offshore wind farm. This invention solves the problem of low accuracy in modeling wind power output characteristics and multi-wind farm power output scenarios caused by the intermittency and randomness of offshore wind power output.

Description

Technical Field

[0001] This invention relates to the field of power output characteristic modeling technology, and in particular to a modeling method and apparatus for typical power output characteristics of multiple offshore wind farms. Background Technology

[0002] Currently, the modeling technology for the output characteristics of onshore wind power is relatively mature. Offshore wind power has the advantages of not occupying land, higher wind speed, and greater power generation compared to onshore wind power. However, the stronger anti-peak shaving of offshore wind power places higher demands on the reasonable modeling of the output curve.

[0003] The commonly used method in the current technology is to fit the random distribution characteristics of wind speed using a two-parameter Weibull distribution based on the historical wind speed data of each wind farm (Xu Qianyao, Kang Chongqing, Zhang Ning, et al. Discussion on the output characteristics and absorption problems of offshore wind power [J]. Automation of Electric Power Systems, 2011, 35(22):54-59.), and generate the output data of the wind farm by combining the characteristics of the units. The random dependence of the output of multiple wind farms is modeled by the Copula function (Papaefthymiou G, Kurowicka D. Using copulas for modeling stochastic dependence in power system uncertainty analysis [J]. IEEE Transactions on Power Systems, 2008, 24(1):40-49.). Finally, cluster analysis of the output scenarios of the wind farm is carried out by clustering methods such as k-means clustering, spectral clustering, and Gaussian mixture clustering.

[0004] Although fitting the random distribution characteristics of wind speed based on the two-parameter Weibull distribution is currently the most commonly used method for wind speed fitting, considering that sea breezes are affected by topography, climate, and seasonal factors differently than land breezes, there are distributions that outperform the two-parameter Weibull distribution in fitting performance. If non-parametric distributions are used for wind speed fitting, the method requires high-quality historical data, and its fitting performance will be affected when the scale and completeness of wind speed data for the wind farm group to be built are limited. For wind farms of different geographical locations and sizes, it is difficult to accurately describe the relevant structure and tail characteristics of the wind farm using a single Copula function. Traditional Copulas are mostly suitable for describing the correlation between two random variables. For multiple wind farms, the number of random variables increases, the amount of data increases, and the calculation becomes more complex. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] To address this issue, this invention proposes a modeling method for typical power output characteristics of multiple offshore wind farms, which solves the problem of low accuracy in modeling wind power output characteristics and multiple wind farm power output scenarios caused by the intermittency and randomness of offshore wind power output.

[0007] Another objective of this invention is to provide a modeling device for the typical power output characteristics of multiple offshore wind farms.

[0008] To achieve the above objectives, this invention proposes a modeling method for typical power output characteristics of multiple offshore wind farms, comprising:

[0009] Multiple types of parameter distribution fitting were performed on the first wind speed data of offshore wind farms to determine the optimal distribution type based on performance evaluation indicators;

[0010] Based on the optimal distribution type, the second wind speed data is transformed to obtain wind farm output data, and the edge distribution data of offshore wind farm output is generated based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data;

[0011] The joint distribution function of the offshore wind farm is obtained based on the marginal distribution data and the correlation functions corresponding to different types of vine structures. The optimal correlation function is then determined based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function.

[0012] The sample data obtained by sampling the optimal correlation function is transformed by inverse function to generate joint output scenario data, and joint typical output scenario data of offshore wind farms is generated according to the preset clustering method and the joint output scenario data.

[0013] In addition, the modeling method for typical power output characteristics of multiple offshore wind farms according to the above embodiments of the present invention may also have the following additional technical features:

[0014] Furthermore, in one embodiment of the present invention, the step of fitting multiple types of parameter distributions to the first wind speed data of the offshore wind farm to determine the optimal distribution type of the fit based on performance evaluation indicators includes:

[0015] Obtain the initial wind speed data for offshore wind farms;

[0016] The second wind speed data is obtained based on the first wind speed data and the surface roughness of the actual ocean conditions.

[0017] The probability density function describing the random distribution characteristics of wind speed is obtained by performing parameter estimation for each type of wind speed data obtained from the second wind speed data classification output calculation.

[0018] The probability density function is fitted with performance indicators to test its performance, and a comprehensive evaluation index is obtained based on the performance test results and the performance indicators to determine the optimal distribution type.

[0019] Further, in one embodiment of the present invention, the step of converting the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generating edge distribution data of offshore wind farm output based on the wind farm output data, includes:

[0020] Probability sampling is performed on the optimal wind speed distribution data in the optimal distribution type to obtain the third wind speed data of the offshore wind farm based on the probability sampling results.

[0021] The least squares method is used to fit the power-wind speed data curve of the wind turbine, and the power data corresponding to the third wind speed data is obtained based on the curve fitting result.

[0022] The power data are arranged in monotonically increasing order, and the marginal distribution data of offshore wind farm output is constructed based on the arrangement results and empirical distribution methods.

[0023] Further, in one embodiment of the present invention, the step of obtaining the joint distribution function of the offshore wind farm based on the edge distribution data and the correlation functions corresponding to different types of vine structures, and determining the optimal correlation function based on the initial optimal correlation function obtained from the joint distribution function and the optimal vine structure, includes:

[0024] A C-vine structure was established, and various types of Copula functions were selected to obtain the joint distribution function of offshore wind farms;

[0025] The Euclidean distance between the fitted optimal bivariate Copula function and the empirical Copula distribution is calculated based on the joint distribution function, and the Copula function with the smallest comparison distance is taken as the initial optimal Copula function.

[0026] The optimal vine structure is determined by comparing the CvM distances corresponding to the C-vine structures, and the optimal Copula function corresponding to the optimal vine structure is determined based on the initial optimal Copula function.

[0027] Furthermore, in one embodiment of the present invention, when the C-vine structure is selected, a multivariate density function is constructed, and the method further includes:

[0028] a) Calculate the rank correlation coefficient between offshore wind farms, select the power data group with a larger average rank correlation coefficient with other wind farms as the main variable of the first-level tree, and the power data group of the remaining wind farms as the secondary variable.

[0029] b) Calculate the Copula function between the main variable and the secondary variable for offshore wind farms, and calculate the Euclidean distance between the Copula function and the empirical Copula function of the sample data to determine the optimal binary Copula function;

[0030] c) Sample the Copula function obtained in b) to obtain a new data set, repeat a) to determine the primary and secondary variables of the next level tree, and repeat b) to iteratively obtain the parameters of the Copula function;

[0031] d) Using the Copula function parameters obtained in each level of the tree in c) as initial values, the multivariate density function is obtained by the maximum likelihood method.

[0032] Furthermore, in one embodiment of the present invention, the step of performing an inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint output scenario data includes:

[0033] The sample data is generated by sampling the optimal Copula function based on the initial power output data of the offshore wind farm;

[0034] The combined power output scenario data of each offshore wind farm is obtained by calculating the sample data based on the inverse functions of each distribution function of the combined distribution function.

[0035] Furthermore, in one embodiment of the present invention, generating joint typical power output scenario data of offshore wind farms based on a preset clustering method and the joint power output scenario data includes:

[0036] The number of clusters is determined based on the preset task requirements;

[0037] The k-means clustering algorithm is used to cluster the joint output scenario data to obtain the joint typical output scenario data with the number of clusters.

[0038] Furthermore, in one embodiment of the present invention, before performing various types of parameter distribution fitting on the initial wind speed data of the offshore wind farm, the method further includes: preprocessing erroneous data and default data in the first wind speed data.

[0039] Furthermore, in one embodiment of the present invention, the performance evaluation indicators include the Akaike Information Criterion, root mean square error, coefficient of determination, and test value.

[0040] To achieve the above objectives, another aspect of the present invention proposes a modeling device for typical power output characteristics of multiple offshore wind farms, comprising:

[0041] The optimal distribution type determination module is used to fit various types of parameter distributions to the first wind speed data of offshore wind farms in order to determine the optimal distribution type of the fit based on performance evaluation indicators.

[0042] The edge distribution data determination module is used to convert the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generate edge distribution data of offshore wind farm output based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data;

[0043] The optimal correlation function determination module is used to obtain the joint distribution function of the offshore wind farm based on the marginal distribution data and the correlation functions corresponding to different types of vine structures, and to determine the optimal correlation function based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function.

[0044] The power output scenario data calculation module is used to perform inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint power output scenario data, and generate joint typical power output scenario data of offshore wind farms according to the preset clustering method and the joint power output scenario data.

[0045] The modeling method and apparatus for typical power output characteristics of multiple offshore wind farms in this invention solves the problem of low accuracy in modeling wind power output characteristics and multiple wind farm power output scenarios caused by the intermittency and randomness of offshore wind power output.

[0046] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0047] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0048] Figure 1 This is a flowchart of a modeling method for typical power output characteristics of multiple offshore wind farms according to an embodiment of the present invention;

[0049] Figure 2 This is a flowchart illustrating the determination of the optimal distribution type for fitting based on performance evaluation metrics according to an embodiment of the present invention.

[0050] Figure 3 This is a flowchart illustrating the generation of edge distribution data of offshore wind farm output based on wind farm output data, according to an embodiment of the present invention.

[0051] Figure 4 This is a flowchart illustrating the determination of the optimal correlation function based on the initial optimal correlation function obtained from the joint distribution function and the optimal vine structure, according to an embodiment of the present invention.

[0052] Figure 5 This is a schematic diagram of a 5-variable and N-variable C-vine Copula model according to an embodiment of the present invention;

[0053] Figure 6 This is a schematic diagram of a 5-variable and N-variable D-vine Copula model according to an embodiment of the present invention;

[0054] Figure 7 This is a schematic diagram of the structure of a modeling device for typical power output characteristics of multiple offshore wind farms according to an embodiment of the present invention. Detailed Implementation

[0055] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0057] The following describes, with reference to the accompanying drawings, a modeling method and apparatus for typical power output characteristics of multiple offshore wind farms according to embodiments of the present invention.

[0058] Figure 1 This is a flowchart of a modeling method for typical power output characteristics of multiple offshore wind farms according to an embodiment of the present invention.

[0059] like Figure 1 As shown, the method includes, but is not limited to, the following steps:

[0060] S1, perform various types of parameter distribution fitting on the first wind speed data of the offshore wind farm, and determine the optimal distribution type of the fitting based on the performance evaluation index;

[0061] S2, transform the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generate edge distribution data of offshore wind farm output based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data;

[0062] S3. Based on the marginal distribution data and the correlation functions corresponding to different types of vine structures, the joint distribution function of the offshore wind farm is obtained, and the optimal correlation function is determined based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function.

[0063] S4. Perform inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint output scenario data, and generate joint typical output scenario data of offshore wind farms based on the preset clustering method and joint output scenario data.

[0064] The modeling method for typical power output characteristics of multiple offshore wind farms according to embodiments of the present invention solves the problem of low accuracy in modeling wind power output characteristics and multiple wind farm power output scenarios caused by the intermittency and randomness of offshore wind power output.

[0065] Figure 2 The flowchart provided in this embodiment of the invention describes how to determine the optimal distribution type based on performance evaluation metrics, as shown below. Figure 2 As shown, it includes the following sub-steps:

[0066] S101, acquire the first wind speed data of the offshore wind farm;

[0067] S102, the second wind speed data is obtained based on the first wind speed data and the surface roughness of the actual ocean conditions;

[0068] S103, perform parameter estimation for each type of wind speed data obtained from the power calculation of the second wind speed data classification to obtain the probability density function used to describe the random distribution characteristics of wind speed;

[0069] S104. The fitting performance of the probability density function is tested based on the performance evaluation index, and a comprehensive evaluation index is obtained based on the performance test results and the performance evaluation index to determine the optimal distribution type.

[0070] Understandably, this invention acquires wind speed data from N offshore wind farms, performs various types of parameter distribution fitting on the wind speed data, and uses the Akaike Information Criterion (AIC), root mean square error (RMSE), and coefficient of determination (R²) to determine the wind speed data. 2 The KS test value and four other indicators are used to determine the comprehensive index, and then the best distribution type for fitting is determined. The output power of the wind farm is obtained by using the wind turbine power-wind speed conversion relationship. Then, 24 marginal distributions of output power are constructed for each wind farm, corresponding to the 24h daily characteristics.

[0071] Figure 3 This is a flowchart illustrating the generation of edge distribution data of offshore wind farm output based on wind farm output data, as provided in an embodiment of the present invention. Figure 3 As shown, it includes the following sub-steps:

[0072] S201, perform probability sampling on the optimal wind speed distribution data in the optimal distribution type to obtain the third wind speed data of the offshore wind farm based on the probability sampling results.

[0073] S202, the least squares method is used to fit the power-wind speed data curve of the wind turbine, and the power data corresponding to the third wind speed data is obtained based on the curve fitting result.

[0074] S203: Arrange the power data in monotonically increasing order, and construct the edge distribution data of offshore wind farm output based on the arrangement results and empirical distribution method.

[0075] Specifically, the steps of determining the optimal distribution type based on performance evaluation indicators and generating edge distribution data of offshore wind farm output based on wind farm output data in the present invention specifically include:

[0076] S1.1) Obtain wind speed data of the offshore wind farm to be built through the climate research center or power generation manufacturer. Specifically, it is necessary to obtain the wind speed v0 at the height h0 of each offshore wind farm and the surface roughness z0 of the actual ocean state. The wind speed v at the height h of the wind turbine hub can be obtained by formula (1).

[0077]

[0078] S1.2) Based on the obtained wind speed data, a 24-hour classification output is performed. For each type of wind speed data, the maximum likelihood method is used to estimate the parameters of various distribution types, resulting in a probability density function describing the random distribution characteristics of wind speed. The distribution types include Rayleigh distribution, Weibull distribution, Gamma distribution, generalized Gamma distribution, and P-Max distribution. The probability density function expressions for each distribution are as follows:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084] In the formula, v is the fitted wind speed, and a, b, and c are the distribution parameters.

[0085] S1.3) Using the Akaike Information Criterion (AIC), root mean square error (RMSE), and coefficient of determination (R²) 2The four indicators, namely the KS test value, are used to test the fitting performance of the probability density function fitted in S1.2). Among them, the Akaike Information Criterion is used to evaluate the influence of the number of parameters on the fitting performance, which can be solved by Equation (7). The smaller the value, the better the fitting performance. The Root Mean Square Error is used to evaluate the dispersion of the data from the true value, which can be obtained by Equation (8). The smaller the value, the better the fitting performance. The Coefficient of Determination is used to evaluate the dispersion of the data from the mean, which can be solved by Equation (9). The closer the value is to 1, the better the fitting performance. The KS test is used to evaluate the maximum difference between the fitted distribution and the true value, which can be solved by Equation (10). The smaller the value, the better the fitting performance.

[0086]

[0087] Where k represents the number of parameters in the density function, n represents the amount of wind speed data fitted, and f(v) i () indicates wind speed v i The density function value.

[0088]

[0089] in For wind speed v i The corresponding cumulative distribution function value.

[0090]

[0091] in is the average value of the cumulative distribution corresponding to wind speed vi.

[0092]

[0093] S1.4) Define indicator variables Where Q represents the type of performance metric, from AIC, RMSE, R 2 In KS, five different distribution types are selected, where a = 1, 2, ..., 5. The AIC values ​​obtained from the five distributions are sorted in monotonically decreasing order to obtain the max(AIC). a ), ..., min(AIC a ),correspond The values ​​range from 1, 2, ..., 5, indicating that the larger the AIC value, the worse the fitting performance. The smaller the value, the better. For R 2 The indicator is taken as 1-R 2 Sort to get The comprehensive index M for wind farm performance evaluation is obtained by fitting five distribution types using equation (11). a The larger the value, the better the fit of this type of distribution.

[0094]

[0095] S1.5) Based on the wind turbine power-wind speed data provided by the wind turbine manufacturer, the wind turbine power-wind speed curve is fitted by the least squares method to obtain the wind turbine power P corresponding to any wind speed, as shown in equation (12).

[0096]

[0097] Where V min V is the cut-in wind speed for the fan. split V is the split wind speed of the fan. rated P is the rated wind speed of the fan. rated V is the rated power of the fan. max To cut off the wind speed of the fan, a i ,b i ,c i ,d i , i = 1, 2 are the target fitting parameters.

[0098] S1.6) Probabilistically sample the optimal wind speed distribution of each offshore wind farm obtained in step S1.3) to obtain n days of wind speed data v for N offshore wind farms. For i = 1, 2, ..., n, and j = 1, 2, ..., N, the power data p of N offshore wind farms over n days can be obtained through step S1.4). i = 1, 2, ..., n, j = 1, 2, ..., N. Power data for wind farm k in the first hour (t = 1). For example, arranging the power data in monotonically increasing order yields (x1, x2, ..., x...). n The power output edge distribution F of wind farm k in the first hour of n days is constructed using the empirical distribution method of equation (13). k (x).

[0099]

[0100] Where n i (i = 1, 2, ..., n) is x i The frequency of occurrence of (i = 1, 2, ..., n), (n1 + n2 + ... + n) n =1), k=1,2,...,n-1

[0101] Ultimately, edge distribution data of power output from multiple offshore wind farms were obtained. t = 1, 2, ..., 24.

[0102] Figure 4The flowchart provided in this embodiment of the invention describes the determination of the optimal correlation function based on the initial optimal correlation function obtained from the joint distribution function and the optimal vine structure. Figure 4 As shown, it includes the following sub-steps:

[0103] S301, establish the C-vine structure and select various types of Copula functions to obtain the joint distribution function of the offshore wind farm;

[0104] S302, calculate the Euclidean distance between the fitted optimal bivariate Copula function and the empirical Copula distribution based on the joint distribution function, and take the Copula function with the smallest comparison distance as the initial optimal Copula function;

[0105] S303, compare the CvM distances corresponding to the C-vine structures to determine the optimal vine structure, and then determine the optimal Copula function corresponding to the optimal vine structure based on the initial optimal Copula function.

[0106] Specifically, C-vine and D-vine Copula structures are established, and Gaussian-Copula, t-Copula, Gumbel-Copula, Clayton-Copula, and Frank-Copula functions are selected respectively to obtain the joint distribution of multiple offshore wind farms according to Equation (14). Then, by comparing the Euclidean distances, the Copula function with the smaller distance is selected as the optimal Copula function. Subsequently, the optimal vine structure is determined by comparing the CvM distances corresponding to the C-vine and D-vine structures, and then the Copula model of the optimal vine structure is determined:

[0107] F(x1,x2,...,x n )=C(F1(x1),F2(x2),...,F N (x n ))(14)

[0108] S2.1) Select C-vine (model such as...) Figure 5 When the structure shown is a vine, the multivariate density function can be decomposed into equation (15):

[0109]

[0110] Where f k Let θ be the density function, c be a bivariate Copula function, and θ be its parameter.

[0111] The specific steps for obtaining this information are as follows:

[0112] S2.1.1) The Kendall rank correlation coefficient τ between each pair of offshore wind farm groups is obtained using equation (16). The power data group with a larger average Kendall rank correlation coefficient with other wind farm groups is selected as the principal variable of the first-level tree, and the power data group of the remaining wind farms is selected as the secondary variable.

[0113]

[0114] Where, N c N represents the sequential logarithm of the power data of the two wind farm groups. d This represents the out-of-order logarithm of the power data for the two wind farm groups.

[0115] S2.1.2) Calculate the Copula function between the main variable and the wind farm group, respectively. Calculate the Euclidean distance between the Copula function and the empirical Copula function of the sample data using equation (17). The smaller the distance, the better the fitting performance. Use this to determine the optimal bivariate Copula function.

[0116]

[0117] Where j and k are two different wind farm group numbers, and 1≤j,k≤N;

[0118] S2.1.3) Sample the Copula function obtained in S2.1.2) to obtain a new data set, repeat S2.1.1) to determine the primary and secondary variables of the next level tree, and repeat S2.1.2) to iteratively obtain the parameters of the Copula function.

[0119] S2.1.4) Using the Copula parameters obtained in each level of the tree in S2.1.3) as initial values, the maximum likelihood method is used to obtain equation (15).

[0120] S2.2) Select D-vine (model such as...) Figure 6 When the structure shown is a vine, the multivariate density function can be decomposed into equation (18):

[0121]

[0122] Considering that the relationships between the variables processed by D are independent, there is no need to select primary and secondary variables. The parameters of each Copula function can be obtained sequentially through iteration. The iteration steps are the same as in S2.1. Finally, equation (18) is obtained by the maximum likelihood method.

[0123] S2.3) Calculate the CvM distance between the Copula function and the empirical Copula function of the sample data under the two vine structures using equation (19). The smaller the value, the better the fitting performance. In this way, the optimal vine structure Copula model can be determined.

[0124]

[0125] Furthermore, the optimal Copula function can be obtained through step S3. t = 1, 2, ..., 24. Let the output of the wind farm numbered n at time t be denoted as . Sample data is generated by sampling each Copula function. n = 1, 2, ..., N. The power output of each wind farm can be calculated using equation (20) based on the inverse function of each distribution function. n = 1, 2, ..., N.

[0126] By repeating the sampling M times, M sets of output data can be obtained. m=1,2,...,M, n=1,2,...,N.

[0127]

[0128] Where F -1 It is the inverse function of the power distribution function. The sample data obtained from sampling,

[0129] n=1,2,...,N, t=1,2,...,24.

[0130] Furthermore, the k-means clustering algorithm is used to reduce the scenarios based on the combined power output data obtained in step S4. The number of clusters k is determined according to the task requirements, resulting in k typical combined power output scenarios for multiple offshore wind farms.

[0131] The modeling method for typical power output characteristics of multiple offshore wind farms according to embodiments of the present invention performs parameter fitting on various distribution types to determine the optimal distribution type, thereby improving the accuracy of wind speed fitting. By using the Copula structure, the method solves the dimensionality curse problem of the traditional Copula function in dealing with the dependence of multiple wind farms. The power output characteristic analysis model of multiple offshore wind farms constructed by the technical solution can better fit the typical power output characteristics of multiple offshore wind farms and improve the accuracy of power output scenario fitting.

[0132] To achieve the above embodiments, such as Figure 7 As shown, this embodiment also provides a modeling device 10 for typical power output characteristics of multiple offshore wind farms. The device 10 includes an optimal distribution type determination module 100, an edge distribution data determination module 200, an optimal correlation function determination module 300, and a power output scenario data calculation module 400.

[0133] The optimal distribution type determination module 100 is used to perform various types of parameter distribution fitting on the first wind speed data of the offshore wind farm in order to determine the optimal distribution type of the fitting based on the performance evaluation index.

[0134] The edge distribution data determination module 200 is used to convert the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generate edge distribution data of offshore wind farm output based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data;

[0135] The optimal correlation function determination module 300 is used to obtain the joint distribution function of the offshore wind farm based on the marginal distribution data and the correlation functions corresponding to different types of vine structures, and to determine the optimal correlation function based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function.

[0136] The power output scenario data calculation module 400 is used to perform inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint power output scenario data, and generate joint typical power output scenario data of offshore wind farms based on the preset clustering method and the joint power output scenario data.

[0137] The modeling device for typical power output characteristics of multiple offshore wind farms according to embodiments of the present invention performs parameter fitting on various distribution types, determines the optimal distribution type, improves the accuracy of wind speed fitting, and solves the dimensionality curse problem of multi-wind farm dependency by using the vine Copula structure. The power output characteristic analysis model of multiple offshore wind farms constructed by the technical solution can better fit the typical power output characteristics of multiple offshore wind farms and improve the accuracy of power output scenario fitting.

[0138] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0139] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A modeling method for typical power output characteristics of multiple offshore wind farms, characterized in that, Includes the following steps: Multiple types of parameter distribution fitting were performed on the first wind speed data of offshore wind farms to determine the optimal distribution type based on performance evaluation indicators; Based on the optimal distribution type, the second wind speed data is transformed to obtain wind farm output data, and the edge distribution data of offshore wind farm output is generated based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data; The joint distribution function of the offshore wind farm is obtained based on the marginal distribution data and the correlation functions corresponding to different types of vine structures. The optimal correlation function is then determined based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function. The sample data obtained by sampling the optimal correlation function is transformed by inverse function to generate joint output scenario data, and joint typical output scenario data of offshore wind farms is generated according to the preset clustering method and the joint output scenario data.

2. The method according to claim 1, characterized in that, The process of fitting various types of parameter distributions to the first wind speed data of offshore wind farms, in order to determine the optimal distribution type based on performance evaluation indicators, includes: Obtain the initial wind speed data for offshore wind farms; The second wind speed data is obtained based on the first wind speed data and the surface roughness of the actual ocean conditions. The probability density function describing the random distribution characteristics of wind speed is obtained by performing parameter estimation for each type of wind speed data obtained from the second wind speed data classification output calculation. The probability density function is fitted with performance indicators to test its performance, and a comprehensive evaluation index is obtained based on the performance test results and the performance indicators to determine the optimal distribution type.

3. The method according to claim 2, characterized in that, The process of converting the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generating edge distribution data of offshore wind farm output based on the wind farm output data, includes: Probability sampling is performed on the optimal wind speed distribution data in the optimal distribution type to obtain the third wind speed data of the offshore wind farm based on the probability sampling results. The least squares method is used to fit the power-wind speed data curve of the wind turbine, and the power data corresponding to the third wind speed data is obtained based on the curve fitting result. The power data are arranged in monotonically increasing order, and the marginal distribution data of offshore wind farm output is constructed based on the arrangement results and empirical distribution methods.

4. The method according to claim 3, characterized in that, The process of obtaining the joint distribution function of the offshore wind farm based on the edge distribution data and the correlation functions corresponding to different types of vine structures, and determining the optimal correlation function based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function, includes: A C-vine structure was established, and various types of Copula functions were selected to obtain the joint distribution function of offshore wind farms; The Euclidean distance between the fitted optimal bivariate Copula function and the empirical Copula distribution is calculated based on the joint distribution function, and the Copula function with the smallest comparison distance is taken as the initial optimal Copula function. The optimal vine structure is determined by comparing the CvM distances corresponding to the C-vine structures, and the optimal Copula function corresponding to the optimal vine structure is determined based on the initial optimal Copula function.

5. The method according to claim 4, characterized in that, When the C-vine structure is selected, a multivariate density function is constructed. The method further includes: a) Calculate the rank correlation coefficient between offshore wind farms, select the power data group with a larger average rank correlation coefficient with other wind farms as the main variable of the first-level tree, and the power data group of the remaining wind farms as the secondary variable. b) Calculate the Copula function between the main variable and the secondary variable for offshore wind farms, and calculate the Euclidean distance between the Copula function and the empirical Copula function of the sample data to determine the optimal binary Copula function; c) Sample the Copula function obtained in b) to obtain a new data set, repeat a) to determine the primary and secondary variables of the next level tree, and repeat b) to iteratively obtain the parameters of the Copula function; d) Using the Copula function parameters obtained in each level of the tree in c) as initial values, the multivariate density function is obtained by the maximum likelihood method.

6. The method according to claim 5, characterized in that, The step of generating joint output scenario data by performing an inverse function transformation on the sample data obtained from sampling the optimal correlation function includes: The sample data is generated by sampling the optimal Copula function based on the initial power output data of the offshore wind farm; The combined power output scenario data of each offshore wind farm is obtained by calculating the sample data based on the inverse functions of each distribution function of the combined distribution function.

7. The method according to claim 1, characterized in that, The step of generating joint typical power output scenario data for offshore wind farms based on a preset clustering method and the joint power output scenario data includes: The number of clusters is determined based on the preset task requirements; The k-means clustering algorithm is used to cluster the joint output scenario data to obtain the joint typical output scenario data with the number of clusters.

8. The method according to claim 1, characterized in that, Before performing various types of parameter distribution fitting on the initial wind speed data of the offshore wind farm, the method further includes: preprocessing erroneous and default data in the first wind speed data.

9. The method according to claim 1, characterized in that, The performance evaluation metrics include the Akaike Information Criterion, root mean square error, coefficient of determination, and test value.

10. A modeling device for typical power output characteristics of multiple offshore wind farms, characterized in that, include: The optimal distribution type determination module is used to fit various types of parameter distributions to the first wind speed data of offshore wind farms in order to determine the optimal distribution type of the fit based on performance evaluation indicators. The edge distribution data determination module is used to convert the second wind speed data based on the optimal distribution type to obtain wind farm output data, and generate edge distribution data of offshore wind farm output based on the wind farm output data; wherein, the second wind speed data is obtained based on the first wind speed data; The optimal correlation function determination module is used to obtain the joint distribution function of the offshore wind farm based on the marginal distribution data and the correlation functions corresponding to different types of vine structures, and to determine the optimal correlation function based on the initial optimal correlation function and the optimal vine structure obtained from the joint distribution function. The power output scenario data calculation module is used to perform inverse function transformation on the sample data obtained by sampling the optimal correlation function to generate joint power output scenario data, and generate joint typical power output scenario data of offshore wind farms according to the preset clustering method and the joint power output scenario data.

Images