Double-fed wind power plant equivalence method and system and medium

Abstract

The invention discloses a doubly-fed wind power plant equivalence method, a system with the doubly-fed wind power plant equivalence method and a medium, and the doubly-fed wind power plant equivalence method is based on a wind power plant equivalence modeling method of principal component analysis and multi-view clustering. And the accurate clustering equivalence of the wind power plant is realized by comprehensively considering the operating characteristic difference of the wind turbine generator from multiple perspectives.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy grid connection, and in particular to a double-fed wind farm equivalent method, system and medium. Background Art

[0002] Currently, the research on equivalent modeling of wind farms mainly focuses on the determination and selection of the number of equivalent units, clustering indicators and clustering algorithms.

[0003] Large-scale wind farms are affected by factors such as operating wind speed, geographical location, fault characteristics and time lag. The operating conditions of wind farms are becoming more and more complex, and there are large differences in the operating status of each wind turbine. The nonlinear characteristics of wind power curves between units are significant. Single-machine equivalents and composite model equivalents are usually unable to accurately characterize the actual operating characteristics of large-scale wind farms, and the equivalent effect is poor. Multi-machine equivalent modeling based on the idea of ​​"coherent equivalent" extracts similar operating characteristics of wind turbines to reasonably divide the fleet, and uses multiple wind turbines to characterize the entire wind farm. It can not only reduce the complexity of the simulation model but also better simulate the overall dynamic operating characteristics of the wind farm. Therefore, it is widely used in wind farm equivalent research.

[0004] Multi-machine equivalent modeling often selects a single clustering index or multiple clustering indexes to divide wind turbines with similar operating characteristics into the same group. The clustering indexes are mainly unit model, Crowbar action, wind speed, rotor speed, electromagnetic torque, mechanical torque, pitch angle, current, voltage and output power. However, the current wind farm equivalent model has the problem of strong subjectivity in the selection of clustering indexes, and no qualitative analysis of the correlation between different clustering indexes has been conducted. Simply increasing the number of clustering indexes not only fails to improve the clustering accuracy of the wind farm, but also increases the amount of calculation of the algorithm. How to effectively select the dominant clustering index and realize accurate clustering of wind turbines is a key issue in improving the accuracy and practicality of the wind farm equivalent model.

[0005] With the extensive use of various clustering algorithms in wind farm clustering research, the clustering efficiency and accuracy of wind turbines have been improved. The clustering algorithms mainly used in current research are K-means clustering, fuzzy clustering, vector machine clustering, density clustering, and hierarchical clustering. The above algorithms mainly implement clustering of wind turbines for clustering index data at a single moment, multiple moments or time series. However, the clustering division strategy based on the operating data at a certain moment cannot guarantee the uniqueness of the clustering results of wind turbines in a certain period or different periods and the accuracy of the clustering equivalence results.

[0006] The clustering results of wind turbines using time series data are significantly related to the length of the time series data and the selection of time periods. Some time series in the wind turbine operation data cannot provide useful information for clustering. How to mine useful clustering feature information from long time series is crucial to improving the accuracy and efficiency of wind farm clustering equivalence. In addition, existing wind farm clustering equivalence studies mostly use a single or multiple clustering indicators to divide the fleet from a single perspective. However, due to the single clustering perspective, the operation data of wind turbines cannot be fully utilized, resulting in incorrect division of the units, which ultimately reduces the accuracy of the equivalence model. Summary of the invention

[0007] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a doubly-fed wind farm equivalent method, which can comprehensively consider the differences in the operating characteristics of wind turbines from multiple perspectives to achieve accurate clustering equivalent of wind farms and improve the equivalent accuracy.

[0008] The present invention also provides a system having the above-mentioned doubly-fed wind farm equivalent method.

[0009] The doubly-fed wind farm equivalent method according to the first aspect of the present invention is characterized by comprising the following steps:

[0010] Quantify the wind turbine operating variables and use principal component analysis to reduce the data dimension and determine the clustering indicators involved in subsequent operations;

[0011] According to the wind turbine clustering index, the wind turbines are divided into groups;

[0012] Equivalence calculation is performed on the divided wind turbine sets and the collector lines corresponding to the sets, and the error caused by the equivalence model is quantitatively evaluated according to the clustering index.

[0013] The doubly-fed wind farm equivalent method according to the embodiment of the present invention has at least the following beneficial effects: the doubly-fed wind farm equivalent method provided by the present invention uses the principal component analysis method to analyze the clustering index, selects the clustering index that best represents the wind turbine operation information, realizes the dimension reduction of the wind turbine clustering index, and clarifies the selection basis of the wind turbine clustering index; on this basis, a multi-perspective neighbor constraint algorithm is used to realize the group division and equivalent of the wind turbine from multiple perspectives. This equivalent method can not only realize the quantitative determination of clustering indicators, but also improve the problem of reduced equivalent accuracy caused by the incorrect division of wind turbine groups due to the use of single-perspective clustering, thereby realizing the refined equivalent of the wind farm.

[0014] According to some embodiments of the present invention, in the step of quantifying the wind turbine operating variables and performing data dimension reduction processing using principal component analysis to determine the clustering index involved in subsequent operations, information entropy is used to calculate the information content of the operating data of each unit, and the formula satisfies:

[0015]

[0016] Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability size when ; n is the segment length of the time series.

[0017] According to some embodiments of the present invention, the step of quantifying the wind turbine operating variables and performing data dimension reduction processing using principal component analysis to determine clustering indicators involved in subsequent operations includes:

[0018] Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X;

[0019] Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the variance contribution rate a of the i-th variable in the variable matrix X i and the cumulative variance contribution rate b of the first i variables i ;

[0020] The principal component factor a ji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as:

[0021] F i =a 1i X1+a 2i X2+....+a pi X p .

[0022] According to some embodiments of the present invention, the variance contribution rate a of the i-th variable in the calculated variable matrix X is i The calculation formula satisfies:

[0023]

[0024] The cumulative variance contribution rate b of the first i variables i The calculation formula satisfies:

[0025]

[0026] Among them, a i is the variance contribution rate of the i-th variable in the variable matrix X; b i is the cumulative variance contribution rate of the first i variables; p is the total number of variables in the variable matrix X; λ i is the ith eigenvalue.

[0027] According to some embodiments of the present invention, the step of dividing the wind turbines into groups according to the wind turbine clustering index includes:

[0028] The cosine distance is used to calculate the size of each element in the distance matrix under a single perspective;

[0029] The distance matrix of each single view is used as a distance matrix with a maximum weight, and a fusion distance matrix with an average weight is defined, and the distance matrix with a maximum weight and the fusion distance matrix with an average weight are combined to obtain an extreme value weight fusion distance matrix;

[0030] The Rolls maximum-minimum criterion based on the distance matrix is ​​used as the cluster clustering evaluation index to obtain the best wind turbine cluster division result.

[0031] According to some embodiments of the present invention, in the process of performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluating the error caused by the equivalent model according to the clustering index, the equivalent parameter calculation formula is as follows:

[0032]

[0033] Where: Z L_k is the impedance parameter of the kth line; Z L_eq P is the impedance parameter after line equalization; i is the active power of the i-th doubly-fed wind turbine generator set; P j is the total active power flowing through the line; N is the number of doubly-fed wind turbines on a single feeder.

[0034] According to some embodiments of the present invention, in the step of performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluating the error caused by the equivalent model according to the clustering index, the mean square error E err1 , average relative error E err2 and the maximum absolute error E err3 As an evaluation and verification indicator for the equivalent results of different equivalent models, it is defined as follows:

[0035]

[0036] Where: A represents the total number of sampling points of the simulation time series; Z (a)Represents simulation data of the equivalent model of a doubly-fed wind farm; represents the simulation data of the detailed model of the doubly-fed wind farm; a represents the sampling point.

[0037] According to the second aspect of the present invention, the doubly-fed wind farm equivalent system is characterized by comprising:

[0038] The quantification and dimension reduction module can quantify the operating variables of wind turbines, and use the principal component analysis method to perform data dimension reduction processing to determine the clustering indicators involved in subsequent operations;

[0039] A group division module, capable of dividing the wind turbines into groups according to the wind turbine clustering index;

[0040] The equivalent value evaluation module can perform equivalent calculations on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluate the errors caused by the equivalent value model according to the clustering index.

[0041] According to some embodiments of the present invention, the quantization and dimensionality reduction module includes a quantization element and a dimensionality reduction element respectively; wherein the quantization element uses information entropy to calculate the information content of the operation data of each unit, and the formula satisfies:

[0042]

[0043] Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability of time; n is the segment length of the time series;

[0044] The working process of the dimension reduction component includes:

[0045] Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X;

[0046] Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the variance contribution rate a of the i-th variable in the variable matrix X i and the cumulative variance contribution rate b of the first i variables i ;

[0047] The principal component factor a ji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as:

[0048] F i =a1i X1+a 2i X2+....+a pi X p .

[0049] According to some embodiments of the present invention, the group division module includes:

[0050] The cosine distance operation component can use the cosine distance to calculate the size of each element in the distance matrix under a single perspective;

[0051] The extreme value weight fusion distance matrix component can take the distance matrix of each single view as the distance matrix with the maximum weight, define the fusion distance matrix with the average weight, combine the distance matrix with the maximum weight and the fusion distance matrix with the average weight, and obtain the extreme value weight fusion distance matrix;

[0052] The Rolls maximum and minimum criterion evaluation component can use the Rolls maximum and minimum criterion based on the distance matrix as a cluster clustering evaluation indicator to obtain the best wind turbine cluster division result.

[0053] An embodiment of the third aspect of the present invention provides a computer-readable storage medium, wherein the medium stores computer-executable instructions, and the computer-executable instructions are used to execute the above-mentioned doubly-fed wind farm equivalent method.

[0054] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0056] Figure 1 A schematic diagram of the steps of a doubly-fed wind farm equivalent method according to an embodiment of the present invention;

[0057] Figure 2 for Figure 1 The flowchart of the multi-view neighbor-constrained hierarchical clustering algorithm provided in the doubly-fed wind farm equivalent method is shown;

[0058] Figure 3 It is a structural block diagram of a doubly-fed wind farm equivalent system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0060] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., and orientations or positional relationships indicated are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0061] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0062] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0063] In order to solve the problem that the selection of clustering indicators in the existing wind farm equivalent modeling research is highly subjective and the clustering perspective is single, resulting in reduced equivalent accuracy, this paper proposes a doubly fed wind farm equivalent method based on principal component analysis and multi-perspective clustering. Figure 1 As shown, the method comprises at least the following steps:

[0064] Step S100: quantifying clustering indicators and performing data dimension reduction processing on wind turbine operating variables.

[0065] Step S200: dividing the wind turbines into groups according to the wind turbine clustering index.

[0066] Step S300: performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and performing quantitative evaluation on the error caused by the equivalent model according to the clustering index.

[0067] Furthermore, in order to explain the doubly-fed wind farm equivalent method provided by the present application in more detail and to explain the purpose of the present application, the steps of the above method are now described in more detail.

[0068] The doubly-fed wind farm equivalent method provided in this application includes:

[0069] Step S100: quantify the wind turbine operating variables and use principal component analysis to perform data dimension reduction to determine the clustering index for subsequent operations. This can be divided into two steps: step S110: quantify the clustering index; step S120: data dimension reduction.

[0070] Step S110: quantification of clustering indexes. This includes:

[0071] In order to achieve quantitative evaluation of the information content in the time series of different variables, the concept of information entropy is used to calculate the information content of the operating data of each unit. Information entropy is often used to describe the uncertainty of the time series of different random variables. The larger the information entropy of the random variable, the greater the uncertainty of the time series of the random variable, and the more information contained in the time series. Its calculation expression is shown in formula (1):

[0072]

[0073] Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability size when ; n is the segment length of the time series.

[0074] Step S120: Data dimension reduction processing. It includes:

[0075] Step S121: Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X.

[0076] Step S122: Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the values ​​of the two indicators of formula (2) and formula (3) respectively:

[0077]

[0078] Where: a i is the variance contribution rate of the i-th variable in the variable matrix X; b i is the cumulative variance contribution rate of the first i variables; p is the total number of variables in the variable matrix X; λ i is the i-th eigenvalue. i The larger the value, the more data information the i-th variable contains. i If it reaches more than 90%, it is considered that the first i variables after dimensionality reduction can represent important information of the original variable data.

[0079] Step S123: Use the principal component factor aji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as:

[0080] F i =a 1i X1+a 2i X2+....+a pi X p (4)

[0081] Step S200: dividing the wind turbines into groups according to the wind turbine clustering index.

[0082] The multi-view neighbor constraint hierarchical clustering algorithm introduces the nearest neighbor constraint and multi-view theory into the traditional hierarchical clustering algorithm to realize the multi-view clustering division of wind farms. When dealing with the multi-view clustering problem, the traditional multi-view clustering algorithm usually fuses the multi-view data by connecting all the features of multiple views or calculating the average similarity matrix. The former increases the dimension of the matrix after data fusion, and the latter cannot represent the best consensus information when there are differences in the importance of each view, which may eventually lead to the clustering result under a single view being better than the clustering result under multiple views. To solve the above problems, the Fusion Distance Matrix with Extreme Weights (FDEW) is used to represent the complementary information between multiple views and the single view information, thereby reducing the impact of the contradiction of the difference in the importance of multiple views on the clustering results of the doubly fed wind turbine.

[0083] First, the commonly used cosine distance is used to calculate the size of each element in the distance matrix under a single perspective. The calculation formula can be expressed as:

[0084]

[0085] Where: A (i) , B (i) is the cosine distance between any two samples in the i-th view; T represents the transpose of the matrix; d(A (i) , B (i) )∈(0,2).

[0086] The distance matrix of each single view is regarded as a distance matrix with extremely large weights, and a fusion distance matrix with average weights is defined. The two are combined to obtain an extreme weight fusion distance matrix. The specific calculation formulas of the three distance matrices can be expressed as formulas (6) to (8):

[0087]

[0088] Where: v is the number of views; D (i) is the distance matrix of the i-th view; D * is the distance matrix of average weight; {FDEW} is the extreme weight fusion distance matrix.

[0089] Hierarchical clustering is one of the commonly used clustering algorithms. There is no need to specify the final number of clusters during clustering. However, after merging class objects during the clustering process, the distance matrix between classes needs to be recalculated. For large wind farms, the computational complexity of clustering is high due to the large amount of data, and a chain effect is easily generated during the clustering process. Hierarchical clustering with the introduction of nearest neighbor constraints can greatly reduce the complexity of the clustering calculation process, thereby obtaining better wind turbine clustering performance. In order to objectively evaluate the quality of the wind turbine clustering results in a doubly fed wind farm and obtain the best wind turbine group division results, this chapter uses the Rawls maximum and minimum criterion based on the distance matrix as the cluster clustering evaluation indicator.

[0090] First, select any two clusters from the multi-view clustering results, select any sample from the two clusters, and define the cosine distance between the two samples as the inter-cluster distance; arbitrarily select the average cosine distance of two samples in a clustering result as the intra-cluster distance. Define the initial Rawls maximum and minimum criterion clustering evaluation index as:

[0091]

[0092] Where: a (i) , b (i) are any sample values ​​of cluster m and cluster n under the i-th perspective; c (i) , e (i) are the two sample values ​​of the kth cluster under the ith perspective; d(a (i) , b (i) ) is the distance between clusters; d(c (i) , e (i) ) is the intra-class distance; ζ k (i) is the kth cluster under the i-th view.

[0093] From formula (9), we can see that in order to obtain the best clustering result of doubly-fed wind turbines, the evaluation index F should be (i) The calculated result is the optimal value, but since the distance between clusters and the distance within the cluster are unknown, and F (i) The calculation results show nonlinear changes, resulting in the optimal F (i) It cannot be determined. According to Rawls's maximum-minimum criterion, the correct decision should maximize the minimum result, so the minimum value of formula (9) can be written as:

[0094]

[0095] According to Rawls's maximum and minimum criteria, the optimal evaluation index of formula (10) can be written as:

[0096]

[0097] Finally, the information entropy of the dominant variable of the clustering index is obtained by using the calculation process of the clustering index information entropy in the first step and the processing flow of the principal component analysis. Formula (11) is used as the evaluation criterion for the optimal result of the multi-view neighbor constraint hierarchical clustering to determine the final division result of the wind turbine group in the doubly fed wind farm, and the equivalent parameter calculation of the doubly fed wind farm is performed. The specific algorithm flow of the multi-view neighbor constraint hierarchical clustering algorithm is as follows: Figure 2 As shown in the figure, L is the number of clustering indicators after dimensionality reduction; n is the number of fault conditions; and m is the number of wind turbines in the doubly fed wind farm.

[0098] Step S300: performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and performing quantitative evaluation on the error caused by the equivalent model according to the clustering index.

[0099] Assuming that there are w doubly-fed wind turbines in the doubly-fed wind farm, which are divided into a group and are equivalent to one doubly-fed wind turbine, the parameters of the generator after the equivalent value can be written as:

[0100]

[0101] Where: S is the capacity of a single doubly-fed wind turbine; P, Q are the active power and reactive power generated by each doubly-fed wind turbine; x m is the generator excitation reactance; x s 、x r are the stator reactance and rotor reactance of the generator respectively; R s , R r are the stator resistance and rotor resistance of the generator respectively; the subscript eq represents the equivalent parameters of the generator. The equivalent parameters of the shaft system of the wind turbine are:

[0102]

[0103] Where: H g , H t are the inertia time constants of the generator and wind turbine shaft systems, respectively; K and D are the shaft system stiffness coefficient and damping coefficient, respectively.

[0104] The equivalent parameters of the transformer in each group are:

[0105]

[0106] Where: S T To divide the capacity of each transformer in the group; Z T is the impedance of each transformer.

[0107] Considering the difference between the terminal voltage and the grid connection point voltage of the doubly-fed wind turbine in each group, the equivalent parameters of the collector line are calculated based on the basic principle that the voltage loss of the collector line before and after the equivalent is consistent. The specific derivation process is taken from the existing technology. The calculation formula of the equivalent parameters of the collector line is shown in formula (15):

[0108]

[0109] Where: Z L_k is the impedance parameter of the kth line; Z L_eq P is the impedance parameter after line equalization; i is the active power of the i-th doubly-fed wind turbine generator set; P j is the total active power flowing through the line; N is the number of doubly-fed wind turbines on a single feeder.

[0110] Wind turbine equivalent usually takes the same total wind energy input before and after equivalent as the basic principle to calculate the equivalent wind speed. For the doubly fed wind turbine with no pitch angle action, the doubly fed wind turbine is running in the maximum power tracking stage. The equivalent wind speed of w doubly fed wind turbines in the same group is V eq_1 , the equivalent wind speed is shown in formula (16):

[0111]

[0112] Where: f is the wind power curve function expression of the maximum power tracking of the doubly fed wind turbine; f -1 It is the inverse function expression of wind power curve for maximum power tracking of doubly-fed wind turbine.

[0113] For the doubly-fed wind turbine with pitch angle action, the output active power is constant, and the doubly-fed wind turbine operates at the maximum power stage. The equivalent wind speed V of this type of doubly-fed wind turbine is eq_2 As shown in formula (17):

[0114]

[0115] In order to quantitatively evaluate the error results of different equivalent models, the simulation results of the detailed model are used as the benchmark and the mean square error (E err1 ), mean relative error (E err2 ) and the maximum absolute error (E err3 ) is used as the evaluation and verification index of the equivalent results of different equivalent models. The definitions of the three evaluation indexes are shown in formulas (18) to (20):

[0116]

[0117] Where: A represents the total number of sampling points of the simulation time series; Z (a)Represents simulation data of the equivalent model of a doubly-fed wind farm; represents the simulation data of the detailed model of the doubly-fed wind farm; a represents the sampling point.

[0118] Another embodiment of the present application provides a doubly-fed wind farm equivalent system, such as Figure 3 As shown, the system 30 includes:

[0119] The quantification and dimension reduction module 301 can quantify the clustering index and perform data dimension reduction processing on the wind turbine operating variables;

[0120] A group division module 302 can divide the wind turbines into groups according to the wind turbine clustering index;

[0121] The equivalent evaluation module 303 can perform equivalent calculations on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluate the errors caused by the equivalent model according to the clustering index.

[0122] Furthermore, the quantization dimension reduction module 301 includes a quantization element and a dimension reduction element. The quantization element uses information entropy to calculate the information content of the operation data of each unit, and the formula satisfies:

[0123]

[0124] Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability of time; n is the segment length of the time series;

[0125] The working process of the dimension reduction component includes:

[0126] Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X;

[0127] Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the variance contribution rate a of the i-th variable in the variable matrix X i and the cumulative variance contribution rate b of the first i variables i ;

[0128] The principal component factor a ji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as:

[0129] F i =a1i X1+a 2i X2+....+a pi X p .

[0130] Furthermore, the group division module 302 includes:

[0131] The cosine distance operation component can use the cosine distance to calculate the size of each element in the distance matrix under a single perspective;

[0132] The extreme value weight fusion distance matrix component can take the distance matrix of each single view as the distance matrix with the maximum weight, define the fusion distance matrix with the average weight, combine the distance matrix with the maximum weight and the fusion distance matrix with the average weight, and obtain the extreme value weight fusion distance matrix;

[0133] The Rolls maximum and minimum criterion evaluation component can use the Rolls maximum and minimum criterion based on the distance matrix as a cluster clustering evaluation indicator to obtain the best wind turbine cluster division result.

[0134] Another aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions for executing the above-mentioned Figure 1 The doubly-fed wind farm equivalent method is shown.

[0135] The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0137] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present application. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A doubly-fed wind farm equivalent method, It is characterized in that The following steps are involved: Quantify the wind turbine operating variables and use principal component analysis to reduce the data dimension and determine the clustering indicators involved in subsequent operations; According to the wind turbine clustering index, the wind turbines are divided into groups; Equivalence calculation is performed on the divided wind turbine sets and the collector lines corresponding to the sets, and the error caused by the equivalence model is quantitatively evaluated according to the clustering index.

2. The method according to claim 1, It is characterized in that In the step of quantifying the wind turbine operating variables and performing data dimension reduction processing using the principal component analysis method to determine the clustering index involved in the subsequent operation, the information entropy is used to calculate the information content of the operating data of each unit, and the formula satisfies: Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability size when ; n is the segment length of the time series.

3. The method according to claim 1, It is characterized in that The step of quantifying the wind turbine operating variables and performing data dimension reduction processing using principal component analysis to determine clustering indicators involved in subsequent operations includes: Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X; Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the variance contribution rate a of the i-th variable in the variable matrix X i and the cumulative variance contribution rate b of the first i variables i ; The principal component factor a ji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as: F i =a 1i X 1 +a 2i X 2 +....+a pi X p 。 4. The method according to claim 3, It is characterized in that The variance contribution rate a of the i-th variable in the calculated variable matrix X is i The calculation formula satisfies: The cumulative variance contribution rate b of the first i variables i The calculation formula satisfies: Among them, a i is the variance contribution rate of the i-th variable in the variable matrix X; b i is the cumulative variance contribution rate of the first i variables; p is the total number of variables in the variable matrix X; λ i is the ith eigenvalue.

5. The method according to claim 1, It is characterized in that The step of dividing the wind turbines into groups according to the wind turbine clustering index comprises: The cosine distance is used to calculate the size of each element in the distance matrix under a single perspective; The distance matrix of each single view is used as a distance matrix with a maximum weight, and a fusion distance matrix with an average weight is defined, and the distance matrix with a maximum weight and the fusion distance matrix with an average weight are combined to obtain an extreme value weight fusion distance matrix; The Rolls maximum-minimum criterion based on the distance matrix is ​​used as the cluster clustering evaluation index to obtain the best wind turbine cluster division result.

6. The method according to claim 1, It is characterized in that In the process of performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluating the error caused by the equivalent model according to the clustering index, the equivalent parameter calculation formula is as follows: Where: Z L_k is the impedance parameter of the kth line; Z L_eq P is the impedance parameter after line equalization; i is the active power of the i-th doubly-fed wind turbine generator set; P j is the total active power flowing through the line; N is the number of doubly-fed wind turbines on a single feeder.

7. The method according to claim 1, It is characterized in that In the step of performing equivalent calculation on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluating the error caused by the equivalent model according to the clustering index, the mean square error E is used. err1 , average relative error E err2 and the maximum absolute error E err3 As an evaluation and verification indicator for the equivalent results of different equivalent models, it is defined as follows: Where: A represents the total number of sampling points of the simulation time series; Z (a) Represents simulation data of the equivalent model of a doubly-fed wind farm; represents the simulation data of the detailed model of the doubly-fed wind farm; a represents the sampling point.

8. A doubly-fed wind farm equivalent system, It is characterized in that include: The quantification and dimension reduction module can quantify the operating variables of wind turbines, and use the principal component analysis method to perform data dimension reduction processing to determine the clustering indicators involved in subsequent operations; A group division module, capable of dividing the wind turbines into groups according to the wind turbine clustering index; The equivalent value evaluation module can perform equivalent calculations on the divided wind turbines and the collector lines corresponding to the turbines, and quantitatively evaluate the errors caused by the equivalent value model according to the clustering index.

9. The system according to claim 1, It is characterized in that The quantization and dimension reduction module includes a quantization element and a dimension reduction element. The quantization element uses information entropy to calculate the information content of the operation data of each unit, and the formula satisfies: Where: M represents a discrete random variable; p(x i ) means variable X is equal to x i The probability of time; n is the segment length of the time series; The working process of the dimension reduction component includes: Arrange the original data variables in order to form a matrix X, perform data standardization on the matrix X, calculate the correlation coefficient matrix of the standardized matrix X, and find the eigenvalues ​​and eigenvectors of the variable matrix X; Sort the eigenvalues ​​of the variable matrix X in descending order, and calculate the variance contribution rate a of the i-th variable in the variable matrix X i and the cumulative variance contribution rate b of the first i variables i ; The principal component factor a ji To characterize the degree of association between each principal component and the original variable, the principal component F j With variable X j The correlation coefficient between The principal component expression can be written as: F i =a 1i X 1 +a 2i X 2 +....+a pi X p 。 10. The system according to claim 9, It is characterized in that The group division module comprises: The cosine distance operation component can use the cosine distance to calculate the size of each element in the distance matrix under a single perspective; The extreme value weight fusion distance matrix component can take the distance matrix of each single view as the distance matrix with the maximum weight, define the fusion distance matrix with the average weight, combine the distance matrix with the maximum weight and the fusion distance matrix with the average weight, and obtain the extreme value weight fusion distance matrix; The Rolls maximum and minimum criterion evaluation component can use the Rolls maximum and minimum criterion based on the distance matrix as a cluster clustering evaluation indicator to obtain the best wind turbine cluster division result.

11. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the method according to any one of claims 1 to 7.

Images