A three-phase unbalanced disturbance source modeling method and device based on monitoring data

Abstract

The present invention provides a method and device for modeling a three-phase unbalanced disturbance source based on monitoring data, the method comprising: obtaining reactive power data and three-phase unbalanced monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle; clustering based on a fuzzy C-means clustering algorithm to obtain a clustering data set corresponding to each clustering result; obtaining a probability model according to the distribution form and function type of the data set and solving it; and constructing a time series model of the three-phase unbalanced disturbance source corresponding to the preset working cycle. Compared with the prior art, clustering and modeling are performed using three-phase unbalanced data and reactive power data, and there is no need to obtain voltage and current through flow calculation to calculate the three-phase unbalanced degree, so it will not be affected by the calculation results of the flow calculation, thereby improving the accuracy and efficiency of the three-phase unbalanced degree evaluation.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular to a method and device for modeling a three-phase unbalanced disturbance source based on monitoring data. Background Art

[0002] With the transformation of power system energy, a high proportion of power electronics and a high proportion of new energy have appeared in the system, reflecting the "double high" characteristics. This has greatly increased the proportion of high-power single-phase loads and power sources, and the three-phase imbalance problem in the power system has become increasingly prominent. Under normal circumstances, the system is three-phase symmetrical, but with the unbalanced access of distributed power sources and the random use of electric vehicle charging piles, the system symmetry will be broken and three-phase imbalance problems will occur, causing additional line losses, reduced distribution transformer output, increased motor losses and other hazards. The above series of factors that cause three-phase imbalance problems are collectively referred to as three-phase imbalance disturbance sources.

[0003] At present, the calculation of three-phase imbalance is mainly based on the flow calculation of three-phase power data. Therefore, the specific method used to calculate the flow data will greatly affect the accuracy of the calculated three-phase imbalance and the modeling of the three-phase imbalance disturbance source. Summary of the invention

[0004] The present invention provides a method and device for modeling a three-phase unbalanced disturbance source based on monitoring data, so as to solve the technical problem of low accuracy of modeling a three-phase unbalanced disturbance source based on power flow calculation.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a three-phase unbalanced disturbance source modeling method based on monitoring data, comprising:

[0006] Obtain reactive power data and three-phase imbalance monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle;

[0007] Based on the fuzzy C-means clustering algorithm, the reactive power data and the three-phase unbalance monitoring data are clustered to obtain a clustering data set corresponding to each clustering result;

[0008] According to the distribution form of each cluster data set, the type of probability density function is selected and solved to obtain the probability model corresponding to each cluster data set;

[0009] According to the probability models corresponding to the cluster data sets, a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle is constructed.

[0010] As a preferred solution, the three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance; the clustering of the reactive power data and the three-phase imbalance monitoring data is specifically as follows:

[0011] Get cluster sample data X:

[0012]

[0013] Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power;

[0014] According to the clustering sample data, a clustering loss function is constructed:

[0015]

[0016] Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, and u in For sample x n The degree of membership to the i-th cluster, c i is the cluster center of the i-th cluster, k is the total number of cluster centers, ||*|| represents the distance, x n is the nth sample;

[0017] Solving by Lagrange multipliers, we obtain the following constraints:

[0018]

[0019]

[0020] Among them, J' is the Lagrangian function, λ i is the weight coefficient;

[0021] Iteratively optimizing the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges;

[0022] The value of k is determined by Dunn's coefficient to obtain the clustering result.

[0023] As a preferred solution, the membership degree and the cluster center are iteratively optimized until the fuzzy C-means clustering algorithm converges, specifically:

[0024] The membership degree and the cluster center are iteratively optimized by the following formula:

[0025]

[0026]

[0027] Until the following convergence conditions are met:

[0028]

[0029] Among them, r is the number of iteration steps, and ε is the preset error threshold.

[0030] As a preferred solution, the value of k is determined by Dunn coefficient, specifically:

[0031] The Dunn coefficient DVI of any two clusters under different k values ​​is calculated according to the following formula:

[0032]

[0033] Calculate the average value of Dunn's coefficient of any two clusters under different k values;

[0034] The k with the largest average value of Dunn's coefficient is taken as the cluster number of the three-phase unbalanced disturbance source.

[0035] As a preferred solution, according to the distribution form of each cluster data set, the type of probability density function is selected and solved respectively, specifically:

[0036] According to the distribution form of each cluster data set, a selection is made from normal distribution, lognormal distribution, exponential distribution and uniform distribution, and a solution is obtained by using the maximum likelihood estimation method.

[0037] Accordingly, an embodiment of the present invention provides a three-phase unbalanced disturbance source modeling device based on monitoring data, comprising:

[0038] A data acquisition module is used to acquire reactive power data and three-phase imbalance monitoring data of a three-phase unbalanced disturbance source to be modeled within a preset working cycle;

[0039] A clustering module, used for clustering the reactive power data and the three-phase unbalance monitoring data based on a fuzzy C-means clustering algorithm to obtain a clustering data set corresponding to each clustering result;

[0040] A solution module, used to select and solve the type of probability density function according to the distribution form of each cluster data set, so as to obtain the probability model corresponding to each cluster data set;

[0041] The model building module is used to build a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability models corresponding to each of the cluster data sets.

[0042] As a preferred solution, the three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance;

[0043] The clustering module clusters the reactive power data and the three-phase imbalance monitoring data, specifically:

[0044] The clustering module obtains clustering sample data X:

[0045]

[0046] Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power;

[0047] According to the clustering sample data, a clustering loss function is constructed:

[0048]

[0049] Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, and u in For sample x n The degree of membership to the i-th cluster, c i is the cluster center of the i-th cluster, k is the total number of cluster centers, ||*|| represents the distance, x n is the nth sample;

[0050] Solving by Lagrange multipliers, we obtain the following constraints:

[0051]

[0052]

[0053] Among them, J' is the Lagrangian function, λ i is the weight coefficient;

[0054] Iteratively optimizing the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges;

[0055] The value of k is determined by Dunn's coefficient to obtain the clustering result.

[0056] As a preferred solution, the clustering module iteratively optimizes the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges, specifically:

[0057] The clustering module iteratively optimizes the membership degree and the cluster center by the following formula:

[0058]

[0059]

[0060] Until the following convergence conditions are met:

[0061]

[0062] Among them, r is the number of iteration steps, and ε is the preset error threshold.

[0063] As a preferred solution, the clustering module determines the value of k by Dunn's coefficient, specifically:

[0064] The clustering module calculates the Dunn coefficient DVI of any two clusters under different k values ​​according to the following formula:

[0065]

[0066] Calculate the average value of Dunn's coefficient of any two clusters under different k values;

[0067] The k with the largest average value of Dunn's coefficient is taken as the cluster number of the three-phase unbalanced disturbance source.

[0068] As a preferred solution, the solution module selects the type of probability density function and solves it according to the distribution form of each cluster data set, specifically:

[0069] The solution module selects from normal distribution, lognormal distribution, exponential distribution and uniform distribution according to the distribution form of each cluster data set, and solves it through maximum likelihood estimation method.

[0070] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0071] The embodiment of the present invention provides a method and device for modeling a three-phase unbalanced disturbance source based on monitoring data, the method comprising: obtaining reactive power data and three-phase unbalanced monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle; clustering the reactive power data and the three-phase unbalanced monitoring data based on the fuzzy C-means clustering algorithm to obtain clustering data sets corresponding to each clustering result; selecting and solving the type of probability density function according to the distribution form of each clustering data set to obtain the probability model corresponding to each clustering data set; and constructing the time series model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability model corresponding to each clustering data set. Compared with the prior art, the three-phase unbalanced data and reactive power data are used for clustering and modeling, and there is no need to obtain voltage and current through power flow calculation to calculate the three-phase unbalanced degree, so it will not be affected by the calculation results of the power flow calculation, and the accuracy and efficiency of the three-phase unbalanced degree evaluation are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 : A flow chart of an embodiment of a three-phase unbalanced disturbance source modeling method provided by the present invention based on monitoring data.

[0073] Figure 2 : A structural schematic diagram of an embodiment of a three-phase unbalanced disturbance source modeling device provided by the present invention based on monitoring data. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0075] Embodiment one:

[0076] Please refer to Figure 1 , Figure 1 A three-phase unbalanced disturbance source modeling method based on monitoring data provided by an embodiment of the present invention includes steps S1 to S4, wherein:

[0077] Step S1, obtaining reactive power data and three-phase imbalance monitoring data of a three-phase unbalanced disturbance source to be modeled within a preset working cycle.

[0078] In this embodiment, a number of power quality monitoring points are set in the power system, and a number of power quality monitoring data are collected for preprocessing, or a large amount of power quality monitoring data can be directly called in the power company. The reactive power data and three-phase imbalance monitoring data of the entire working cycle of the three-phase unbalanced disturbance source to be modeled are selected. The reactive power data and three-phase imbalance monitoring data are both recorded as effective values ​​every 3 minutes, and the missing data are filled in by the mean compensation method. Among them, the three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance.

[0079] Step S2: clustering the reactive power data and the three-phase imbalance monitoring data based on a fuzzy C-means clustering algorithm to obtain a clustering data set corresponding to each clustering result.

[0080] In a complete working cycle, the power of the three-phase unbalanced disturbance source is not exactly the same as the three-phase imbalance. For example, a single-phase photovoltaic power generation system is affected by the actual light intensity and temperature, so their power size and three-phase imbalance in a day can be divided into multiple situations. Therefore, this embodiment clusters the reactive power data and three-phase imbalance monitoring data of the three-phase unbalanced disturbance source through the fuzzy C-means clustering algorithm (Fuzzy C-Means, abbreviated FCM), which can accurately divide the operation cycle of the disturbance source and obtain data sets under different clustering results.

[0081] Exemplarily, clustering the reactive power data and the three-phase imbalance monitoring data is specifically performed as follows:

[0082] Get cluster sample data X:

[0083]

[0084] Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power;

[0085] Set the initial cluster center to C = {c1,c 2, …,c k}, and construct a clustering loss function based on the clustering sample data:

[0086]

[0087] Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, which is used to adjust the influence of the membership index in the optimization target, and the preferred value is 2, u in For sample xn The membership degree of the i-th cluster, c i is the cluster center of the i-th cluster and has three dimensions, k is the total number of cluster centers, ||*|| represents the distance (which can be any distance metric such as Euclidean distance, D8 distance, etc.), x n is the nth sample and has three-dimensional features.

[0088] Solving by Lagrange multipliers, we obtain the following constraints:

[0089]

[0090]

[0091] Among them, J' is the Lagrangian function, λ i is the weight coefficient;

[0092] The membership degree and the cluster center are iteratively optimized until the fuzzy C-means clustering algorithm converges, specifically:

[0093] Since fuzzy C-means clustering is a process of continuously iteratively calculating membership and cluster centers, the membership and cluster centers are iteratively optimized by the following formula:

[0094]

[0095]

[0096] Until the following convergence conditions are met, the best cluster center is obtained:

[0097]

[0098] Among them, r is the number of iteration steps, and ε is the preset error threshold.

[0099] The value of k is determined by the Dunn coefficient to obtain the clustering result. Preferably, the Dunn coefficient DVI of any two clusters under different k values ​​is calculated according to the following formula:

[0100]

[0101] Calculate the average value of the Dunn coefficient of any two clusters under the different k values; wherein, before the calculation, k can be set to 2, 3, ..., 10, and the membership degree and cluster center under different k values ​​are repeatedly iteratively optimized until convergence, so that the clustering results corresponding to different k values ​​can be obtained. By calculating the average value of the Dunn coefficient of any two clusters under different k values, the k with the largest average value of the Dunn coefficient can be selected as the cluster number of the three-phase unbalanced disturbance source, that is, the optimal cluster number. In the implementation of the embodiment of the present application, since the three-phase unbalanced disturbance source is exposed to different light and the temperature of the environment in different environments, there may be different division methods. Taking this uncertainty into account, the FCM algorithm is improved by the Dunn coefficient to determine the k value, which can achieve accurate clustering of the monitoring data of the three-phase unbalanced disturbance source, so as to further improve the accuracy and effectiveness of modeling.

[0102] Step S3, according to the distribution form of each cluster data set, respectively select the type of probability density function and solve it to obtain the probability model corresponding to each cluster data set.

[0103] In this embodiment, the types of probability distribution functions include normal distribution, lognormal distribution, exponential distribution and uniform distribution, see Table 1:

[0104] Table 1 Types of probability distribution functions

[0105]

[0106] According to the distribution form of each cluster data set, one of the above probability distribution functions is selected. As an example of this embodiment, the selection can be made manually or automatically according to the matching degree between the distribution form of each cluster data set and each type of probability density distribution function.

[0107] Furthermore, the Matlaba command mle of the maximum likelihood estimation method is used to solve the parameters corresponding to the function type, and the solved parameters are substituted into the corresponding probability distribution function to obtain the probability model of the corresponding clustering data set.

[0108] Step S4: constructing a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability models corresponding to the clustered data sets.

[0109] Based on the analysis of the probability models corresponding to different clustering data sets, the time series model of the three-phase unbalanced disturbance source divided according to the clustering results of the three-phase unbalance degree and power size can be obtained.

[0110] Accordingly, refer to Figure 2 The embodiment of the present invention provides a three-phase unbalanced disturbance source modeling device based on monitoring data, comprising:

[0111] The data acquisition module 101 is used to acquire reactive power data and three-phase imbalance monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle;

[0112] A clustering module 102, configured to cluster the reactive power data and the three-phase unbalance monitoring data based on a fuzzy C-means clustering algorithm to obtain a clustering data set corresponding to each clustering result;

[0113] A solution module 103 is used to select and solve the type of probability density function according to the distribution form of each cluster data set, so as to obtain the probability model corresponding to each cluster data set;

[0114] The model building module 104 is used to build a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability models corresponding to each of the cluster data sets.

[0115] In this embodiment, the three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance;

[0116] The clustering module 102 clusters the reactive power data and the three-phase imbalance monitoring data, specifically:

[0117] The clustering module 102 obtains clustering sample data X:

[0118]

[0119] Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power;

[0120] According to the clustering sample data, a clustering loss function is constructed:

[0121]

[0122] Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, and u in For sample x n The degree of membership to the i-th cluster, c i is the cluster center of the i-th cluster, k is the total number of cluster centers, ||*|| represents the distance, x n is the nth sample;

[0123] Solving by Lagrange multipliers, we obtain the following constraints:

[0124]

[0125]

[0126] Among them, J' is the Lagrangian function, λ i is the weight coefficient;

[0127] Iteratively optimizing the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges;

[0128] The value of k is determined by Dunn's coefficient to obtain the clustering result.

[0129] In this embodiment, the clustering module 102 iteratively optimizes the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges, specifically:

[0130] The clustering module 102 iteratively optimizes the membership degree and the cluster center by the following formula:

[0131]

[0132]

[0133] Until the following convergence conditions are met:

[0134]

[0135] Among them, r is the number of iteration steps, and ε is the preset error threshold.

[0136] In this embodiment, the clustering module 102 determines the value of k by Dunn's coefficient, which is specifically:

[0137] The clustering module 102 calculates the Dunn coefficient DVI of any two clusters under different k values ​​according to the following formula:

[0138]

[0139] Calculate the average value of Dunn's coefficient of any two clusters under different k values;

[0140] The k with the largest average value of Dunn's coefficient is taken as the cluster number of the three-phase unbalanced disturbance source.

[0141] In this embodiment, the solving module 103 selects the type of probability density function and solves it according to the distribution form of each cluster data set, specifically:

[0142] The solution module 103 selects from normal distribution, lognormal distribution, exponential distribution and uniform distribution according to the distribution form of each cluster data set, and solves it through maximum likelihood estimation method.

[0143] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0144] The embodiment of the present invention provides a method and device for modeling a three-phase unbalanced disturbance source based on monitoring data, the method comprising: obtaining reactive power data and three-phase unbalanced monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle; clustering the reactive power data and the three-phase unbalanced monitoring data based on the fuzzy C-means clustering algorithm to obtain clustering data sets corresponding to each clustering result; selecting and solving the type of probability density function according to the distribution form of each clustering data set to obtain the probability model corresponding to each clustering data set; and constructing the time series model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability model corresponding to each clustering data set. Compared with the prior art, the three-phase unbalanced data and reactive power data are used for clustering and modeling, and there is no need to obtain voltage and current through power flow calculation to calculate the three-phase unbalanced degree, so it will not be affected by the calculation results of the power flow calculation, and the accuracy and efficiency of the three-phase unbalanced degree evaluation are improved.

[0145] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A three-phase unbalanced disturbance source modeling method based on monitoring data, characterized in that: include: Obtain reactive power data and three-phase imbalance monitoring data of the three-phase unbalanced disturbance source to be modeled within a preset working cycle; Based on the fuzzy C-means clustering algorithm, the reactive power data and the three-phase unbalance monitoring data are clustered to obtain a clustering data set corresponding to each clustering result; According to the distribution form of each cluster data set, the type of probability density function is selected and solved to obtain the probability model corresponding to each cluster data set; Constructing a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability models corresponding to each of the clustered data sets; The three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance; The clustering of the reactive power data and the three-phase imbalance monitoring data is specifically as follows: Get cluster sample data X: Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power; According to the clustering sample data, a clustering loss function is constructed: Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, and u in For sample x n The membership degree of the i-th cluster, c i is the cluster center of the i-th cluster, k is the total number of cluster centers, ||*|| represents the distance, x n is the nth sample; Solving by Lagrange multipliers, we obtain the following constraints: Among them, J' is the Lagrangian function, λ i is the weight coefficient; Iteratively optimizing the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges; The value of k is determined by Dunn's coefficient to obtain the clustering result; The iterative optimization of the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges is specifically as follows: The membership degree and the cluster center are iteratively optimized by the following formula: Until the following convergence conditions are met: Among them, r is the number of iteration steps, and ε is the preset error threshold.

2. A three-phase unbalanced disturbance source modeling method based on monitoring data as claimed in claim 1, characterized in that: The value of k is determined by Dunn's coefficient, specifically: The Dunn coefficient DVI of any two clusters under different k values ​​is calculated according to the following formula: Calculate the average value of Dunn's coefficient of any two clusters under different k values; The k with the largest average value of Dunn's coefficient is taken as the cluster number of the three-phase unbalanced disturbance source.

3. A three-phase unbalanced disturbance source modeling method based on monitoring data according to any one of claims 1 to 2, characterized in that: According to the distribution form of each cluster data set, the type of probability density function is selected and solved respectively, specifically: According to the distribution form of each cluster data set, a selection is made from normal distribution, lognormal distribution, exponential distribution and uniform distribution, and a solution is obtained by using the maximum likelihood estimation method.

4. A three-phase unbalanced disturbance source modeling device based on monitoring data, characterized in that: include: A data acquisition module is used to acquire reactive power data and three-phase imbalance monitoring data of a three-phase unbalanced disturbance source to be modeled within a preset working cycle; A clustering module, used for clustering the reactive power data and the three-phase unbalance monitoring data based on a fuzzy C-means clustering algorithm to obtain a clustering data set corresponding to each clustering result; A solution module, used to select and solve the type of probability density function according to the distribution form of each cluster data set, so as to obtain the probability model corresponding to each cluster data set; A model building module, used to build a timing model of the three-phase unbalanced disturbance source corresponding to the preset working cycle according to the probability models corresponding to each of the cluster data sets; The three-phase imbalance monitoring data includes negative sequence voltage imbalance and negative sequence current imbalance; The clustering module clusters the reactive power data and the three-phase imbalance monitoring data, specifically: The clustering module obtains clustering sample data X: Wherein, l is the lth power quality monitoring data sampling point, N is the total number of sampling points in the preset working cycle, ε U is the negative sequence voltage unbalance, ε I Negative sequence current imbalance, q is reactive power; According to the clustering sample data, a clustering loss function is constructed: Among them, J is the value function of the fuzzy C-means clustering algorithm, m is the weighting coefficient, and u in For sample x n The membership degree of the i-th cluster, c i is the cluster center of the i-th cluster, k is the total number of cluster centers, ||*|| represents the distance, x n is the nth sample; Solving by Lagrange multipliers, we obtain the following constraints: Among them, J' is the Lagrangian function, λ i is the weight coefficient; Iteratively optimizing the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges; The value of k is determined by Dunn's coefficient to obtain the clustering result; The clustering module iteratively optimizes the membership degree and the cluster center until the fuzzy C-means clustering algorithm converges, specifically: The clustering module iteratively optimizes the membership degree and the cluster center by the following formula: Until the following convergence conditions are met: Among them, r is the number of iteration steps, and ε is the preset error threshold.

5. A three-phase unbalanced disturbance source modeling device based on monitoring data as claimed in claim 4, characterized in that: The clustering module determines the value of k through the Dunn coefficient, specifically: The clustering module calculates the Dunn coefficient DVI of any two clusters under different k values ​​according to the following formula: Calculate the average value of Dunn's coefficient of any two clusters under different k values; The k with the largest average value of Dunn's coefficient is taken as the cluster number of the three-phase unbalanced disturbance source.

6. A three-phase unbalanced disturbance source modeling device based on monitoring data according to any one of claims 4 to 5, characterized in that: The solving module selects the type of probability density function and solves it according to the distribution form of each cluster data set, specifically: The solution module selects from normal distribution, lognormal distribution, exponential distribution and uniform distribution according to the distribution form of each cluster data set, and solves it through maximum likelihood estimation method.

Images