Method for detecting abnormal voltage data in power grid

Abstract

The present invention discloses a method for detecting abnormal voltage data in a power grid, comprising the steps of: S1. collecting voltage data; S2. using an adaptive variant bat algorithm to process the voltage data to obtain local abnormal data of the voltage data; S3. clustering the local abnormal data of the voltage data to obtain clustered data; S4. classifying and detecting the clustered data to obtain abnormal voltage data. The method for detecting abnormal voltage data in a power grid of the present invention has better clustering effect and abnormal detection performance, and improves the detection efficiency of voltage abnormality.

Description

Technical Field

[0001] The present invention relates to the field of power grids, and in particular to a method for detecting abnormal voltage data in a power grid. Background Art

[0002] With the continuous advancement of modern power technology, the power grid is developing rapidly towards intelligence, informatization and digitization. As the amount of data to be processed by various devices in the power grid system gradually increases, the amount of power grid data is growing rapidly, which increases the difficulty of statistical analysis of the power grid. In particular, it is difficult to analyze abnormal data of the power grid, and problems such as abnormal power grid data will seriously affect the safety of the power grid.

[0003] Among them, power grid data anomalies are often reflected in voltage data anomalies, but there are currently no effective means or technologies for detecting voltage data anomalies. Summary of the invention

[0004] In view of this, the purpose of the present invention is to overcome the defects in the prior art and provide a method for detecting voltage data anomalies in a power grid, which has better clustering effect and anomaly detection performance and improves the detection efficiency of voltage anomalies.

[0005] The method for detecting abnormal voltage data in a power grid of the present invention comprises the following steps:

[0006] S1. Collect voltage data;

[0007] S2. Processing the voltage data using an adaptive mutant bat algorithm to obtain local abnormal data of the voltage data;

[0008] S3. Clustering the local abnormal data of the voltage data to obtain clustered data;

[0009] S4. Classify and detect the clustered data to obtain abnormal voltage data.

[0010] Further, the step S2 specifically includes:

[0011] S21. Initialize the frequency f in the adaptive mutation bat algorithm i =0, speed and location Among them, f i is the frequency of the i-th voltage data conversion, and are the transformation speed and position of the i-th voltage data at time t, respectively;

[0012] S22. According to the adaptive variant bat algorithm formula W(t), the voltage data is iteratively processed; the formula W(t) is:

[0013] Where W(t) is the distribution weight of abnormal data at time t; R is the number of equal regions into which the current voltage data is divided; r k is the probability of abnormal data appearing in the kth region;

[0014] S23. Repeat step S22 to iteratively correct the change speed of the i-th voltage data at time t+1 and location Abnormal values ​​that locally deviate from normal voltage data and the number of the abnormal values ​​are obtained.

[0015] Further, the frequency f of the i-th voltage data transformation of the adaptive variant bat algorithm is determined according to the following formula: i :

[0016]

[0017] Among them, F avg is the average adaptation value of all voltage data, F best is the abnormal data in the current voltage data, f min is the minimum frequency, α∈[-1,1] is a random vector;

[0018] The transformation speed of the i-th voltage data of the adaptive mutant bat algorithm at time t is determined according to the following formula:

[0019]

[0020] Where ω is the inertia weight, is the change speed of the ith voltage data at time t, x * is the abnormal data location;

[0021] The position of the i-th voltage data of the adaptive mutant bat algorithm at time t+1 is determined according to the following formula:

[0022]

[0023] Among them, δ∈[-1,1] is a random vector, A (t) is the loudness at time t, and s is the adaptive parameter.

[0024] Further, the step S3 specifically includes:

[0025] S31. Construct clustering loss model:

[0026]

[0027] Among them, F(U,S) is the clustering loss function, R is the number of equal regions into which the current voltage data is divided, and t i is the execution time of the ith voltage data, U is the membership matrix, and U={ρ j (x i )},x i is the voltage data set, the x i ={x1,x2,…,x n}, k is the number of categories, S is the Laplace coefficient, The i-th voltage data belongs to cluster center c j The weighting coefficient, c j is the jth regional cluster center, ρ j (x i ) is the j-th class membership corresponding to the i-th voltage data, b is the number of iterations, ‖x i -c j ‖ 2 is the distance between the ith voltage data and the jth regional cluster center c j The distance is the distribution entropy and average distance mapping weight of voltage data;

[0028] S32. Solve the minimum value of the clustering loss model to obtain the cluster center c j And the membership function ρ j (x i );

[0029] The cluster center c j for:

[0030] The membership function ρ j (x i )for:

[0031] S33. Iterative calculation of cluster center c j And the membership function ρ j (x i ) until the cluster loss model converges, and the abnormal voltage data corresponding to the cluster loss model after convergence is used as the clustered data.

[0032] The beneficial effects of the present invention are as follows: the present invention discloses a method for detecting voltage data anomalies in a power grid, which performs cluster analysis on voltage data collected from the power grid by utilizing a clustering method optimized by an adaptive variant bat algorithm, obtains anomaly classification of voltage data, and further realizes detection of voltage data anomalies; the present invention has better clustering effect and anomaly detection performance, and improves the detection efficiency of voltage anomalies. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0034] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0035] The present invention is further described below in conjunction with the accompanying drawings. Figure 1 As shown:

[0036] The method for detecting abnormal voltage data in a power grid of the present invention comprises the following steps:

[0037] S1. Obtaining voltage data in the power grid from various data sensors provided in the power grid system;

[0038] S2. Processing the voltage data using an adaptive mutant bat algorithm to obtain local abnormal data of the voltage data;

[0039] S3. Clustering the local abnormal data of the voltage data to obtain clustered data;

[0040] S4. Classify and detect the clustered data to obtain abnormal voltage data.

[0041] In this embodiment, the step S2 specifically includes:

[0042] S21. Initialize the frequency f in the adaptive mutation bat algorithm i =0, speed and location Among them, f i is the frequency of the i-th voltage data conversion, and are the transformation speed and position of the i-th voltage data at time t, respectively;

[0043] S22. According to the adaptive variant bat algorithm formula W(t), the voltage data is iteratively processed; the formula W(t) is:

[0044] Where W(t) is the distribution weight of abnormal data at time t. The smaller the value of W(t), the more uneven the distribution of abnormal data. When W(t) is 0, the abnormal data are concentrated in the same place. R is the number of equal regions into which the current voltage data is divided. The abnormal voltage data contained in each region is N1, N2, ..., N k ; r k is the probability of abnormal data appearing in the kth region; S is the Laplace coefficient;

[0045] S23. Repeat step S22 to iteratively correct the change speed of the i-th voltage data at time t+1 and location The abnormal value that locally deviates from the normal voltage data and the number of the abnormal values ​​are obtained; wherein the number of iterations can be set according to the actual situation.

[0046] In this embodiment, the frequency f of the i-th voltage data transformation of the adaptive variant bat algorithm is determined according to the following formula: i :

[0047]

[0048] Among them, F avg is the average adaptation value of all voltage data, F best is the abnormal data in the current voltage data, f min is the minimum frequency, α∈[-1,1] is a random vector;

[0049] The transformation speed of the i-th voltage data of the adaptive mutant bat algorithm at time t is determined according to the following formula:

[0050]

[0051] Among them, ω is the inertia weight, which can affect the local and global optimization ability of the algorithm. is the change speed of the ith voltage data at time t, x * is the abnormal data location;

[0052] The position of the i-th voltage data of the adaptive mutant bat algorithm at time t+1 is determined according to the following formula:

[0053]

[0054] Among them, δ∈[-1,1] is a random vector, A (t) is the loudness at time t, s is an adaptive parameter, and the parameter s can be adjusted according to the specific use of the adaptive variant bat algorithm.

[0055] If an optimal solution is obtained in the adaptive mutation bat algorithm, the position of each voltage data is further updated as:

[0056] x new =x old +τA (t) ;

[0057] Among them, x new represents the new position, x oldrepresents the old position, τ∈[-1,1].

[0058] In this embodiment, the step S3 specifically includes:

[0059] S31. Construct clustering loss model:

[0060]

[0061] Among them, F(U,S) is the clustering loss function, R is the number of equal regions into which the current voltage data is divided, and t i is the execution time of the ith voltage data, U is the membership matrix, and U={ρ j (x i )},x i is the voltage data set, the x i ={x1,x2,…,x n}, k is the number of categories, S is the Laplace coefficient, The i-th voltage data belongs to cluster center c j The weighting coefficient, c j is the jth regional cluster center, ρ j (x i ) is the j-th class membership corresponding to the i-th voltage data, b is the number of iterations, ‖x i -c j ‖ 2 is the distance between the ith voltage data and the jth regional cluster center c j When the distance is 0, the i-th voltage data is abnormal data. is the distribution entropy and average distance mapping weight of voltage data;

[0062] S32. Let the clustering loss function F(U,S) in the clustering loss model be j and ρ j (x i ) is 0, and the minimum value of the clustering loss model is obtained, and the cluster center c is obtained. j And the membership function ρ j (x i );

[0063] The cluster center c j for:

[0064] The membership function ρ j (x i )for:

[0065] S33. Iterative calculation of cluster center c j And the membership function ρj (x i ) until the cluster loss model converges, and the abnormal voltage data corresponding to the cluster loss model converges is used as the clustered data. The number of iterations is set according to the actual situation; if the cluster loss model fails to converge after reaching the maximum number of iterations, the process ends.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for detecting abnormal voltage data in a power grid, characterized in that: The steps include: S1. Collect voltage data; S2. Processing the voltage data using an adaptive mutant bat algorithm to obtain local abnormal data of the voltage data; The step S2 specifically includes: S21. Initialize the frequency f in the adaptive mutation bat algorithm i =0, speed and location Among them, f i is the frequency of the i-th voltage data conversion, and are the transformation speed and position of the i-th voltage data at time t, respectively; The frequency f of the i-th voltage data transformation of the adaptive mutant bat algorithm is determined according to the following formula i : Among them, F avg is the average adaptation value of all voltage data, F best is the abnormal data in the current voltage data, f min is the minimum frequency, α∈[-1, 1] is a random vector; The transformation speed of the i-th voltage data of the adaptive mutant bat algorithm at time t is determined according to the following formula: Where ω is the inertia weight, is the change speed of the ith voltage data at time t, x * is the abnormal data location; The position of the i-th voltage data of the adaptive mutant bat algorithm at time t+1 is determined according to the following formula: Among them, δ∈[-1, 1] is a random vector, A (t) is the loudness at time t, s is the adaptive parameter; S22. According to the adaptive variant bat algorithm formula W(t), the voltage data is iteratively processed; the formula W(t) is: Where W(t) is the distribution weight of abnormal data at time t; R is the number of equal regions into which the current voltage data is divided; r k is the probability of abnormal data appearing in the kth region; S23. Repeat step S22 to iteratively correct the change speed of the i-th voltage data at time t+1 and location Obtaining abnormal values ​​that locally deviate from normal voltage data and the number of the abnormal values; S3. Clustering the local abnormal data of the voltage data to obtain clustered data; The step S3 specifically includes: S31. Construct clustering loss model: Among them, F(U, S) is the clustering loss function, R is the number of equal regions into which the current voltage data is divided, and t i is the execution time of the ith voltage data, U is the membership matrix, and U={ρ j (x i )), x i is the voltage data set, the x i ={x1, x2, ..., x n }, k is the number of categories, S is the Laplace coefficient, The i-th voltage data belongs to cluster center c j The weighting coefficient, c j is the jth regional cluster center, ρ j (x i ) is the j-th class membership corresponding to the ith voltage data, b is the number of iterations, ||x i -c j || 2 is the distance between the ith voltage data and the jth regional cluster center c j The distance is the distribution entropy and average distance mapping weight of voltage data; S32. Solve the minimum value of the clustering loss model to obtain the cluster center c j And the membership function ρ j (x i ); The cluster center c j for: The membership function ρ j (x i )for: S33. Iterative calculation of cluster center c j And the membership function ρ j (x i ), until the cluster loss model converges, and the abnormal voltage data corresponding to the cluster loss model converges as the clustered data; S4. Classify and detect the clustered data to obtain abnormal voltage data.

Images