An unsupervised dimension reduction voltage transformer secondary circuit anomaly online monitoring method

Abstract

The application provides a kind of unsupervised dimension reduction voltage transformer secondary circuit exception online monitoring method, real-time acquisition time series voltage data of voltage transformer, according to time window segmentation data, obtain time domain, frequency domain high-dimensional data sample set in each time window, can take into account the time domain characteristics of data and the frequency domain characteristics under abnormal conditions, use the construction time domain similarity matrix and frequency domain similarity matrix, extract the feature vector of similarity matrix and reduce dimension, finally use the method of unsupervised mixture Gaussian model to distinguish normal and abnormal time data, using KL divergence can accurately distinguish normal and abnormal transformer secondary circuit, realize more accurate abnormal secondary circuit positioning, further guarantee the normal operation of power system.

Description

Technical Field

[0001] The present invention relates to the field of power system transformer monitoring, and more particularly to an unsupervised dimension reduction-based online monitoring method for abnormalities in a voltage transformer secondary circuit. Background Art

[0002] Gateway energy metering devices, consisting of a high-voltage voltage transformer, secondary circuit, and energy meter, are the core equipment for energy measurement. Their long-term stable operation directly impacts the stability and accuracy of energy metering data. Furthermore, as the sole data source for trade settlements, transactions, and economic and technical indicator assessments between power generation companies, power grid companies, and electricity users, the reliability of their measurement results directly impacts the fairness of energy trade. Therefore, research on the metering performance and operating status of gateway energy metering devices holds great potential.

[0003] Over the long term, the performance of gateway energy metering devices is susceptible to a variety of factors, leading to a gradual decline in performance. Due to the complex structure of high-voltage transformers and energy meters, and their susceptibility to external interference, metering equipment can gradually lose accuracy, leading to deviations in energy calculations. This can impact the fairness of energy trade and even the steady development of the electricity market. For this reason, the State Administration for Market Regulation has included gateway instrument transformers and energy meters in the "Catalogue of Measuring Instruments Subject to Mandatory Management," which includes them in the mandatory inspection catalogue.

[0004] In recent years, with the continuous advancement of online monitoring methods for gateway transformers, online monitoring equipment installed on the secondary side has become a critical component of gateway energy metering at substations. However, the transformer and online monitoring device are connected via a secondary circuit, and the monitoring data contains error information about the secondary circuit. When anomalies such as loose terminals occur in the secondary circuit, the online monitoring device may miss or misjudge them, seriously affecting the assessment of the voltage transformer's operating status. Therefore, how to use collected data to accurately determine whether secondary circuit anomalies exist has become a pressing issue. Solving this problem is crucial for improving the reliability and accuracy of gateway energy metering devices. Summary of the Invention

[0005] The present invention addresses the problem of inaccurate monitoring of abnormal voltage transformer secondary circuits in the prior art and provides an unsupervised dimension reduction online monitoring method for abnormal voltage transformer secondary circuits, comprising:

[0006] Step 1: Collect data based on voltage transformer online monitoring device The voltage transformers are used to monitor the time series voltage data online, wherein each voltage transformer group includes three voltage transformers, and the time series voltage data is divided into Different time windows are obtained to obtain the secondary circuit time domain high-dimensional data sample set under each time window , perform Fourier transform on the time domain high-dimensional data sample set to obtain a frequency domain high-dimensional data sample set

[0007] Step 2: respectively analyze the time domain high-dimensional data sample set and the frequency domain high-dimensional data sample set Perform similarity analysis to obtain the time domain similarity matrix under each time window and frequency domain similarity matrix

[0008] Step 3: respectively calculate the time domain similarity matrix and frequency domain similarity matrix Perform matrix decomposition to obtain the time domain eigenvector and frequency domain eigenvectors The time domain feature vector and frequency domain eigenvectors Input into the dimensionality reduction algorithm to obtain the 1-dimensional time domain eigenvalues ​​under each time window and 1D frequency domain eigenvalues After performing maximum and minimum normalization on the 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue, the two-dimensional eigenvector is obtained. is the 1-dimensional time domain eigenvalue after normalization, is the 1-dimensional frequency domain eigenvalue after normalization

[0009] Step 4: convert the two-dimensional feature vector Input into a mixture Gaussian model to identify the abnormal occurrence time window of the secondary circuit, wherein the mixture Gaussian model is used as an unsupervised classification model;

[0010] Step 5: Calculate the abnormality occurrence time window and its surroundings The time domain differential eigenvector of each voltage transformer under a normal time window is , g , take 2 The mean of the time domain differential characteristic vectors of all voltage transformers in a normal time window is used to obtain the benchmark differential characteristic mean vector , calculate the probability distribution of the time domain difference eigenvector of each voltage transformer in the abnormal time window Probability distribution of the mean vector of the differential feature with the benchmark The KL divergence between them is used to identify abnormal voltage transformer secondary circuits based on the KL divergence.

[0011] The application provides a voltage transformer secondary circuit abnormality online monitoring method based on unsupervised dimension reduction, which can consider time domain characteristics of data and wide and high frequency characteristics under abnormal conditions through mixed feature analysis of time and frequency domains, uses a similarity matrix to extract a feature vector and reduce dimensions, and finally uses an unsupervised Gaussian mixture model to distinguish normal and abnormal time data. The KL divergence can be used to accurately distinguish normal and abnormal transformers and realize more accurate abnormal secondary circuit positioning. The monitoring result obtained by the method has the characteristics of rapidness, accuracy and strong interpretability. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A flowchart of the voltage transformer secondary circuit abnormality online monitoring method based on unsupervised dimension reduction provided by the application is provided.

[0013] Figure 2 An abnormal data monitoring result graph provided by the embodiment of the application is provided.

[0014] Figure 3 A KL divergence calculation flowchart provided by the embodiment of the application is provided.

[0015] Figure 4-1 And Figure 4-2 An abnormal transformer KL divergence analysis result graph provided by the embodiment of the application is provided. DETAILED DESCRIPTION

[0016] To make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application. In addition, the technical features in each embodiment or single embodiment provided by the application can be combined with each other to form a feasible technical scheme, and the combination is not restricted by the order of steps and / or structure mode, but should be based on the realization of those skilled in the art. When the combination of technical schemes appears contradictory or unfeasible, it should be considered that the combination of technical schemes does not exist and is not within the protection scope of the application.

[0017] At present, although the voltage transformer online monitoring device can evaluate the running state of the voltage transformer in real time, there are still some deficiencies. One of the key problems is how to effectively monitor the abnormal situation of the secondary circuit of the voltage transformer to realize more accurate monitoring of the running state of the voltage transformer. In order to solve this problem, the present application proposes an innovative method, that is, an online monitoring method for abnormal data of the secondary circuit of the voltage transformer based on dimension reduction feature extraction and Gaussian mixture distribution. The method can accurately identify the time and position of the abnormal data in the secondary circuit, so as to realize more comprehensive and reliable evaluation of the performance of the voltage transformer. The method can improve the practicability and reliability of the voltage transformer online monitoring device to a certain extent, and provide strong guarantee for the stable operation of the power system.

[0018] The present application provides an unsupervised dimension reduction method for online monitoring of voltage transformer secondary circuit abnormalities, and a flow chart is shown as Figure 1 The abnormal monitoring method comprises the following steps:

[0019] Step 1, based on the voltage transformer online monitoring device, a group of voltage transformer online monitoring time series voltage data is collected , wherein each group of voltage transformers includes three voltage transformers, the time series voltage data is divided into different time windows, and the secondary circuit time domain high-dimensional data sample set under each time window is obtained , and the time domain high-dimensional data sample set is subjected to Fourier transform to obtain a frequency domain high-dimensional data sample set .

[0020] It can be understood that the voltage transformer online monitoring device is a multifunctional monitoring device, and the data acquisition unit thereof is connected with the voltage transformer through the secondary circuit to collect various key parameters of the voltage transformer in real time, including the time stamp of data acquisition, the output voltage value of the secondary side of the voltage transformer.

[0021] The voltage transformer online monitoring device collects time series voltage data with a size of output by a group of transformers (each group of transformers is composed of ABC three transformers) at intervals of the same time period . With a fixed time period as the time window length, all data in the time period are extracted from the set starting , and the time domain high-dimensional data sample set with a size of is obtained by completing the segmentation based on the time window , wherein . .

[0022] ​The Fourier transform is performed on the data in the time-domain high-dimensional data sample set in each time window, and the transform formula is:

[0023]

[0024] wherein is an imaginary unit, is a time index, is a frequency index, is a sample number.

[0025] Then, the frequency-domain voltage amplitude data is extracted, and the extraction formula is as follows:

[0026]

[0027] wherein, is the frequency-domain voltage amplitude data in the frequency-domain high-dimensional data sample set, is the frequency-domain voltage data in the form of a complex number obtained after the Fourier transform, is a frequency index, is the i-th frequency, represents a real part of the complex number, represents an imaginary part of the complex number.

[0028] Step 2, similarity analysis is respectively performed on the time-domain high-dimensional data sample set and the frequency-domain high-dimensional data sample set , to obtain a time-domain similarity matrix and a frequency-domain similarity matrix under each time window.

[0029] It can be understood that, according to the time-domain high-dimensional data sample set and the frequency-domain high-dimensional data sample set under each time window obtained in step 1, cosine similarity analysis is respectively performed on the time-domain high-dimensional data sample set and the frequency-domain high-dimensional data sample set . Specifically, the cosine similarity between each two columns of data in the time-domain high-dimensional data sample set under each time window is calculated, to obtain a time-domain similarity matrix under each time window; and the cosine similarity between each two columns of data in the frequency-domain high-dimensional data sample set under each time window is calculated, to obtain a frequency-domain similarity matrix under each time window, wherein the size of the time-domain similarity matrix and the frequency-domain similarity matrix under each time window is​​​ .

[0030] wherein the cosine similarity is calculated by the formula:

[0031]

[0032] wherein denotes the dot product of two vectors, and denote the modulus of the vector and the vector , respectively, is the angle between the vector and the vector .

[0033] Step 3, respectively, the time domain similarity matrix and the frequency domain similarity matrix are decomposed to obtain the time domain eigenvector and the frequency domain eigenvector The time domain eigenvector and the frequency domain eigenvector are input into the dimension reduction algorithm to obtain the 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue After maximum and minimum normalization processing of the 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue, respectively, a two-dimensional eigenvector is obtained, wherein is the normalized 1-dimensional time domain eigenvalue, is the normalized 1-dimensional frequency domain eigenvalue.

[0034] It can be understood that the time domain similarity matrix and the frequency domain similarity matrix under each time window are respectively decomposed to obtain the time domain eigenvector and the frequency domain eigenvector .

[0035] wherein the time domain similarity matrix and the frequency domain similarity matrix under each time window are respectively decomposed to obtain the eigenvector, which specifically includes:

[0036] The time domain similarity matrix and the frequency domain similarity matrix are respectively decomposed to obtain the eigenvector composed of the eigenvalue, which has a size of , i.e. and , that is, the characteristic equation is solved: ​

[0037]

[0038] wherein is a similarity matrix, is an eigenvector composed of eigenvalues, is an identity matrix of the same size as the similarity matrix.

[0039] extracting principal component analysis (PCA), mean, kurtosis or median of the time-domain eigenvector under each time window to obtain a 1-dimensional time-domain eigenvalue under each time window; and extracting principal component analysis, mean, kurtosis or median of the frequency-domain eigenvector under each time window to obtain a 1-dimensional frequency-domain eigenvalue under each time window. .

[0040] performing maximum-minimum normalization on the 1-dimensional time-domain eigenvalue and the 1-dimensional frequency-domain eigenvalue under each time window, with the formula being:

[0041]

[0042] wherein, is the original data, and are the minimum value and the maximum value in the 1-dimensional time-domain eigenvalue or the 1-dimensional frequency-domain eigenvalue under all time windows, is the 1-dimensional time-domain eigenvalue and the 1-dimensional frequency-domain eigenvalue after maximum-minimum normalization.

[0043] taking the normalized 1-dimensional time-domain eigenvalue as the abscissa and the normalized 1-dimensional frequency-domain eigenvalue data as the ordinate to obtain a two-dimensional eigenvector .

[0044] Step 4: inputting the two-dimensional eigenvector into a mixture Gaussian model to identify the abnormal occurrence time window of the secondary circuit, the mixture Gaussian model being used as an unsupervised classification model to classify and identify whether the two-dimensional eigenvector is abnormal.

[0045] It can be understood that inputting the two-dimensional eigenvector into the mixture Gaussian model, the mixture Gaussian model can obtain a Gaussian distribution center and calculate the distance between the two-dimensional eigenvector and the Gaussian distribution center.

[0046] inputting the two-dimensional eigenvector ​Input to the mixture Gaussian model, the mixture Gaussian model outputs a two-dimensional feature vector The covariance matrix of , standard deviation , based on 3 Criteria threshold =3 .

[0047] The two-dimensional feature vector Distance from the center of the Gaussian distribution With threshold To compare; if > , then the two-dimensional eigenvector The time window where the anomaly occurs is the time window where the anomaly occurs; otherwise, the two-dimensional feature vector The time window where the abnormality occurs is the normal time window. The Gaussian mixture model can be used to identify the abnormal time window of the secondary circuit from all time windows.

[0048] Among them, the mixed Gaussian model includes the probability density function composed of multiple Gaussian distributions, and the probability density function of each Gaussian distribution is expressed as:

[0049]

[0050] in, is the probability density function, representing the two-dimensional feature vector The probability distribution of is the number of Gaussian distributions, For the The weight coefficient of the Gaussian distribution, For the The density function of a Gaussian distribution, For the The mean of a Gaussian distribution, For the The covariance matrix of a Gaussian distribution.

[0051] Step 5: Calculate the abnormality occurrence time window and its surroundings The time domain differential eigenvector of each voltage transformer under a normal time window is , g , take 2 The mean of the time domain differential characteristic vectors of all voltage transformers in a normal time window is used to obtain the benchmark differential characteristic mean vector , calculate the probability distribution of the time domain difference eigenvector of each voltage transformer in the abnormal time window Probability distribution of the mean vector of the differential feature with the benchmark KL divergence between the two, identifying the abnormal secondary circuit of the voltage transformer based on the KL divergence.

[0052] It can be understood that the abnormal time window can be identified by step 4, and the time domain high-dimensional sample data of each voltage transformer in the abnormal time window and the normal time window before and after the abnormal time window are obtained. The time domain high-dimensional sample data of each voltage transformer in each time window is subjected to difference processing. The difference data of the two data is obtained by subtracting the former data from the latter data in the time domain high-dimensional sample data of each voltage transformer in each time window. The time domain difference feature vector of each transformer is of size , The number of time series data in each time window.

[0053] The time domain difference feature vectors of each voltage transformer in all time windows are normalized. The maximum and minimum values in all data of the reference difference feature mean vector and the time domain difference feature vector of each voltage transformer in the abnormal time window are found. The reference difference feature mean vector and the time domain difference feature vector of each voltage transformer are normalized to obtain the normalized reference difference feature mean vector and the time domain difference feature vector of each voltage transformer.

[0054] The probability distribution of the normalized reference difference feature mean vector and the time domain difference feature vector of each voltage transformer is calculated. is divided into left-closed right-open intervals, and and are composed of +2) data intervals. The normalized reference difference feature mean vector and the time domain difference feature vector of each voltage transformer are divided into data intervals.

[0055] The number of data in each interval is divided by the total number of data in all intervals to obtain the probability distribution of the time domain difference feature vector of each voltage transformer in the abnormal time window and the probability distribution of the reference difference feature mean vector .

[0056] The probability distribution of the time domain difference feature vector of each voltage transformer in the abnormal time window and the probability distribution of the reference difference feature mean vector KL divergence between the two, including:

[0057]

[0058] wherein, represents the time-domain differential feature vector probability distribution of each voltage transformer under the abnormal occurrence time window, represents the reference differential feature mean vector probability distribution, represents the corresponding position of the calculation element in the distribution, is and the KL divergence between the two.

[0059] The KL divergence between the time-domain differential feature vector probability distribution of each voltage transformer under the abnormal occurrence time window and the reference differential feature mean vector probability distribution is compared with the threshold value When the KL divergence exceeds the threshold value , the corresponding voltage transformer is an abnormal voltage transformer secondary circuit, so the abnormal occurrence time window and the voltage transformer secondary circuit where the abnormality occurs are found, and the abnormal time and location of the abnormal voltage transformer are located.

[0060] The method provided by the application will be described below with a specific embodiment.

[0061] A group of three-phase continuous three-month voltage amplitude data collected by an online monitoring device of a certain transformer substation can be obtained, and the amplitude data matrix is :

[0062] ;

[0063] ;

[0064] ;

[0065] ;

[0066] Each data point is data collected once every 1 minute, and 1 hour is taken as the time window, so there are 60 rows of data under each time window, and the time-domain high-dimensional data sample set is obtained. Fourier decomposition is performed on the time-domain sample data set to obtain the frequency-domain high-dimensional data sample set .

[0067] The cosine similarity analysis is performed on the time-domain high-dimensional data sample set and the frequency-domain high-dimensional data sample set , respectively, and a time-domain similarity matrix of size is obtained under each time window. and the frequency domain similarity matrix . The time domain similarity matrix and the frequency domain similarity matrix are analyzed to obtain the eigenvectors of the time domain similarity matrix and the eigenvectors of the frequency domain similarity matrix , respectively referred to as the time domain eigenvector and the frequency domain eigenvector.

[0068] The time domain eigenvector and the frequency domain eigenvector are respectively input into the PCA algorithm in the dimension reduction method, and n_components=1 is set, that is, the principal component with the highest ranking is obtained, so that each eigenvector corresponds to a 1-dimensional eigenvalue, and a 1-dimensional time domain eigenvalue and a 1-dimensional frequency domain eigenvalue are obtained.

[0069] The 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue are subjected to maximum and minimum normalization processing to obtain normalized 1-dimensional time domain eigenvalues and 1-dimensional frequency domain eigenvalues in the range of 0-1. The normalized 1-dimensional time domain eigenvalues are taken as the axis, and the normalized 1-dimensional frequency domain eigenvalues are taken as the axis to obtain a two-dimensional eigenvector of the current time window.

[0070] An unsupervised classification model of the two-dimensional eigenvector is established using a Gaussian mixture model, and the number of Gaussian distributions n_components=2 is set, that is, two Gaussian distributions are used to fit the number, and a proper threshold value is taken according to the 3 criterion to distinguish between abnormal time windows and normal time windows. If the distance d between the two-dimensional eigenvector Figure 2 of a certain time window and the center of the Gaussian distribution is greater than the threshold value, the time window is an abnormal occurrence time window, and if the distance d is less than the threshold value, the time window is a normal time window. The identification result is shown in

[0071] The first abnormal occurrence time window is taken, and the time domain differential eigenvectors of each voltage transformer in the abnormal occurrence time window and the nearest time windows before and after the abnormal occurrence time window are calculated , and the mean value of the time domain differential eigenvectors of all transformers in the normal time window is taken to obtain a reference differential eigenvector mean vector with a size of , and the reference differential eigenvector mean vector The maximum value and the minimum value in the time-domain difference feature vector of each mutual inductor in the abnormal occurrence time window are normalized by maximum and minimum normalization to obtain a normalized reference difference feature mean vector and a time-domain difference feature vector of each mutual inductor in the abnormal occurrence time window.

[0072] For the normalized reference difference feature mean vector and the time-domain difference feature vector of each mutual inductor in the abnormal occurrence time window, the eight left-open right-closed intervals between 、 and are taken as data intervals to divide the data into three intervals, and the number of data in each interval is divided by the total number of data in the three intervals to obtain the corresponding probability distribution of the reference difference feature mean vector and the time-domain difference feature vector of each mutual inductor in the abnormal occurrence time window and . The KL divergence between and is calculated, and 0.1 is taken as the threshold value . The flow of calculating the KL divergence between and can be seen in Figure 3 . When the KL divergence between and is greater than the threshold value , the corresponding voltage mutual inductor secondary circuit is an abnormal voltage mutual inductor secondary circuit, otherwise, the voltage mutual inductor secondary circuit is a normal voltage mutual inductor secondary circuit, and the final analysis and identification result is shown in Figure 4-1 and Figure 4-2 , the mutual inductor 2 secondary circuit is an abnormal mutual inductor secondary circuit, and other abnormal time window processing modes are the same.

[0073] The application provides an unsupervised dimension reduction voltage mutual inductor secondary circuit anomaly online monitoring method, which acquires time series voltage data in real time through a voltage mutual inductor online monitoring device, then divides and performs frequency domain decomposition according to a time window to obtain a high-dimensional data sample set in the time domain and the frequency domain, thereby reducing the calculation cost. The cosine similarity method is used to extract features from the data sample set in the time and frequency domains to obtain a similarity matrix. The matrix decomposition is used to obtain the feature vectors of the time and frequency domain similarity matrices. The time and frequency domain feature vectors are reduced in dimension to obtain a one-dimensional time domain and frequency domain feature value of each time window as a two-dimensional feature vector of the current time window, which can more accurately extract data features and reduce the influence of redundant information. A mixed Gaussian model is used to establish an unsupervised classification model of the two-dimensional feature vector, and the three The criterion takes a proper threshold value to distinguish abnormal data and normal data, obtains a secondary circuit abnormal occurrence time window, and performs KL divergence analysis according to a time domain difference feature, so that the abnormal secondary circuit is obtained, the result can be more in line with the actual application scene and can be continuously updated, and the monitoring accuracy is improved.

[0074] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0075] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems, or computer program products. Therefore, the application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0076] The application is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flow Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0077] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including instruction devices that implement the flow Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0078] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the flow Figure 1 one or more flows and / or blocks Figure 1steps of the functions specified in the block or blocks.

[0079] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the preferred embodiments by those skilled in the art once they learn of the basic inventive concepts. Therefore, the appended claims are intended to encompass within their scope all such variations and modifications as are included within the scope of the application.

[0080] It is apparent that those skilled in the art can make modifications and variations to the application without departing from the scope of the application. Therefore, the application is intended to cover all such modifications and variations as come within the scope of the following claims and their equivalents.

Claims

1. A method for online monitoring of abnormalities in the secondary circuit of a voltage transformer using unsupervised dimensionality reduction, characterized in that: include: Step 1: Collect data based on voltage transformer online monitoring device The voltage transformers are used to monitor the time series voltage data online, wherein each voltage transformer group includes three voltage transformers, and the time series voltage data is divided into Different time windows are obtained to obtain the secondary circuit time domain high-dimensional data sample set under each time window , perform Fourier transform on the time domain high-dimensional data sample set to obtain a frequency domain high-dimensional data sample set Step 2: respectively analyze the time domain high-dimensional data sample set and the frequency domain high-dimensional data sample set Perform similarity analysis to obtain the time domain similarity matrix under each time window and frequency domain similarity matrix ; Step 3: respectively calculate the time domain similarity matrix and frequency domain similarity matrix Perform matrix decomposition to obtain the time domain eigenvector and frequency domain eigenvectors The time domain feature vector and frequency domain eigenvectors Input into the dimensionality reduction algorithm to obtain the 1-dimensional time domain eigenvalues ​​under each time window and 1D frequency domain eigenvalues After performing maximum and minimum normalization on the 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue, the two-dimensional eigenvector is obtained. is the 1-dimensional time domain eigenvalue after normalization, is the 1D frequency domain eigenvalue after normalization Step 4: convert the two-dimensional feature vector Input into a mixture Gaussian model to identify the abnormal occurrence time window of the secondary circuit, wherein the mixture Gaussian model is used as an unsupervised classification model; Step 5: Calculate the abnormality occurrence time window and its surroundings The time domain differential eigenvector of each voltage transformer under a normal time window is , g , take 2 The mean of the time domain differential characteristic vectors of all voltage transformers in a normal time window is used to obtain the benchmark differential characteristic mean vector , calculate the probability distribution of the time domain difference eigenvector of each voltage transformer in the abnormal time window Probability distribution of the mean vector of the differential feature with the benchmark The KL divergence between them is used to identify abnormal voltage transformer secondary circuits based on the KL divergence.

2. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: In step 1, the time series voltage data is divided into Different time windows are obtained to obtain the secondary circuit time domain high-dimensional data sample set under each time window ,include: At the same time interval Collect once The output of the voltage transformer is Timing voltage data , with a fixed time period The length of the time window, starting from the set start time Start by extracting All the time series voltage data within the time period complete the data segmentation based on the time window, and obtain the size of Time domain high-dimensional data sample set ,in .

3. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: In step 1, the time domain high-dimensional data sample set is Fourier transformed to obtain a frequency domain high-dimensional data sample set. ,include: Perform Fourier transform on the time domain high-dimensional data samples in each time window. The transformation formula is: in, is an imaginary unit, is the time index, is the frequency index, It is frequencies, is the number of samples; Extract its frequency domain voltage amplitude data, the extraction formula is as follows: in, is the frequency domain voltage amplitude data in the frequency domain high-dimensional data sample set, is the complex frequency domain voltage data obtained after Fourier transform, express The real part of express The imaginary part of .

4. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: In step 2, the time domain high-dimensional data sample set is respectively and the frequency domain high-dimensional data sample set Perform similarity analysis to obtain the time domain similarity matrix under each time window and frequency domain similarity matrix ,include: Calculate the time domain high-dimensional data sample set under each time window The cosine similarity between each two columns of data is used to obtain the time domain similarity matrix under each time window ; And calculate the frequency domain high-dimensional data sample set under each time window The cosine similarity between each two columns of data is used to obtain the frequency domain similarity matrix under each time window , where the time domain similarity matrix under each time window is and frequency domain similarity matrix The size of ; The calculation formula for cosine similarity is: in represents the dot product of two vectors, and Represents vectors and vector The model, is a vector and vector The angle between them.

5. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: The time domain feature vector in step 3 and frequency domain eigenvectors The size of , the time domain feature vector Frequency domain eigenvector Input into the dimensionality reduction algorithm to obtain the 1-dimensional time domain eigenvalues ​​under each time window and 1D frequency domain eigenvalues ,include: Extract the time domain feature vector under each time window The principal component analysis, mean, kurtosis or median of the time window is used to obtain the 1-dimensional time domain eigenvalues ​​under each time window. , ; And extract the frequency domain feature vector under each time window The principal component analysis, mean, kurtosis or median of the eigenvalues ​​of the 1D frequency domain in each time window is obtained. , .

6. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: After the 1-dimensional time domain eigenvalue and the 1-dimensional frequency domain eigenvalue are respectively subjected to maximum and minimum normalization processing, the two-dimensional eigenvector is obtained. ,include: The maximum and minimum normalization is performed on the 1-dimensional time domain eigenvalues ​​and 1-dimensional frequency domain eigenvalues ​​under all time windows. The formula is: in, is the original data, and are the minimum and maximum values ​​of the 1-dimensional time domain eigenvalues ​​or 1-dimensional frequency domain eigenvalues ​​under all time windows, respectively. is the 1-dimensional time domain eigenvalue and 1-dimensional frequency domain eigenvalue after maximum and minimum normalization; The normalized 1D time domain eigenvalue is taken as the horizontal coordinate, and the normalized 1D frequency domain eigenvalue data is taken as the vertical coordinate to obtain the two-dimensional eigenvector .

7. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: In step 4, the two-dimensional feature vector Input into the Gaussian mixture model to identify the abnormal time window of the secondary circuit, including: The two-dimensional feature vector Input to the Gaussian mixture model and output the two-dimensional feature vector The covariance matrix of Based on the covariance matrix, the feature quantity is obtained , standard deviation , threshold =3 ; The two-dimensional feature vector Distance from the center of the Gaussian distribution With threshold Make comparisons; like > , then the two-dimensional feature vector The time window where the anomaly occurs is the time window where the anomaly occurs; otherwise, the two-dimensional feature vector The time window is a normal time window; The mixed Gaussian model includes a probability density function composed of multiple Gaussian distributions, and the probability density function of each Gaussian distribution is expressed as: in, is the probability density function, representing the two-dimensional feature vector The probability distribution of is the number of Gaussian distributions, For the The weight coefficient of the Gaussian distribution, For the The density function of a Gaussian distribution, For the The mean of a Gaussian distribution, For the The covariance matrix of a Gaussian distribution.

8. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: Step 5: Calculate the abnormality occurrence time window and its surroundings The time domain differential eigenvector of each transformer under a normal time window is ,include: The time window of abnormal occurrence and its surroundings The time domain high-dimensional sample data of each voltage transformer in each time window of a normal time window is differentiated to obtain the size of The time domain differential eigenvector of each voltage transformer , The number of time series data in each time window.

9. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: Calculate the probability distribution of the time domain difference eigenvector of each voltage transformer within the abnormality occurrence time window Probability distribution of the mean vector of the differential feature with the benchmark ,include: Find the base differential feature mean vector And the time domain differential eigenvector of each voltage transformer in the abnormal time window The maximum and minimum values ​​of all data, the benchmark differential feature mean vector and the time domain differential eigenvector of each voltage transformer Normalization is performed to obtain a normalized reference differential feature mean vector and a time domain differential feature vector of each voltage transformer; Will Divided into left-open and right-closed intervals, and and composition( +2) data intervals, with ( +2) data intervals are used to divide the normalized reference differential feature mean vector and the time domain differential feature vector of each voltage transformer; The probability of each interval is obtained by dividing the number of data in each interval by the total number of data in all intervals, and the time domain differential eigenvector probability distribution of each voltage transformer in the abnormal time window is obtained. and the probability distribution of the mean vector of the benchmark differential feature .

10. The method for online monitoring of abnormalities in the secondary circuit of a voltage transformer according to claim 1, characterized in that: The probability distribution of the time domain differential eigenvector of each voltage transformer in the time window of abnormality occurrence is calculated Probability distribution of the mean vector of the differential feature with the benchmark The KL divergence between , including: in, represents the probability distribution of the time domain differential eigenvector of each voltage transformer in the abnormality occurrence time window, represents the probability distribution of the benchmark differential feature mean vector, Indicates the position of the calculated element in the distribution, for and The KL divergence between The probability distribution of the time domain differential eigenvector of each voltage transformer in the abnormal time window is Probability distribution of the mean vector of the differential feature with the benchmark The KL divergence between When the KL divergence exceeds the threshold , then the corresponding voltage transformer is the abnormal voltage transformer secondary circuit.

Images