Online monitoring method for abnormal operation of voltage transformer and storage medium

Abstract

The invention discloses an online monitoring method for abnormal operation of a voltage transformer, and the method comprises the steps: firstly collecting the secondary side voltage data of a CVT in real time, forming a time sequence, building a bilateral sliding window with a to-be-measured point as a center, and sliding along a time axis to cover a complete sequence; then, K-Means clustering initialization is executed on each sliding window, the clustering number K is set to be equal to 2, two initial centroids are randomly selected, and the distance between each piece of voltage data and the two centroids is calculated; thirdly, dividing the voltage value in the window into the closest centroid category, and recalculating a new centroid until the position of the centroid is stable or reaches the maximum number of iterations; and finally, calculating a current window clustering centroid distance in real time, comparing the current window clustering centroid distance with a dynamic threshold value obtained by historical data training, and judging whether the CVT has an internal fault or not. The method utilizes the existing voltage data, does not need an additional measuring instrument, can adapt to different operation conditions, avoids the over-alarm problem of a traditional fixed threshold value method, and improves the accuracy and robustness of fault diagnosis.

Description

Technical Field

[0001] The invention relates to the technical field of power equipment monitoring, and in particular to an online monitoring method and storage medium for abnormal operation of a voltage transformer. Background Art

[0002] Capacitor voltage transformer (CVT) is an important measuring device in the power system. The stability of its operating state has a vital impact on the safe operation of the power system. Therefore, online monitoring and fault analysis of CVT has always been the focus of the power industry.

[0003] Traditionally, the fault diagnosis of CVT relies on manual monitoring and analysis of specific electrical parameters. For example, Wang Hengshan of Quanzhou Electric Power Bureau proposed a method to determine the fault condition of CVT by monitoring the current I2 flowing through the low-voltage capacitor C2 and the voltage U2 on the primary winding of the intermediate transformer. Although this method can directly determine the type of fault, it requires the installation of corresponding measuring instruments and relies on manual analysis of monitoring data, lacking the ability of automatic processing.

[0004] With the development of power grid monitoring systems, some researchers have begun to use existing real-time power grid monitoring systems to achieve online monitoring of CVT. Gu Zhongde of Changshu Power Supply Company, in "Online Monitoring of CVT Based on OPEN3000 System", implanted judgment and analysis software, used three-phase voltage measurement values ​​to calculate deviations and compared them with preset alarm limits, and realized online monitoring of CVT faults. This method does not require additional equipment, but relies on preset fixed alarm limits, which may cause over-alarm or under-alarm problems under different operating conditions.

[0005] In order to further improve the accuracy and intelligence level of CVT online monitoring, Lin Hao and others from the China Electric Power Research Institute established a fault expert diagnosis system, using the CVT secondary voltage data in the EMS system to monitor the operating status of the CVT and issue early warnings. This method studies the relationship between secondary voltage changes and CVT faults, and considers the impact of grid voltage fluctuations on secondary voltage, providing a new idea for CVT online monitoring technology. However, the implementation of this method requires the support of a complex expert system and a large amount of historical data.

[0006] In addition, Cai Bingbing from Huazhong University of Science and Technology proposed a method of comprehensive parameter analysis using multiple monitoring signals in "Research on CVT Live Detection Technology Based on Multi-Parameters". Although this method is well described, it is not conducive to its promotion and application in actual engineering due to its complexity.

[0007] In summary, the existing CVT online monitoring technology has the following problems: first, it relies on additional measuring instruments or preset fixed alarm limits, lacks adaptability and accuracy; second, it relies on complex expert systems or multi-parameter comprehensive analysis, which is not conducive to promotion and application; third, it lacks the ability to detect local anomalies in real-time voltage data, making it difficult to maintain a high fault detection rate under noise interference. Summary of the invention

[0008] The purpose of the present invention is to provide an online monitoring method for abnormal operation of a voltage transformer in view of the problems existing in the prior art. The method utilizes the existing voltage data without installing additional measuring instruments, and can adapt to different operating conditions, avoid the over-alarm problem of the traditional fixed threshold method, and improve the accuracy and robustness of fault diagnosis.

[0009] To achieve the above object, the technical solution adopted by the present invention is:

[0010] An online monitoring method for abnormal operation of a voltage transformer comprises the following steps:

[0011] S1, real-time collection of CVT secondary side voltage data to form a time series Dn, where each data point includes a voltage value v i and timestamp t i ; Take the point to be measured d i Construct a bilateral sliding window η with a length of 2k+1 as the center i (k) , sliding along the time axis with a fixed step length r to cover the entire sequence;

[0012] S2. Perform K-Means clustering initialization on each sliding window, set the number of clusters K = 2, randomly select two initial centroids c1 and c2 within the window voltage value range, and calculate each voltage data v i The distances to the two centroids c1 and c2 respectively;

[0013] S3, classify each voltage value in the window into the nearest centroid category, recalculate the mean of the two categories of voltage values ​​as the new centroid, and repeat until the centroid position is stable or the maximum number of iterations is reached;

[0014] S4, real-time calculation of the current window cluster centroid distance l i , compared with the dynamic threshold τ, when l i >τ, it is judged as a CVT internal fault; where τ is obtained by training with historical data.

[0015] In step S2, each voltage data v is calculated i The distances to the two centroids c1 and c2, respectively, include:

[0016] Use Euclidean distance to calculate each voltage data vi The distances to the two centroids c1 and c2 are respectively given by:

[0017]

[0018] Step S3 includes:

[0019] S3.1, for all voltage values ​​v1, v2, ..., v in the window 2k+1 , calculate the distance to the two centroids c1 and c2 one by one; divide each data point into the class corresponding to the nearest centroid, forming two sets C1 and C2.

[0020] S3.2. For each class C1 and C2, calculate the mean of all its data points as the new centroid:

[0021]

[0022] S3.3. Repeat until the centroid position is stable or the maximum number of iterations is reached.

[0023] In step S4, the setting of the threshold τ includes the following steps:

[0024] S4.1. Use the LOF algorithm to clean historical normal data and remove abnormal points with LOF values ​​greater than 1.5;

[0025] S4.2. Slide the window with a fixed step size r on the cleaned data, perform K-Means clustering, and calculate the distance l between the centroids of the two categories in each window i =|c i1 -c i2 ∣;

[0026] S4.3. Take all l i The maximum value of the dynamic threshold τ=max{l i}.

[0027] Step S4.1 specifically includes:

[0028] S4.1.1. Set data point v i The number of neighbors is k, calculate the point v i The distance to its kth nearest neighbor is denoted as k-distance (υ i );

[0029] S4.1.2. Calculate v i The mean reachable distance to all its neighbors is:

[0030] reach-dist(υ i , j )=max{k-distance(υ j),||υ i -υ j ||};

[0031] Among them, v j It is point v i The kth distance neighborhood N k (v i ) in any neighbor point, ||υ i -υ j || represents v i With v j The Euclidean distance of

[0032] S4.1.3. Define point v i The local reachable density is:

[0033]

[0034] S4.1.4. Calculate point v i LOF value:

[0035]

[0036] That is, by comparing v i The density (lrd(v i ))With neighbors v j The density (lrd(v j )), quantize v i The degree of abnormality;

[0037] S4.1.5, set when LOF (v i )>1.5, determine v i outliers and removed.

[0038] In step S4.1.1, the number of neighbors k is set to 5, forming a density calculation unit containing 5 nearest neighbor points.

[0039] In step S1, the window length parameter k of the bilateral sliding window satisfies k≥5, forming an analysis window containing at least 11 data points; and the fixed step length r is set to be 1 data point interval.

[0040] The method further includes: step S5, when three consecutive sliding windows detect that li>τ, a CVT capacitor breakdown fault alarm signal is generated and the protection device is triggered to lock.

[0041] A computer-readable storage medium stores a computer program, wherein the computer program implements the above method steps when executed by a processor.

[0042] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the above method steps are implemented when the processor executes the computer program.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. By real-time collection and analysis of CVT secondary side voltage data, combined with bilateral sliding window and K-Means clustering algorithm, it can accurately capture abnormal changes in voltage data and realize timely diagnosis of CVT internal faults; no additional measuring instruments are required, only existing voltage data is used, which reduces monitoring costs;

[0045] 2. Through local cluster analysis of the sliding window, the centroid distance of the voltage cluster in the sliding window is used as the fault feature, the normal voltage fluctuation is regarded as intra-class noise, and the voltage offset caused by the fault is identified as inter-class separation, which conforms to the real environment and achieves a high fault detection rate in the presence of noise interference;

[0046] 3. Through the setting of dynamic thresholds, τ is obtained by training historical data, so that the system can adapt to different operating conditions, avoid the over-alarm problem of the traditional fixed threshold method, and improve the accuracy and robustness of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A schematic diagram of CVT voltage data in an embodiment of the present application;

[0049] Figure 2 A schematic diagram of a sliding window in an embodiment of the present application;

[0050] Figure 3 This is a schematic diagram of the centroid of the class after K-means clustering in an embodiment of the present application;

[0051] Figure 4 Schematic diagram of the centroid of the CVT voltage abnormality interval in the embodiment of the present application. DETAILED DESCRIPTION

[0052] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings 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.

[0053] The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0054] In the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish descriptions and cannot be understood as indicating or suggesting relative importance. It should also be understood that although the terms "first", "second", etc. are used to describe various elements in some embodiments of the present application in the text, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, the first table can be named as the second table, and similarly, the second table can be named as the first table without departing from the scope of the various described embodiments. The first table and the second table are both tables, but they are not the same table.

[0055] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0056] The existing CVT online monitoring technology has the following problems: first, it relies on additional measuring instruments or preset fixed alarm limits, lacks adaptability and accuracy; second, it relies on complex expert systems or multi-parameter comprehensive analysis, which is not conducive to promotion and application; third, it lacks the ability to detect local anomalies in real-time voltage data, making it difficult to maintain a high fault detection rate under noise interference.

[0057] In view of the above technical problems, in a first aspect of an embodiment of the present application, a method for online monitoring of abnormal operation of a voltage transformer is provided, comprising the following steps:

[0058] S1. Real-time collection of CVT secondary voltage data to form a time series D n ={d1(v1,t1),...,d n (v n ,t n )}, where each data point includes a voltage value v i and timestamp t i ,like Figure 1 As shown; take the point to be measured d i Construct a bilateral sliding window η with a length of 2k+1 as the center i (k) ={d i-k ,...,d i+k} Slide along the time axis with a fixed step length r to cover the entire sequence. The sliding window is as follows Figure 2 shown.

[0059] For example, the window length parameter k of the bilateral sliding window satisfies k≥5, forming an analysis window containing at least 11 data points; the fixed step length r is set to 1 data point interval.

[0060] Window center point v i The abnormality of is determined by the degree of cluster separation of its neighborhood data. When a fault occurs inside the CVT, the voltage distribution in the window before and after the fault point will show significant heterogeneity, resulting in a sudden change in the distance between cluster centroids.

[0061] S2. Perform K-Means clustering initialization on each sliding window, set the number of clusters K = 2, randomly select two initial centroids c1 and c2 within the window voltage value range, and calculate each voltage data v i The distances to the two centroids c1 and c2 respectively. The centroids of the classes after K-means clustering are as follows Figure 3 shown.

[0062] S3. Classify each voltage value in the window into the nearest centroid category, recalculate the mean of the two categories of voltage values ​​as the new centroid, and repeat until the centroid position is stable or the maximum number of iterations is reached.

[0063] S4, real-time calculation of the current window cluster centroid distance l i , compared with the dynamic threshold τ, when l i >τ, it is determined as a CVT internal fault, where τ is obtained by training with historical data. The centroid of the class within the abnormal CVT voltage interval is as follows: Figure 4 shown.

[0064] This embodiment scheme can accurately capture abnormal changes in voltage data and realize timely diagnosis of CVT internal faults by real-time collection and analysis of CVT secondary side voltage data, combined with bilateral sliding window and K-Means clustering algorithm. No additional measuring instruments are required, only existing voltage data is used, which reduces monitoring costs. At the same time, the accuracy and robustness of fault diagnosis are improved by setting dynamic thresholds.

[0065] Through local clustering analysis of the sliding window, the voltage cluster centroid distance within the sliding window is used as the fault feature, the normal voltage fluctuation is regarded as intra-class noise, and the voltage offset caused by the fault is identified as inter-class separation, which is in line with the real environment and achieves a high fault detection rate in the presence of noise interference.

[0066] By setting a dynamic threshold τ, which is obtained by training historical data, the system can adapt to different operating conditions and avoid the over-alarm problem of the traditional fixed threshold method.

[0067] In step S2, each voltage data v is calculated i The distances to the two centroids c1 and c2 include:

[0068] Use Euclidean distance to calculate each voltage data v i The distances to the two centroids c1 and c2 are respectively given by:

[0069]

[0070] In the above steps, in order to effectively divide the voltage data in the window into different clusters, the Euclidean distance is used to calculate the distance from each voltage data point vi to the two initial centroids c1 and c2. Euclidean distance is a commonly used distance measurement method that can intuitively reflect the spatial distance between data points, thereby ensuring that the data points can be accurately divided into the nearest centroid category, providing a basis for subsequent centroid updates and cluster division.

[0071] Step S3 specifically includes:

[0072] S3.1, for all voltage values ​​v1, v2, ..., v in the window 2k+1 , calculate the distance to the two centroids c1 and c2 one by one; divide each data point into the class corresponding to the nearest centroid, forming two sets C1 and C2.

[0073] This step achieves the preliminary classification of data points and provides a basis for subsequent centroid update.

[0074] S3.2. For each class C1 and C2, calculate the mean of all its data points as the new centroid:

[0075]

[0076] The new centroid position is the average position of all data points in the class, which can better represent the overall characteristics of the class. By updating the centroid position, the clustering results can be made more accurate and stable.

[0077] S3.3. Repeat steps S3.1 and S3.2 until the centroid position is stable or the maximum number of iterations is reached.

[0078] The stability of the centroid position means that the clustering results have converged and no longer change significantly; reaching the maximum number of iterations is to prevent the algorithm from falling into an infinite loop and ensure that the algorithm can be completed within a limited time.

[0079] Exemplarily, the initial window data is set to: [210V, 215V, 220V, 218V, 225V], and the initial random centroid is: c1 = 215V, c2 = 225V;

[0080] Through the first iteration, [210V, 215V, 218V] is assigned to C1 (close to 215V), and [220V, 225V] is assigned to C2 (close to 225V); the new centroid c1 = 214.33V, c2 = 222.5V is calculated;

[0081] By the second iteration, [210V, 215V, 218V] is assigned to C1 (close to 214.33V), and [220V, 225V] is assigned to C2 (close to 222.5V)

[0082] Repeat the iteration until the centroid no longer changes or the maximum number of iterations is reached, and the iteration is terminated.

[0083] In step S4, the setting of the threshold τ includes the following steps:

[0084] S4.1. Use the LOF algorithm to clean historical normal data and remove abnormal points with LOF values ​​greater than 1.5;

[0085] S4.2. Slide the window with a fixed step size r on the cleaned data, perform K-Means clustering, and calculate the distance l between the centroids of the two categories in each window i =|c i1 -c i2 ∣;

[0086] S4.3. Take all l i The maximum value of the dynamic threshold τ=max{l i}.

[0087] In the above steps, in order to accurately set the dynamic threshold τ for fault judgment, first, the local outlier factor (LOF) algorithm is used to clean the historical normal data, and the abnormal points with LOF values ​​greater than 1.5 are removed. Then, the cleaned data is processed by sliding window with a fixed step size r, and the K-Means clustering algorithm is executed in each window to calculate the distance l between the two types of centroids i , i.e., the centroid c i1 With c i2 The absolute value difference between them can capture the clustering characteristics of the data in a normal state and provide a reliable basis for setting the threshold. Finally, the maximum value of the centroid distance li in all windows is taken as the dynamic threshold τ, that is, τ = max{l i This threshold can reflect the maximum possible value of the centroid distance of the data under normal conditions, providing an accurate reference standard for subsequent fault judgment.

[0088] Exemplarily, after cleaning the historical data, the centroid distances of 10 window clusterings are: [3.2V, 2.8V, 3.5V, 3.0V, 3.6V]; then the dynamic threshold τ=3.6V.

[0089] Step S4.1 specifically includes:

[0090] S4.1.1. Set data point v i The number of neighbors is k, calculate the point v i The distance to its kth nearest neighbor is denoted as k-distance (υ i );

[0091] S4.1.2. Calculate v i The mean reachable distance to all its neighbors is:

[0092] reach-dist(υ i , j )=max{k-distance(υ j ),||υ i -υ j ||};

[0093] Among them, v j It is point v i The kth distance neighborhood N k (v i ) in any neighbor point, ||υ i -υ j || represents v i With v j The Euclidean distance of

[0094] S4.1.3. Define point v i The local reachable density is:

[0095]

[0096] S4.1.4. Calculate point v i LOF value:

[0097]

[0098] That is, by comparing v i The density (lrd(v i ))With neighbors v j The density (lrd(v j )), quantize v i The degree of abnormality;

[0099] S4.1.5, set when LOF (v i )>1.5, determine v i outliers and removed.

[0100] In step S4.1, in order to clean the historical normal data and remove the abnormal points, the local outlier factor (LOF) algorithm is used. Specifically, step S4.1.1 determines the local neighborhood range of each data point, which provides a basis for subsequent calculations. In step S4.1.2, the point v is calculated. i The mean reachable distance to all its neighbors, taking into account v i to its neighbor v j The actual distance to the neighbor v j In step S4.1.3, we define a point v i The local reachable density (lrd(v i )), reflects v i The density of the surrounding data points. In step S4.1.4, calculate the point v i LOF value, that is, by comparing v i The density of its neighbors v j The density of v i The larger the LOF value, the higher the abnormality. i The more abnormal the point is relative to its neighbors. Finally, in step S4.1.5, set LOF(v i )>1.5, determine v i It can effectively identify and remove outliers in the data set.

[0101] In step S4.1.1, the number of neighbors k is set to 5, forming a density calculation unit containing 5 nearest neighbor points.

[0102] In step S4.1.1, in order to determine the local neighborhood range of each data point, the number of neighbors k is set to 5. For each point v in the data set i , they all search for their five nearest neighbor points to form a density calculation unit containing the five nearest neighbor points, which can not only ensure the representativeness of the local neighborhood, but also avoid the computational complexity and accuracy loss caused by too large a neighborhood.

[0103] The method of this embodiment also includes the following steps:

[0104] S5, when three consecutive sliding windows detect l i >τ, a CVT capacitor breakdown fault alarm signal is generated and the protection device is triggered to lock.

[0105] The detection mechanism of three consecutive sliding windows can ensure the accuracy and stability of fault judgment and avoid false alarms caused by single data points or accidental factors. Secondly, the fault alarm signal is generated and the protection device is triggered to lock, which can quickly cut off the fault circuit, prevent the fault from expanding, and protect the safe operation of power equipment.

[0106] According to a second aspect of an embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the above-mentioned method steps are implemented.

[0107] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method steps when executing the computer program.

[0108] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An online monitoring method for abnormal operation of a voltage transformer, characterized in that: The following steps are involved: S1, real-time collection of CVT secondary side voltage data to form a time series Dn, where each data point includes a voltage value v i and timestamp t i ; Take the point to be measured d i Construct a bilateral sliding window η with a length of 2k+1 as the center i (k) , sliding along the time axis with a fixed step length r to cover the entire sequence; S2. Perform K-Means clustering initialization on each sliding window, set the number of clusters K = 2, randomly select two initial centroids c1 and c2 within the window voltage value range, and calculate each voltage data v i The distances to the two centroids c1 and c2 respectively; S3, classify each voltage value in the window into the nearest centroid category, recalculate the mean of the two categories of voltage values ​​as the new centroid, and repeat until the centroid position is stable or the maximum number of iterations is reached; S4, real-time calculation of the current window cluster centroid distance l i , compared with the dynamic threshold τ, when l i >τ, it is judged as a CVT internal fault; where τ is obtained by training with historical data.

2. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: In step S2, each voltage data v is calculated i The distances to the two centroids c1 and c2, respectively, include: Use Euclidean distance to calculate each voltage data v i The distances to the two centroids c1 and c2 are respectively given by:

3. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: Step S3 includes: S3.1, for all voltage values ​​v1, v2, ..., v in the window 2k+1 , calculate the distance to the two centroids c1 and c2 one by one; divide each data point into the class corresponding to the nearest centroid, forming two sets C1 and C2; S3.

2. For each class C1 and C2, calculate the mean of all its data points as the new centroid: S3.

3. Repeat until the centroid position is stable or the maximum number of iterations is reached.

4. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: In step S4, the setting of the threshold τ includes the following steps: S4.

1. Use the LOF algorithm to clean historical normal data and remove abnormal points with LOF values ​​greater than 1.5; S4.

2. Slide the window with a fixed step size r on the cleaned data, perform K-Means clustering, and calculate the distance l between the centroids of the two categories in each window i =|c i1 -c i2 ∣; S4.

3. Take all l i The maximum value of the dynamic threshold τ=max{l i }.

5. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: Step S4.1 specifically includes: S4.1.

1. Set data point v i The number of neighbors is k, calculate the point v i The distance to its kth nearest neighbor is denoted as k-distance(v i ); S4.1.

2. Calculate v i The mean reachable distance to all its neighbors is: reach-dist(v i ,v j )=max{k-distance(v j ),||v i -v j ||}; Among them, v j It is point v i The kth distance neighborhood N k (v i ) in any neighbor point, ||v i -v j || represents v i With v j The Euclidean distance of S4.1.

3. Define point v i The local reachable density is: S4.1.

4. Calculate point v i LOF value: That is, by comparing v i The density (lrd(v i ))With neighbors v j The density (lrd(v j )), quantize v i The degree of abnormality; S4.1.5, set when LOF (v i )>1.5, determine v i outliers and removed.

6. The method for online monitoring of abnormal operation of a voltage transformer according to claim 5, characterized in that: In step S4.1.1, the number of neighbors k is set to 5, forming a density calculation unit containing 5 nearest neighbor points.

7. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: In step S1, the window length parameter k of the bilateral sliding window satisfies k≥5, forming an analysis window containing at least 11 data points; and the fixed step length r is set to be 1 data point interval.

8. The method for online monitoring of abnormal operation of a voltage transformer according to claim 1, characterized in that: Also includes: S5, when three consecutive sliding windows detect l i >τ, a CVT capacitor breakdown fault alarm signal is generated and the protection device is triggered to lock.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the method steps of any one of claims 1 to 8 when executed by a processor.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method steps of any one of claims 1 to 8 are implemented.

Images