Frequency response data change feature generation method and device for transformer winding fault category identification, and storage medium

Abstract

The invention discloses a frequency response data change feature generation method and device for transformer winding fault category identification and a storage medium. The method comprises the following steps: respectively converting normal frequency response amplitude-frequency data and frequency response amplitude-frequency data of each fault winding into respective corresponding frequency response amplitude-frequency polar coordinate graphs; calculating multiple types of geometric features of all the frequency response amplitude-frequency polar coordinate graphs, and calculating a difference value between the radius of each point in the frequency response amplitude-frequency polar coordinate graph of each fault winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph; the feature vector for the machine learning model is formed by the multiple types of geometric features and the radius difference values. The equipment and the storage medium are used for implementing the method. The method can provide important reference and guidance for the feature extraction problem in transformer winding fault diagnosis based on a frequency response analysis method and a machine learning method.

Description

Technical Field

[0001] This invention relates to the field of data generation methods for fault diagnosis of power equipment, specifically a method, device, and storage medium for generating frequency response data change characteristics for identifying fault categories in transformer windings. Background Technology

[0002] With the continuous increase in power grid load, the short-circuit current level of the power grid is also increasing, making the problem of transformer damage caused by short-circuit current impacts more prominent. Frequency response analysis, as a sensitive method for detecting transformer winding deformation, has been widely used. As a comparative method, the frequency response method mainly diagnoses transformer winding faults by comparing the current winding frequency response data with normal winding data and analyzing the changes between the data. Currently, with the development of artificial intelligence technology, machine learning methods combined with the frequency response method have become a way to achieve quantitative diagnosis of transformer winding faults. When the two methods are combined, accurately calculating the changing characteristics of the transformer winding frequency response data is crucial to the results of the machine learning diagnostic model. However, existing methods for calculating the changing characteristics of frequency response data often do not obtain comprehensive features, resulting in low accuracy of machine learning-based identification. Summary of the Invention

[0003] This invention provides a method, device, and storage medium for generating frequency response data change features for transformer winding fault category identification, in order to solve the problem that the existing frequency response data change feature calculation methods are not comprehensive enough.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A method for generating frequency response data variation features for transformer winding fault category identification, the process is as follows:

[0006] Acquire the normal frequency response amplitude-frequency data of the winding, as well as the frequency response amplitude-frequency data of each faulty winding;

[0007] Convert the normal frequency response amplitude-frequency data into a normal frequency response amplitude-frequency polar plot; convert the frequency response amplitude-frequency data of each faulty winding into its corresponding faulty winding frequency response amplitude-frequency polar plot.

[0008] Calculate multiple geometric features of the normal frequency response amplitude-frequency polar plot; calculate the same multiple geometric features of the frequency response amplitude-frequency polar plot for each faulty winding;

[0009] Furthermore, the difference between the radius of each point in the frequency response amplitude-frequency polar coordinate diagram of each faulty winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate diagram is calculated.

[0010] The feature vector for machine learning models is composed of multiple geometric features of the normal frequency response amplitude-frequency polar coordinate plot, multiple geometric features of the frequency response amplitude-frequency polar coordinate plot of each fault winding, and the difference between the radius of each point in the frequency response amplitude-frequency polar coordinate plot of each fault winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate plot.

[0011] Furthermore, when converting the normal frequency response amplitude-frequency data or the frequency response amplitude-frequency data of the faulty winding into the corresponding polar coordinate graph, each frequency point is used as a point in the corresponding polar coordinate graph.

[0012] Based on the number of frequency points in the frequency response, the angle θ in polar coordinates is defined as follows:

[0013]

[0014] Where n is the number of frequency points in the frequency response data:

[0015] The amplitude corresponding to each frequency point in the frequency response data is taken as the amplitude of each point in the polar coordinates, as shown in the following formula:

[0016]

[0017] Where r(i) is the radius of the i-th point in the frequency response amplitude-frequency polar coordinate graph; H(i) is the value corresponding to the i-th frequency point in the frequency response amplitude-frequency data, i=[1,n].

[0018] Furthermore, the various geometric features of the normal frequency response amplitude-frequency polar coordinate diagram and the fault winding frequency response amplitude-frequency polar coordinate diagram are all circularity, standard deviation, and perimeter-to-area ratio.

[0019] Furthermore, the roundness Yd is calculated as follows:

[0020]

[0021] Where, μ r This is the average radius of all points in the amplitude-frequency polar plot of the corresponding frequency response, and has... .

[0022] Furthermore, the standard deviation is calculated as follows:

[0023] .

[0024] Furthermore, the perimeter-to-area ratio CR is calculated as follows:

[0025]

[0026] in,

[0027]

[0028] In the above formula, θ n+1 This represents the angle of the i-th point in the amplitude-frequency polar coordinate graph of the corresponding frequency response; and when i=n, ​​i+1=n+1, at which point θ n+1 =θ1、r(i+1) = r(1).

[0029] Furthermore, the feature vector is obtained by concatenating the various geometric features of the normal frequency response amplitude-frequency polar coordinate diagram, the frequency response amplitude-frequency polar coordinate diagram of each fault winding, and the radius difference between each point in the frequency response amplitude-frequency polar coordinate diagram of each fault winding and each point in the normal frequency response amplitude-frequency polar coordinate diagram.

[0030] An electronic device includes a processor and a memory, wherein program instructions in the memory are read and executed to perform the above-described method for generating frequency response data change features for identifying transformer winding fault categories.

[0031] A storage medium storing program instructions, which, when read and executed, perform the above-described method for generating frequency response data change features for identifying transformer winding fault categories.

[0032] This invention maps frequency response data to polar coordinates, extracts multi-dimensional geometric features such as roundness, standard deviation, and perimeter-to-area ratio, as well as the radius difference between points in different polar coordinates, thereby constructing a feature vector that integrates two-dimensional polar coordinate morphological features and one-dimensional data deviation features.

[0033] By employing features such as circularity, standard deviation, perimeter area, and the radius difference between points in different polar coordinate plots, this invention not only preserves the features of the original one-dimensional data but also includes the two-dimensional image features transformed into polar coordinate plots. The resulting feature vectors enrich the composition of subsequent machine learning models. Therefore, the feature vectors obtained by this invention significantly improve the representational ability of the features, providing a more comprehensive and discriminative input for machine learning models, thereby improving the accuracy and reliability of transformer winding fault type identification. Attached Figure Description

[0034] Figure 1 This is a flowchart of the method according to an embodiment of the present invention.

[0035] Figure 2 This is a polar coordinate graph of the normal frequency response data in an embodiment of the present invention.

[0036] Figure 3 This is a polar coordinate plot of the frequency response data of an axial displacement fault according to an embodiment of the present invention.

[0037] Figure 4 This is a polar coordinate plot of the frequency response data for inter-pie spacing faults according to an embodiment of the present invention.

[0038] Figure 5 This is a polar coordinate plot of the frequency response data of a radial fault according to an embodiment of the present invention. Detailed Implementation

[0039] The present invention will be further described in detail below with reference to specific embodiments. These descriptions are illustrative rather than limiting. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments in the specification without inventive effort are within the scope of protection of this application.

[0040] like Figure 1 As shown in the figure, this embodiment discloses a method for generating frequency response data change features for transformer winding fault category identification, the process of which is as follows:

[0041] Step S01: Convert the frequency response amplitude-frequency data of different faulty windings into a frequency response amplitude-frequency polar coordinate graph of the faulty winding, and convert the normal frequency response amplitude-frequency data of the winding into a normal frequency response amplitude-frequency polar coordinate graph.

[0042] Specifically, each frequency point in the frequency response amplitude-frequency data is taken as a point in the corresponding frequency response amplitude-frequency polar coordinate graph, and the polar coordinate angle θ is shown in formula (1):

[0043] (1)

[0044] Where n is the number of frequency points in the frequency response data.

[0045] Based on the amplitude value corresponding to each frequency point in the frequency response amplitude-frequency data, the amplitude value of each point in the corresponding frequency response amplitude-frequency polar coordinates is obtained, as shown in formula (2):

[0046] (2)

[0047] Where r(i) is the radius of the i-th point in the frequency response amplitude-frequency polar coordinate graph; H(i) is the value corresponding to the i-th frequency point in the frequency response amplitude-frequency data, i=[1,n].

[0048] Figure 2 This is a polar plot of the normal frequency response amplitude-frequency data, converted from normal frequency response amplitude-frequency data. Figures 3 to 5These are polar coordinate graphs of the frequency response amplitude-frequency data of different faulty windings, converted from the original data. The faults include axial displacement winding faults, inter-winding spacing faults, and radial bulge faults in the windings.

[0049] Step S02: Calculate the three geometric characteristic parameters—circularity, standard deviation, and perimeter-to-area ratio—of the frequency response amplitude-frequency polar plot for each faulty winding. Also calculate the three geometric characteristic parameters—circularity, standard deviation, and perimeter-to-area ratio—of the normal frequency response amplitude-frequency polar plot. Wherein:

[0050] The roundness Yd is calculated as shown in formula (2):

[0051] (3)

[0052] Where, μ r The average value of the radii of all points in either the amplitude-frequency polar plot of the frequency response of the faulty winding or the amplitude-frequency polar plot of the normal frequency response is equal to... .

[0053] The standard deviation is calculated as shown in formula (4):

[0054] (4)

[0055] The perimeter-to-area ratio is calculated as shown in formula (5):

[0056] (5)

[0057] in,

[0058]

[0059] In the above formula, θ n+1 This represents the angle of the i-th point in the polar coordinate graph; and when i=n, ​​i+1=n+1, at which point θ n+1 =θ1、r(i+1) = r(1).

[0060] Furthermore, the difference between the radius of each point in the frequency response amplitude-frequency polar plot of each faulty winding and the radius of each point in the normal frequency response amplitude-frequency polar plot is calculated. As shown in formula (6):

[0061] (6)

[0062] in, The radius of the i-th point in the frequency response amplitude-frequency polar coordinate graph under normal conditions; Let be the radius of the i-th point in the frequency response amplitude-frequency polar coordinate graph of the fault condition.

[0063] Step S03: The circularity, standard deviation, and perimeter-to-area ratio of the calculated normal frequency response amplitude-frequency polar coordinate diagram and the frequency response amplitude-frequency polar coordinate diagram of each fault winding, as well as the difference between the radius of each point in the frequency response amplitude-frequency polar coordinate diagram of each fault winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate diagram, are concatenated and combined to form the feature vector T for the machine learning model, as shown in formula (7):

[0064] T=[ (1), (2),…, (n), Yd, σ, CR] (7).

[0065] This embodiment also discloses an electronic device, including a processor and a memory. Program instructions in the memory are read and executed by the processor to perform steps S01-S03 of the frequency response data change feature generation method for transformer winding fault category identification described above. The program instructions include a polar coordinate graph generation module, a feature calculation module, and a feature vector generation module. The polar coordinate graph generation module generates a corresponding frequency response amplitude-frequency polar coordinate graph based on the frequency response amplitude-frequency data. The feature calculation module calculates three types of geometric features—circularity, standard deviation, and perimeter-to-area ratio—based on the frequency response amplitude-frequency polar coordinate graph, and calculates the radius differences between points in different frequency response amplitude-frequency polar coordinate graphs. The feature vector generation module organizes multiple types of geometric features and radius differences into a feature vector according to a predetermined concatenation order.

[0066] This embodiment also discloses a storage medium that stores program instructions. When the program instructions are read and run, steps S01-S03 of the frequency response data change feature generation method for transformer winding fault category identification are executed.

[0067] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. These embodiments are merely descriptions of preferred embodiments and are not intended to limit the scope or concept of the invention. The specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. Such combinations, as long as they do not violate the spirit of the present invention, should also be considered as part of this disclosure. To avoid unnecessary repetition, the present invention will not further describe the various possible combinations.

[0068] This invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this invention and without departing from the design idea of ​​this invention, all modifications and improvements made by those skilled in the art to the technical solutions of this invention should fall within the protection scope of this invention. The technical content for which protection is sought in this invention has been fully described in the claims.

Claims

1. A method for generating frequency response data variation features for transformer winding fault category identification, characterized in that, The process is as follows: Acquire the normal frequency response amplitude-frequency data of the winding, as well as the frequency response amplitude-frequency data of each faulty winding; Convert the normal frequency response amplitude-frequency data into a normal frequency response amplitude-frequency polar plot; convert the frequency response amplitude-frequency data of each faulty winding into its corresponding faulty winding frequency response amplitude-frequency polar plot. Calculate multiple geometric features of the amplitude-frequency polar plot of the normal frequency response; Calculate the same multiple geometric features of the amplitude-frequency polar plot of the frequency response for each faulted winding; Furthermore, the difference between the radius of each point in the frequency response amplitude-frequency polar coordinate diagram of each faulty winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate diagram is calculated. The feature vector for machine learning models is composed of multiple geometric features of the normal frequency response amplitude-frequency polar coordinate plot, multiple geometric features of the frequency response amplitude-frequency polar coordinate plot of each fault winding, and the difference between the radius of each point in the frequency response amplitude-frequency polar coordinate plot of each fault winding and the radius of each point in the normal frequency response amplitude-frequency polar coordinate plot.

2. The method for generating frequency response data change features for transformer winding fault category identification according to claim 1, characterized in that, When converting normal frequency response amplitude-frequency data or faulty winding frequency response amplitude-frequency data into the corresponding polar coordinate graph, each frequency point is used as a point in the corresponding polar coordinate graph. Based on the number of frequency points in the frequency response, the angle θ in polar coordinates is defined as follows: ， Where n is the number of frequency points in the frequency response data: The amplitude corresponding to each frequency point in the frequency response data is taken as the amplitude of each point in the polar coordinates, as shown in the following formula: ， Where r(i) is the radius of the i-th point in the frequency response amplitude-frequency polar coordinate graph; H(i) is the value corresponding to the i-th frequency point in the frequency response amplitude-frequency data, i=[1,n].

3. The method for generating frequency response data variation features for transformer winding fault category identification according to claim 1, characterized in that, The various geometric features of the normal frequency response amplitude-frequency polar coordinate diagram and the various geometric features of the fault winding frequency response amplitude-frequency polar coordinate diagram are all circularity, standard deviation, and perimeter-area ratio.

4. The method for generating frequency response data variation features for transformer winding fault category identification according to claim 3, characterized in that, The roundness Yd is calculated as follows: ， Where, μ r This is the average radius of all points in the amplitude-frequency polar plot of the corresponding frequency response, and has... .

5. The method for generating frequency response data variation features for transformer winding fault category identification according to claim 3, characterized in that, The standard deviation is calculated as follows: 。 6. The method for generating frequency response data variation features for transformer winding fault category identification according to claim 3, characterized in that, The perimeter-to-area ratio CR is calculated as follows: ， in, , ， In the above formula, θ n+1 This represents the angle of the i-th point in the amplitude-frequency polar coordinate graph of the corresponding frequency response; and when i=n, ​​i+1=n+1, at which point θ n+1 =θ1、r(i+1) = r(1).

7. A method for generating frequency response data variation features for transformer winding fault category identification according to any one of claims 1-6, characterized in that, The feature vector is obtained by concatenating the various geometric features of the normal frequency response amplitude-frequency polar coordinate diagram, the frequency response amplitude-frequency polar coordinate diagram of each fault winding, and the radius difference between each point in the frequency response amplitude-frequency polar coordinate diagram of each fault winding and each point in the normal frequency response amplitude-frequency polar coordinate diagram.

8. An electronic device comprising a processor and a memory, characterized in that, When the program instructions in the memory are read and executed, the frequency response data change feature generation method for transformer winding fault category identification, as described in any one of claims 1-7, is performed.

9. A storage medium storing program instructions, characterized in that, When the program instructions are read and executed, a method for generating frequency response data change features for identifying transformer winding fault categories, as described in any one of claims 1-7, is performed.

Images