CN-122020110-A - Frequency response data change characteristic generation method, equipment and storage medium for transformer winding fault category identification

CN122020110ACN 122020110 ACN122020110 ACN 122020110ACN-122020110-A

Abstract

The invention discloses a frequency response data change characteristic generation method, equipment and storage medium for transformer winding fault category identification, which are used for respectively converting normal frequency response amplitude-frequency data and frequency response amplitude-frequency data of each fault winding into frequency response amplitude-frequency polar coordinate graphs corresponding to each fault winding; calculating the multi-class geometrical characteristics of all frequency response amplitude-frequency polar coordinate graphs, calculating the difference value between the radius of each point in each fault winding frequency response amplitude-frequency polar coordinate graph and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph, and forming a characteristic vector for a machine learning model by the multi-class geometrical characteristics and the radius difference value. The method can provide important references and guidance for the feature extraction problem in the fault diagnosis of the transformer winding based on the frequency response analysis method and the machine learning method.

Inventors

XIONG JUNHUI
YE FAN
Xiong xinheng
QI JINWEI
ZHU YI
SHU LIAN
ZHANG HUAZHAO
YOU MIN

Assignees

国网江西省电力有限公司南昌供电分公司

Dates

Publication Date: 20260512
Application Date: 20251212

Claims (9)

1. A frequency response data change characteristic generation method for transformer winding fault category identification is characterized by comprising the following steps: Acquiring normal frequency response amplitude-frequency data of windings and frequency response amplitude-frequency data of each fault winding; Converting the frequency response amplitude-frequency data of each fault winding into a frequency response amplitude-frequency polar coordinate graph of each corresponding fault winding; Calculating the same multi-class geometric features of the frequency response amplitude-frequency polar coordinate graph of each fault winding; And calculating the difference value between the radius of each point in each fault winding frequency response amplitude-frequency polar coordinate graph and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph; and forming a feature vector for the machine learning model by using the multiple types of geometric features of the normal frequency response amplitude-frequency polar coordinate graph, the multiple types of geometric features of each fault winding frequency response amplitude-frequency polar coordinate graph and the difference value between the radius of each point in each fault winding frequency response amplitude-frequency polar coordinate graph and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph.
2. The method for generating the frequency response data change characteristics for transformer winding fault class identification according to claim 1, wherein each frequency point is used as a point in a corresponding polar coordinate graph when the normal frequency response amplitude-frequency data or the frequency response amplitude-frequency data of the fault winding is converted into the corresponding polar coordinate graph; According to the number of frequency points of the frequency response, the angle theta of the polar coordinate is defined as shown in the following formula: , wherein n is the number of frequency points in the frequency response data: taking the amplitude corresponding to each frequency point in the frequency response data as the amplitude of each point in the polar coordinates, wherein the amplitude is shown as the following formula: , Wherein r (i) is the radius of the ith point in the frequency response amplitude-frequency polar graph, and H (i) is the value corresponding to the ith frequency point in the frequency response amplitude-frequency data, i= [1, n ].
3. The method for generating the frequency response data change characteristics for transformer winding fault category identification according to claim 1, wherein the calculated multiple types of geometric characteristics of the normal frequency response amplitude-frequency polar graph and the calculated multiple types of geometric characteristics of the fault winding frequency response amplitude-frequency polar graph are all circularity, standard deviation and perimeter area ratio.
4. A frequency response data change characteristic generation method for transformer winding fault class identification according to claim 3, wherein the circularity Yd is calculated as follows: , Wherein mu r is the average value of all the point radiuses in the amplitude-frequency polar graph of the corresponding frequency response and has 。
5. A frequency response data change signature generation method for transformer winding fault class identification as recited in claim 3 wherein said standard deviation is calculated as: 。
6. A frequency response data change characteristic generation method for transformer winding fault class identification according to claim 3, wherein the perimeter-to-area ratio CR is calculated as: , Wherein, the , , In the above formula, θ n+1 represents the angle of the ith point in the polar graph of the corresponding frequency response amplitude-frequency, and when i=n, i+1=n+1, and when θ n+1 =θ 1 and r (i+1) =r (1) are present.
7. A method for generating frequency response data variation characteristics for transformer winding fault class identification according to any one of claims 1-6, wherein a normal frequency response amplitude-frequency polar graph, multiple classes of geometric characteristics of each fault winding frequency response amplitude-frequency polar graph, and a radius difference between each point in each fault winding frequency response amplitude-frequency polar graph and each point in the normal frequency response amplitude-frequency polar graph are connected in series, thereby obtaining said characteristic vector.
8. An electronic device comprising a processor and a memory, wherein program instructions in the memory, when read and executed, perform a frequency response data change signature generation method for transformer winding fault class identification as claimed in any one of claims 1-7.
9. A storage medium storing program instructions which, when read and executed, perform a frequency response data change signature generation method for transformer winding fault class identification as claimed in any one of claims 1 to 7.

Description

Frequency response data change characteristic generation method, equipment and storage medium for transformer winding fault category identification Technical Field The invention relates to the field of data generation methods for power equipment fault diagnosis, in particular to a frequency response data change characteristic generation method, equipment and storage medium for transformer winding fault category identification. Background As the load of the power grid increases, the short-circuit current level of the power grid also increases, and the problem that the transformer is damaged due to short-circuit current impact is more and more remarkable. The frequency response analysis method has been widely used as a sensitive transformer winding deformation detection method. The frequency response method is used as a comparison method, and mainly comprises the steps of comparing current winding frequency response data with normal data of windings, and diagnosing faults of the windings of the transformer by analyzing changes among the data. At present, with the development of artificial intelligence technology, a machine learning method is combined with a frequency response method to become a way for realizing quantitative diagnosis of transformer winding faults. When combining the two methods, accurately calculating the change characteristics of the transformer winding frequency response data is critical to the results of the machine learning diagnostic model. However, the variation characteristics obtained by the existing frequency response data variation characteristic calculation method are often not comprehensive enough, so that the recognition accuracy based on machine learning is not high. Disclosure of Invention The invention provides a frequency response data change characteristic generation method, equipment and a storage medium for transformer winding fault category identification, which are used for solving the problem that the change characteristic of a frequency response data change characteristic calculation method in the prior art is not comprehensive enough. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: A frequency response data change characteristic generation method for transformer winding fault category identification comprises the following steps: Acquiring normal frequency response amplitude-frequency data of windings and frequency response amplitude-frequency data of each fault winding; Converting the frequency response amplitude-frequency data of each fault winding into a frequency response amplitude-frequency polar coordinate graph of each corresponding fault winding; Calculating the same multi-class geometric features of the frequency response amplitude-frequency polar coordinate graph of each fault winding; And calculating the difference value between the radius of each point in each fault winding frequency response amplitude-frequency polar coordinate graph and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph; and forming a feature vector for the machine learning model by using the multiple types of geometric features of the normal frequency response amplitude-frequency polar coordinate graph, the multiple types of geometric features of each fault winding frequency response amplitude-frequency polar coordinate graph and the difference value between the radius of each point in each fault winding frequency response amplitude-frequency polar coordinate graph and the radius of each point in the normal frequency response amplitude-frequency polar coordinate graph. Further, when the normal frequency response amplitude-frequency data or the frequency response amplitude-frequency data of the fault winding are converted into the corresponding polar coordinate graph, each frequency point is used as a point in the corresponding polar coordinate graph; According to the number of frequency points of the frequency response, the angle theta of the polar coordinate is defined as shown in the following formula: wherein n is the number of frequency points in the frequency response data: taking the amplitude corresponding to each frequency point in the frequency response data as the amplitude of each point in the polar coordinates, wherein the amplitude is shown as the following formula: Wherein r (i) is the radius of the ith point in the frequency response amplitude-frequency polar graph, and H (i) is the value corresponding to the ith frequency point in the frequency response amplitude-frequency data, i= [1, n ]. Further, the calculated multi-class geometric features of the normal frequency response amplitude-frequency polar coordinate graph and the calculated multi-class geometric features of the fault winding frequency response amplitude-frequency polar coordinate graph are all circularity, standard deviation and perimeter area ratio. Further, the circularity Yd is calculated as follows: Wh