CN-121978171-A - Transformer core defect detection method, device and equipment

CN121978171ACN 121978171 ACN121978171 ACN 121978171ACN-121978171-A

Abstract

The application discloses a transformer core defect detection method, device and equipment, which comprise the steps of collecting multi-mode data in the operation process of a transformer, wherein the multi-mode data comprise vibration signals, temperature fields, gas concentration and harmonic currents, carrying out feature fusion processing on the multi-mode data according to modal weights and a preset cascade fusion network to obtain fusion feature vectors, constructing a space impedance distribution equation according to the fusion feature vectors, carrying out electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix, and carrying out defect detection analysis according to the fusion feature vectors and the impedance distribution matrix based on a constraint reinforcement learning strategy and a preset core diagram model to obtain a defect evaluation grade. The method can solve the technical problems that the prior art is difficult to realize multi-source heterogeneous data fusion, and the coupling influence among cross physical fields is not considered, so that the defect detection result lacks pertinence, accuracy and reliability.

Inventors

TANG QI
CAO DEFA
QIU WEI
HUANG GUOPING
HUANG JING

Assignees

广东电网有限责任公司佛山供电局

Dates

Publication Date: 20260505
Application Date: 20260127

Claims (10)

1. A method for detecting a defect in a transformer core, comprising: Collecting multi-mode data in the operation process of the transformer, wherein the multi-mode data comprises vibration signals, a temperature field, gas concentration and harmonic current; Performing feature fusion processing on the multi-mode data according to the modal weight and a preset cascade fusion network to obtain a fusion feature vector; constructing a space impedance distribution equation according to the fusion feature vector, and performing electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix; And performing defect detection analysis according to the fusion feature vector and the impedance distribution matrix based on a constraint reinforcement learning strategy and a preset iron core diagram model to obtain a defect evaluation grade.
2. The method for detecting defects of a transformer core according to claim 1, wherein the collecting multi-modal data during operation of the transformer comprises: acquiring a vibration spectrum of the transformer in the operation process according to a preset window length through FFT (fast Fourier transform), and obtaining a vibration signal; Acquiring an infrared thermal image space gradient in the operation process of the transformer based on a thermal imager, and determining a temperature field; determining the concentration of dissolved gas in the transformer oil based on a preset line chromatograph to obtain the gas concentration; and collecting current harmonic waves in the operation process of the transformer in a measurement mode to obtain harmonic current.
3. The method for detecting a defect of a transformer core according to claim 1, wherein the performing feature fusion processing on the multi-mode data according to the mode weight and a preset cascade fusion network to obtain a fusion feature vector comprises: performing kernel density estimation calculation according to the multi-modal data to obtain modal weights; And performing cascade feature fusion calculation based on a preset cascade fusion network according to the modal weight and the multi-modal data to obtain a fusion feature vector.
4. The method for detecting defects of a transformer core according to claim 1, wherein constructing a spatial impedance distribution equation according to the fused eigenvector and performing electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix comprises: Carrying out joule heating effect analysis and harmonic excitation analysis according to the temperature field and harmonic current in the fusion characteristic vector, and generating a space impedance distribution equation; And performing electromagnetic-thermal coupling effect iterative analysis based on a preset time step according to the space impedance distribution equation to obtain an impedance distribution matrix.
5. The method for detecting a defect of a transformer core according to claim 1, wherein the performing defect detection analysis based on the constraint reinforcement learning strategy and a preset core diagram model according to the fusion feature vector and the impedance distribution matrix to obtain a defect evaluation level comprises: Constructing an iron core three-dimensional grid topological graph for the transformer to obtain a preset iron core graph model; inputting the fusion feature vector, the impedance distribution matrix and the vibration propagation function into the preset iron core diagram model for convolution calculation to obtain an abnormal probability diagram; And carrying out defect detection analysis according to the abnormal probability map based on a constraint reinforcement learning strategy to obtain a defect evaluation grade.
6. The method for detecting a defect of a transformer core according to claim 5, wherein the inputting the fusion eigenvector, the impedance distribution matrix, and the vibration propagation function into the preset core pattern model for convolution calculation to obtain an anomaly probability map comprises: Constructing an adjacency matrix according to the vibration propagation function; calculating a temperature mask based on a temperature field gradient in the fused feature vector; And updating and calculating the graph convolution characteristics of the preset iron core graph model according to the adjacent matrix, the temperature mask and the impedance distribution matrix to obtain an abnormal probability graph.
7. The method for detecting defects of a transformer core according to claim 5, wherein the performing defect detection analysis according to the anomaly probability map based on the constraint reinforcement learning strategy to obtain a defect evaluation level comprises: Based on Fourier law, configuring a double-constraint reward function according to the gas change rate and the heat flow divergence; And based on a constraint reinforcement learning strategy, performing defect detection analysis based on action-state updating operation according to the abnormal probability map, the impedance distribution matrix and the double-constraint reward function to obtain a defect evaluation grade.
8. The method for detecting a defect of a transformer core according to claim 5, wherein the performing defect detection analysis based on the constraint reinforcement learning strategy and the preset core pattern model according to the fusion feature vector and the impedance distribution matrix to obtain a defect evaluation level further comprises: And performing joint inversion analysis according to the defect evaluation grade and the abnormal probability in the abnormal probability map through a physical information neural network to obtain defect information, wherein the defect information comprises defect positions, defect areas and defect depths.
9. A transformer core defect detection apparatus, comprising: The data acquisition unit is used for acquiring multi-mode data in the operation process of the transformer, wherein the multi-mode data comprises vibration signals, a temperature field, gas concentration and harmonic current; the feature fusion unit is used for carrying out feature fusion processing on the multi-mode data according to the modal weight and a preset cascade fusion network to obtain a fusion feature vector; The coupling analysis unit is used for constructing a space impedance distribution equation according to the fusion characteristic vector and carrying out electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix; And the defect detection unit is used for carrying out defect detection analysis according to the fusion characteristic vector and the impedance distribution matrix based on a constraint reinforcement learning strategy and a preset iron core diagram model to obtain a defect evaluation grade.
10. A transformer core defect detection apparatus, the apparatus comprising a processor and a memory; the memory is used for storing program codes and transmitting the program codes to the processor; The processor is configured to perform the transformer core defect detection method of any one of claims 1-8 according to instructions in the program code.

Description

Transformer core defect detection method, device and equipment Technical Field The present application relates to the field of transformer fault monitoring, and in particular, to a method, an apparatus, and a device for detecting a transformer core defect. Background With the expansion of the grid-connected scale of the ultra-high voltage power grid and the new energy, the problem of unplanned shutdown caused by the defect of the transformer core is increasingly outstanding. The traditional detection technology relies on a single-mode threshold value alarm, such as a detection scheme of oil chromatography or infrared temperature measurement, and has obvious technical defects of high false alarm rate (15%), rough positioning (centimeter level) and the like. In recent years, the development of digital twin and multi-physical field simulation technology provides a new thought for iron core state evaluation, but at present, there are still some problems that firstly, multi-source heterogeneous data are difficult to effectively fuse, and secondly, coupling influence among cross physical fields is not considered. These problems lead to the lack of pertinence, accuracy and reliability of the iron core defect detection results, and further cannot meet the application requirements of actual scenes. Disclosure of Invention The application provides a method, a device and equipment for detecting defects of a transformer iron core, which are used for solving the technical problems that the prior art is difficult to realize multi-source heterogeneous data fusion, and the defect detection result lacks pertinence, accuracy and reliability because coupling influence among cross physical fields is not considered. In view of the foregoing, a first aspect of the present application provides a method for detecting a defect of a transformer core, including: Collecting multi-mode data in the operation process of the transformer, wherein the multi-mode data comprises vibration signals, a temperature field, gas concentration and harmonic current; Performing feature fusion processing on the multi-mode data according to the modal weight and a preset cascade fusion network to obtain a fusion feature vector; constructing a space impedance distribution equation according to the fusion feature vector, and performing electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix; And performing defect detection analysis according to the fusion feature vector and the impedance distribution matrix based on a constraint reinforcement learning strategy and a preset iron core diagram model to obtain a defect evaluation grade. Preferably, the collecting multi-mode data during operation of the transformer includes: acquiring a vibration spectrum of the transformer in the operation process according to a preset window length through FFT (fast Fourier transform), and obtaining a vibration signal; Acquiring an infrared thermal image space gradient in the operation process of the transformer based on a thermal imager, and determining a temperature field; determining the concentration of dissolved gas in the transformer oil based on a preset line chromatograph to obtain the gas concentration; and collecting current harmonic waves in the operation process of the transformer in a measurement mode to obtain harmonic current. Preferably, the feature fusion processing is performed on the multi-mode data according to the mode weight and a preset cascade fusion network to obtain a fusion feature vector, which includes: performing kernel density estimation calculation according to the multi-modal data to obtain modal weights; And performing cascade feature fusion calculation based on a preset cascade fusion network according to the modal weight and the multi-modal data to obtain a fusion feature vector. Preferably, the constructing a spatial impedance distribution equation according to the fusion eigenvector and performing electromagnetic-thermal coupling effect iterative analysis to obtain an impedance distribution matrix includes: Carrying out joule heating effect analysis and harmonic excitation analysis according to the temperature field and harmonic current in the fusion characteristic vector, and generating a space impedance distribution equation; And performing electromagnetic-thermal coupling effect iterative analysis based on a preset time step according to the space impedance distribution equation to obtain an impedance distribution matrix. Preferably, the performing defect detection analysis based on the constraint reinforcement learning strategy and the preset iron core diagram model according to the fusion feature vector and the impedance distribution matrix to obtain a defect evaluation level includes: Constructing an iron core three-dimensional grid topological graph for the transformer to obtain a preset iron core graph model; inputting the fusion feature vector, the impedance distribution matrix and the