Search

CN-121995170-A - Transformer discharge detection method and system with anti-interference function

CN121995170ACN 121995170 ACN121995170 ACN 121995170ACN-121995170-A

Abstract

The invention belongs to the technical field of power equipment state monitoring, and discloses a transformer discharge detection method and system with an anti-interference function, wherein the method comprises the steps of obtaining a transformer multi-mode discharge original signal, extracting initial multi-dimensional characteristics through channel synchronization coefficients and anchoring synchronous interference signals, and carrying out weighted average fusion on the initial multi-dimensional characteristics according to the channel synchronization coefficients to obtain an anti-interference characteristic vector; the method comprises the steps of constructing a feature screening-interference immunity lightweight neural network model, inputting anti-interference features, taking a channel synchronization coefficient as an interference immunity factor, outputting a discharge type identification result, combining peak amplitude values of a transformer multi-mode discharge original signal after noise elimination pretreatment to obtain an evaluation index, and obtaining a discharge risk grade based on the evaluation index. The invention realizes accurate identification and risk level evaluation of transformer discharge and solves the problems of poor anti-interference performance and the like.

Inventors

  • SHANG XIANFEI
  • ZHANG HAIJING

Assignees

  • 鑫大变压器有限公司

Dates

Publication Date
20260508
Application Date
20260128

Claims (10)

  1. 1. The transformer discharge detection method with the anti-interference function is characterized by comprising the following steps of: acquiring a multi-mode discharge original signal of the transformer, and carrying out noise elimination pretreatment on the multi-mode discharge original signal of the transformer by calculating a channel synchronization coefficient and anchoring a synchronous interference signal; based on the transformer multi-mode discharge original signal after noise elimination pretreatment, extracting initial multi-dimensional characteristics, and carrying out weighted average fusion on the initial multi-dimensional characteristics according to channel synchronization coefficients to obtain an anti-interference characteristic vector; Constructing a lightweight neural network model, adding an interference immunity layer to obtain a feature screening-interference immunity lightweight neural network model, inputting an anti-interference feature vector, taking a channel synchronization coefficient as an interference immunity factor, and outputting a discharge type identification result; according to the discharge type identification result, the evaluation index is obtained by combining the peak amplitude of the transformer multi-mode discharge original signal after the noise elimination pretreatment, and the discharge risk level is obtained based on the evaluation index.
  2. 2. The method for detecting the transformer discharge with the anti-interference function according to claim 1, wherein the transformer multi-mode discharge original signal comprises an electromagnetic pulse signal, a vibration signal and a transient voltage signal, the noise elimination preprocessing is specifically that based on the transformer multi-mode discharge original signal, channel synchronization coefficients of three channel signals are calculated firstly, synchronous interference signals are anchored according to the channel synchronization coefficients, then a channel with the weakest interference signal strength is selected as a reference channel, and differential operation is carried out on signals of the other two channels and the reference channel signal to obtain the preprocessed transformer multi-mode discharge original signal.
  3. 3. The method for detecting discharge of a transformer with anti-interference function as claimed in claim 2, wherein the channel synchronization coefficient is calculated by: ; Wherein, the For the amplitude of the electromagnetic pulse signal, For the amplitude of the vibration signal, For the amplitude of the transient voltage signal, Is the average value of the amplitude of the electromagnetic pulse signal at all moments, Is the average value of the amplitude of the vibration signal at all times, Is the average value of the transient voltage signal amplitude at all moments, In order to sample the total length of time, Is the sampling instant.
  4. 4. The method for detecting discharge of a transformer with anti-interference function according to claim 1, wherein the weighted average fusion of the initial multidimensional features according to the channel synchronization coefficient is expressed as: wherein, the method comprises the steps of, Is the first The dimensions of the features are fused together, Is the first The channel synchronization coefficients corresponding to the dimensional characteristics, Is the first -Maintaining an initial multidimensional feature; And arranging all the fusion features in a fixed sequence to obtain an anti-interference feature vector.
  5. 5. The transformer discharge detection method with the anti-interference function according to claim 1, wherein the feature screening-interference immunity lightweight neural network model sequentially comprises an input layer, a convolution layer, an interference immunity layer and an output layer, wherein anti-interference feature vectors are input into the input layer, normalized and output to the convolution layer, local key features are extracted, the convolution layer is accessed into an activation function, pooling operation is performed, the convolution feature vectors are output to the interference immunity layer, channel synchronization coefficients are used as interference immunity factors, weighting correction is performed on the convolution feature vectors, feature vectors after interference immunity are output to the output layer, and the output layer adopts a full-connection layer structure and outputs a discharge type identification result.
  6. 6. The method for detecting discharge of transformer with anti-interference function as claimed in claim 5, wherein the formula for performing weighted correction on the convolution eigenvector is as follows: wherein, the method comprises the steps of, In order to interfere with the post-immunization characteristics, In order to convolve the feature vector, Is a channel synchronization coefficient.
  7. 7. The method for detecting discharge of a transformer with anti-interference function according to claim 1, wherein the evaluation index has a calculation formula as follows: wherein, the method comprises the steps of, In order to evaluate the index of the light, In order to normalize the peak amplitude of the original multi-mode discharge signal of the transformer after the noise elimination pretreatment, Is a discharge type weighting coefficient.
  8. 8. A transformer discharge detection system with anti-interference function, comprising: The data acquisition and preprocessing module is configured to acquire a multi-mode discharge original signal of the transformer, and perform noise elimination preprocessing on the multi-mode discharge original signal of the transformer by calculating a channel synchronization coefficient and anchoring a synchronous interference signal; the characteristic extraction module is configured to extract initial multidimensional characteristics based on the transformer multi-mode discharge original signals subjected to noise elimination pretreatment, and perform weighted average fusion on the initial multidimensional characteristics according to the channel synchronization coefficients to obtain anti-interference characteristic vectors; The identification detection module is configured to construct a lightweight neural network model, add an interference immunity layer, obtain a feature screening-interference immunity lightweight neural network model, input an anti-interference feature vector, take a channel synchronization coefficient as an interference immunity factor and output a discharge type identification result; the evaluation module is configured to obtain an evaluation index according to the discharge type identification result and the peak amplitude of the transformer multi-mode discharge original signal after the noise elimination pretreatment, and obtain a discharge risk level based on the evaluation index.
  9. 9. An electronic device comprising a memory and a processor, and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform a method of detecting a transformer discharge with anti-tamper function as claimed in any one of claims 1 to 7.
  10. 10. A computer readable storage medium storing computer instructions which, when executed by a processor, perform a transformer discharge detection method with anti-interference function as claimed in any one of claims 1 to 7.

Description

Transformer discharge detection method and system with anti-interference function Technical Field The invention relates to the technical field of power equipment state monitoring, in particular to a transformer discharge detection method and system with an anti-interference function. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. The transformer is the core equipment of the power system, and the running state of the transformer is directly related to the safe and stable running of the power grid. Partial discharge is an important precursor of transformer insulation degradation, timely and accurately detects the type of partial discharge of a transformer and evaluates the severity of discharge, and can effectively prevent transformer faults and reduce the operation and maintenance cost of a power system, so that the detection technology of the partial discharge of the transformer is always the research focus in the field of power equipment state monitoring. In the prior art, a detection scheme for a multi-mode discharge signal of a transformer generally utilizes UHF, AE, TEV multi-channel sensors to collect multi-mode original signals in the operation process of the transformer, then adopts traditional algorithms such as filtering, wavelet denoising and the like to perform denoising pretreatment on the original signals to remove environmental interference and self noise of equipment, then extracts characteristic parameters such as peak value, kurtosis, pulse width and the like of the denoised signals, and finally inputs the extracted characteristics into a machine learning model such as a convolutional neural network and the like to realize discharge type identification and severity assessment. However, the research finds that the prior art has a plurality of defects in practical application, and the defects are expressed as follows: The existing denoising algorithm lacks pertinence, only can remove noise in a fixed frequency band, and synchronous interference signals similar to the characteristics of discharge signals cannot be effectively identified and filtered, so that a large amount of interference components still remain in the denoised signals, and the accuracy of subsequent feature extraction is affected; in the prior art, in the stage of feature fusion, the characteristics corresponding to interference signals are misled on the identification result due to simple splicing of the multichannel characteristics, so that the reliability of discharge type identification and severity assessment is further reduced; The characteristic extraction and model identification links in the prior art are mutually independent, relevant parameters of interference characteristics are not reserved in the noise elimination process, so that the model cannot adaptively adjust the identification strategy according to the interference intensity, the identification accuracy is greatly reduced when facing complex interference scenes, the existing identification network model is complex in structure, large in parameter quantity and long in training period, the characteristics of the multi-mode discharge signals are not subjected to light weight design, and the requirements of on-site real-time detection of the power equipment are difficult to meet. Disclosure of Invention In order to solve the problems, the invention provides a transformer discharge detection method with an anti-interference function, which realizes accurate identification and risk level assessment of transformer discharge through synchronous interference anchoring noise elimination, weighting feature fusion and interference immunity lightweight model, and solves the problems of poor anti-interference performance and the like. In order to achieve the above purpose, the present invention adopts the following technical scheme: in a first aspect, the present invention provides a method for detecting discharge of a transformer with an anti-interference function, including the steps of: acquiring a multi-mode discharge original signal of the transformer, and carrying out noise elimination pretreatment on the multi-mode discharge original signal of the transformer by calculating a channel synchronization coefficient and anchoring a synchronous interference signal; based on the transformer multi-mode discharge original signal after noise elimination pretreatment, extracting initial multi-dimensional characteristics, and carrying out weighted average fusion on the initial multi-dimensional characteristics according to channel synchronization coefficients to obtain an anti-interference characteristic vector; Constructing a lightweight neural network model, adding an interference immunity layer to obtain a feature screening-interference immunity lightweight neural network model, inputting an anti-interference feature vector, taking a channel synchronization co