Transformer state analysis method and system based on deep learning

By using deep learning methods to automatically process transformer characteristic gas data, the problem of insufficient feature extraction capability in transformer condition analysis is solved, enabling accurate early fault identification and rapid response, thus meeting the intelligent operation and maintenance needs of power systems.

CN122432843APending Publication Date: 2026-07-21INNER MONGOLIA ELECTRIC POWER (GRP) CO LTD XILIN GOL ULTRA-HIGH VOLTAGE POWER SUPPLY BRANCH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
INNER MONGOLIA ELECTRIC POWER (GRP) CO LTD XILIN GOL ULTRA-HIGH VOLTAGE POWER SUPPLY BRANCH
Filing Date
2026-04-24
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing technologies lack the ability to process characteristic gas concentrations in depth and automatically in transformer condition analysis, which causes early and weak fault characteristics to be masked by noise, making it difficult to achieve accurate fault identification and rapid response, and failing to meet the intelligent and refined operation and maintenance needs of power systems.

Method used

A deep learning-based transformer state analysis method is adopted, which uses convolutional neural networks to automate the processing of characteristic gas data, including data preprocessing, convolutional layer feature extraction, pooling layer dimensionality reduction, fully connected layer combination, and backpropagation optimization, to form a trained state analysis model and realize the automated identification of transformer state.

Benefits of technology

It improves the accuracy and anti-interference capability of transformer condition identification, realizes the automation and efficiency of the analysis process, avoids missed judgments and misjudgments, and adapts to the intelligent operation and maintenance needs of diverse power production scenarios such as urban core substations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122432843A_ABST
    Figure CN122432843A_ABST
Patent Text Reader

Abstract

The application provides a transformer state analysis method and system based on deep learning, and relates to the technical field of power equipment fault diagnosis.The method comprises the following steps: acquiring characteristic gas content data dissolved in transformer oil, and constructing an initial feature data set; performing linear normalization mapping processing on each gas content value in the initial feature data set through a data preprocessing unit to obtain a standardized feature vector sequence; sequentially performing local feature convolution operation on the standardized feature vector sequence through a convolution layer of a preset convolutional neural network model, and then performing down-sampling dimension reduction operation through a pooling layer to obtain a high-dimensional feature mapping; performing non-linear feature combination operation on the high-dimensional feature mapping through a fully connected layer of the preset convolutional neural network model, and then transmitting the high-dimensional feature mapping from the fully connected layer to an output layer to obtain a preliminary state discrimination result.The application can improve the accuracy and efficiency of transformer state discrimination, and adapt to intelligent and refined operation and maintenance requirements of a power system.
Need to check novelty before this filing date? Find Prior Art