Power transformer voiceprint anomaly detection method and system

By applying conditions based on hidden vectors and reconstruction features in power transformers to generate an adversarial network, abnormal detection and quantitative risk assessment of transformer voiceprints are realized, and the problem of difficulty in detecting mechanical abnormal states in a timely manner is solved in the existing technology, and the accuracy of detection is improved.

CN119673204APending Publication Date: 2025-03-21STATE GRID INTELLIGENCE TECHNOLOGY CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411647075.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-18
Publication Date
2025-03-21

AI Technical Summary

Technical Problem

The existing non-break detection methods are difficult to detect mechanical abnormalities in power transformers in a timely manner, such as DC bias, core loosening, winding loosening and winding deformation, resulting in potential defects not being discovered in time, which may cause major accidents.

Method used

The adversarial network is generated based on the conditions of hidden vectors and reconstruction features, and the risk value is calculated by reconstructing the loss and abnormality scores, and abnormality detection and quantitative risk assessment of the transformer voiceprint are realized.

Benefits of technology

The detection of abnormal signals is achieved under the condition of few samples, which improves the accuracy of voiceprint abnormal detection, and solves the problems of low accuracy and insufficient generalization of deep learning models when there are insufficient fault samples.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119673204A_ABST
    Figure CN119673204A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of equipment defect diagnosis, and provides a voiceprint anomaly detection method and system for a power transformer, and the method comprises the steps: obtaining a risk value of a to-be-detected transformer through a reasoning network based on the input characteristics of the to-be-detected transformer, and judging that the to-be-detected transformer is abnormal when the risk value of the to-be-detected transformer exceeds a risk threshold value; wherein the reasoning network comprises a generation encoder, and a generation decoder and an abnormal score calculation network which are respectively connected with the generation encoder; the generation encoder processes the input characteristics of the transformer to be detected to obtain implicit vectors; the generation decoder processes the implicit vector to obtain a reconstruction feature; on the basis of the implicit vectors and the reconstruction features, reconstruction loss is obtained through calculation; the abnormal score calculation network processes the implicit vector to obtain an abnormal score; the risk value is the weighted sum of the reconstruction loss and the abnormal score. And the accuracy of voiceprint anomaly detection is effectively improved.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Cited By

  • Wind power blade fault detection method and system based on audio potential characterization

    CN120071968A