Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear

Deep neural networks have proven to be highly efficient in fault detection and classification using partial discharges (PDs) in gas-insulated switchgear (GIS). However, previous studies have not fully addressed the issue of limited labeled training data, which significantly impacts the performance o...

Full description

Saved in:
Bibliographic Details
Main Authors: Ho Trong Tai, Young-Woo Youn, Hyeon-Soo Choi, Yong-Hwa Kim
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10945843/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849702129205248000
author Ho Trong Tai
Young-Woo Youn
Hyeon-Soo Choi
Yong-Hwa Kim
author_facet Ho Trong Tai
Young-Woo Youn
Hyeon-Soo Choi
Yong-Hwa Kim
author_sort Ho Trong Tai
collection DOAJ
description Deep neural networks have proven to be highly efficient in fault detection and classification using partial discharges (PDs) in gas-insulated switchgear (GIS). However, previous studies have not fully addressed the issue of limited labeled training data, which significantly impacts the performance of these models. Existing semi-supervised learning (SSL) approaches typically discard low-confidence pseudo-labels from unlabeled PD samples, because these unreliable labels can mislead the model. This gap in current research overlooks the potential value of low-confidence samples, which, despite their uncertain class, are unlikely to belong to the classes with the lowest probabilities. In this study, we aim to overcome this limitation by proposing a novel semi-supervised contrastive complementary learning (SCCL) method. Our SCCL approach generates a larger set of reliable negative pairs using complementary labels, allowing us to utilize the entire set of unlabeled PD samples effectively. We validate the feasibility of SCCL using phase-resolved PDs (PRPDs) and onsite noise data collected through an ultrahigh-frequency (UHF) PD measurement system. Experimental results indicate that the SCCL method achieves an impressive accuracy of 96.23% by effectively leveraging low-confidence unlabeled PD samples to improve classification performance in GIS under restricted labeling conditions.
format Article
id doaj-art-c1f2138a2beb459fb86bbeca4be4cc9e
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c1f2138a2beb459fb86bbeca4be4cc9e2025-08-20T03:17:44ZengIEEEIEEE Access2169-35362025-01-0113587225873410.1109/ACCESS.2025.355635310945843Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated SwitchgearHo Trong Tai0https://orcid.org/0009-0002-9089-5081Young-Woo Youn1https://orcid.org/0000-0002-4207-971XHyeon-Soo Choi2Yong-Hwa Kim3https://orcid.org/0000-0003-2183-5085Department of Computer Science and Information, Korea National University of Transportation, Uiwang-si, Gyeonggi-do, South KoreaSmart Grid Research Division, Korea Electrotechnology Research Institute, Gwangju-si, Republic of KoreaGenad System, Naju-si, Jeollanam-do, Republic of KoreaDepartment of Computer Science and Information, Korea National University of Transportation, Uiwang-si, Gyeonggi-do, South KoreaDeep neural networks have proven to be highly efficient in fault detection and classification using partial discharges (PDs) in gas-insulated switchgear (GIS). However, previous studies have not fully addressed the issue of limited labeled training data, which significantly impacts the performance of these models. Existing semi-supervised learning (SSL) approaches typically discard low-confidence pseudo-labels from unlabeled PD samples, because these unreliable labels can mislead the model. This gap in current research overlooks the potential value of low-confidence samples, which, despite their uncertain class, are unlikely to belong to the classes with the lowest probabilities. In this study, we aim to overcome this limitation by proposing a novel semi-supervised contrastive complementary learning (SCCL) method. Our SCCL approach generates a larger set of reliable negative pairs using complementary labels, allowing us to utilize the entire set of unlabeled PD samples effectively. We validate the feasibility of SCCL using phase-resolved PDs (PRPDs) and onsite noise data collected through an ultrahigh-frequency (UHF) PD measurement system. Experimental results indicate that the SCCL method achieves an impressive accuracy of 96.23% by effectively leveraging low-confidence unlabeled PD samples to improve classification performance in GIS under restricted labeling conditions.https://ieeexplore.ieee.org/document/10945843/Semi-supervised learning (SSL)complementary labelscontrastive learningfault diagnosisphase-resolved partial discharge (PRPD)gas-insulated switchgear (GIS)
spellingShingle Ho Trong Tai
Young-Woo Youn
Hyeon-Soo Choi
Yong-Hwa Kim
Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
IEEE Access
Semi-supervised learning (SSL)
complementary labels
contrastive learning
fault diagnosis
phase-resolved partial discharge (PRPD)
gas-insulated switchgear (GIS)
title Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
title_full Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
title_fullStr Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
title_full_unstemmed Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
title_short Partial Discharge Diagnosis Using Semi-Supervised Learning and Complementary Labels in Gas-Insulated Switchgear
title_sort partial discharge diagnosis using semi supervised learning and complementary labels in gas insulated switchgear
topic Semi-supervised learning (SSL)
complementary labels
contrastive learning
fault diagnosis
phase-resolved partial discharge (PRPD)
gas-insulated switchgear (GIS)
url https://ieeexplore.ieee.org/document/10945843/
work_keys_str_mv AT hotrongtai partialdischargediagnosisusingsemisupervisedlearningandcomplementarylabelsingasinsulatedswitchgear
AT youngwooyoun partialdischargediagnosisusingsemisupervisedlearningandcomplementarylabelsingasinsulatedswitchgear
AT hyeonsoochoi partialdischargediagnosisusingsemisupervisedlearningandcomplementarylabelsingasinsulatedswitchgear
AT yonghwakim partialdischargediagnosisusingsemisupervisedlearningandcomplementarylabelsingasinsulatedswitchgear