Deep Learning Neural Networks Diagnosis of Power Transformer through Its DGA Data

Hansel Matthew, Aqila Dzikra Ayu, Iman Herlambang Suherman, Aries Subiantoro, Benyamin Kusumoputro

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Dissolved gas analysis (DGA) is the most common method used to diagnose faults in thermal power plant transformers by monitoring the concentration of various gas in transformer oil and using it to interpret the type of fault. In this study, neural networks using SGD and Adam optimizers are proposed to classify and predict electrical faults based on DGA analysis data due to the complicated interpretation of the data. Those data are obtained from IEC TC 10 databases combined with the data from a Jakarta Government's Steam Power Plant. The network designed produced a reliable fault detector for power transformer condition diagnosis.

Original languageEnglish
Title of host publicationProceedings - 9th International Conference on Electrical Engineering, Computer Science and Informatics, EECSI 2022
EditorsMochammad Facta, Mohammad Syafrullah, Munawar Agus Riyadi, Imam Much Ibnu Subroto, Irawan Irawan
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages292-296
Number of pages5
ISBN (Electronic)9786239213558
DOIs
Publication statusPublished - 2022
Event9th International Conference on Electrical Engineering, Computer Science and Informatics, EECSI 2022 - Jakarta, Indonesia
Duration: 6 Oct 20227 Oct 2022

Publication series

NameInternational Conference on Electrical Engineering, Computer Science and Informatics (EECSI)
Volume2022-October
ISSN (Print)2407-439X

Conference

Conference9th International Conference on Electrical Engineering, Computer Science and Informatics, EECSI 2022
Country/TerritoryIndonesia
CityJakarta
Period6/10/227/10/22

Keywords

  • fault detection
  • neural network
  • power transformers

Fingerprint

Dive into the research topics of 'Deep Learning Neural Networks Diagnosis of Power Transformer through Its DGA Data'. Together they form a unique fingerprint.

Cite this