Título: On the use of Machine Learning for predictive maintenance of power transformers
Autores: Carla Pacheco, Vagner Paes, Marcelo de Carvalho, Felipe Lopes, Gabriel Machado, Alex Garcia, Edmilson Neto, Jefferson Santos, Edward Haeusler, Ana Marotti
Resumo: This paper focuses on the use of machine learning algorithms to assist predictive maintenance aiming at reducing downtime and maintenance costs associated with power transformers. The paper presents two ML predictive indicators, Chromatographic Assay Indicator (CAI) and Electrical Failure Risk Indicator (EFRI), which use chromatographic and sensors data, respectively. The CAI evaluation showed a significant improvement in predicting failures compared with classical methods, whereas the EFRI tests showed it can be helpful to help maintenance team in identifying potential problems. The proposed solution integrates classical chromatographic analysis techniques with these ML indicators and aims at supporting maintenance specialists in decision-making processes, leading to more efficient maintenance management and reduced costs associated with equipment downtime
Palavras-chave: predictive maintenance, machine learning, power transformers, random forest algorithm, energy transmission
Páginas: 8
Código DOI: 10.21528/CBIC2023-031
Artigo em pdf: CBIC_2023_paper031.pdf
Arquivo BibTeX: CBIC_2023_031.bib