Diagnostics de l’état des transformateurs de puissance par l’application de l’apprentissage automatique

Abstract

The power transformer operates on Faraday's principles, facilitating the transfer of energy over long distances. Its maintenance and diagnosis are crucial for the stability of the electrical grid. In this study, transformer diagnostics were approached using oil analysis methods, including dissolved gas analysis. Two analytical approaches were employed: the traditional method adhering to international standards such as CEI, and the use of artificial intelligence. The latter relies on machine learning algorithms, which have proven effective in predicting oil condition.