Amélioration du diagnostic des défauts des transformateurs électriques par la fusion d'algorithmes de machine learning utilisant les données AGD

Abstract

This work focuses on the diagnosis of power transformers, which are key components in the stability and performance of electrical networks. These transformers, operating according to Faraday’s principles, enable the transmission of energy over long distances. Their monitoring is therefore crucial to prevent serious failures. The study is primarily based on Dissolved Gas Analysis (DGA) in insulating oil, which is an effective method for detecting internal faults in power transformers. In this work, a modern approach based on artificial intelligence, particularly machine learning, was explored. This method enables algorithms to learn from data in order to predict the condition of the oil and detect potential faults.The results show that integrating machine learning significantly improves diagnostic accuracy, enabling more reliable predictive maintenance and better management of electrical equipment.