École Nationale Supérieure de
Techniques Avancées

Résumé

Machine health monitoring refers to techniques for assessing the condition of rotating machinery. The goal is to extend the lifespan of components and reduce maintenance costs by implementing predictive maintenance. Deep Learning emerges as an effective method for extracting useful representations from the vibrational signals of bearings. These representations are reused in classification tasks to detect defaults and to estimate the remaining useful life using statistical models. The objective of this internship is to explore selfsupervised learning methods and determine if they can balance out for the lack of labeled data by outperforming supervised models. Moreover, this internship aims to establish a viable method for predicting the remaining useful life of a bearing.