Deep learning for predictive maintenance of rolling bearings

Abstract

[en] The monitoring of machine health has become of great importance in the industry in the recent years. Unexpected equipment failures can lead to catastrophic consequences, such as production downtime and costly equipment replacement. Rolling bearings are one of the most delicate components of rotating equipment, being a common cause of machine failures. For this reason, predictive maintenance techniques of rolling bearings are fundamental to preserve the health of a machine. In this project, we present a deep learning approach to predict bearing failures in their early development. All methodologies are data-driven, therefore they do not assume any expert knowledge on the field nor require any information about the equipment’s operating conditions. For this reason, this approach is versatile and can be used to diagnose multiple machines.