Rolling Bearing Diagnosis using Cyclostationary tools and neural networks

Abstract

Rotating machines are ubiquitous in the context of industrial control; early detection of their mechanical defects plays an important role for productivity and economic high gain. These defects are currently characterized by randomness and hidden periodicities. We propose in this regard is the study of the cyclical aspect of these defects by studying statistical tools in areas such as coherence spectrum. Applications on bearings are designed to show the impact of these tools.