Condition monitoring and prognostics of bearings considering uncertainties

Abstract

Rotating systems represent a class of machines with great application in industry. Among the main component of rotating systems, bearing elements are the most susceptible to faults. Many techniques that have been developed for condition monitoring and prognostics of bearings use model-based responses. In this context, the development of robust and representative models for each component of the rotating system becomes crucial, attending the new tendencies demanded by Industry 4.0. Furthermore, faults in bearing are impacted by considerable stochastic fluctuation of the machine operation conditions and the material properties. Therefore, it is also imperative to consider probabilistic and uncertainty quantification approaches in order to develop robust condition monitoring and prognostics strategies of bearing elements. From these motivations, the research question proposed for this work is: Can condition monitoring and prognostics strategies of bearings be improved by using dynamic modeling and uncertainty quantification approaches? In order to respond the research question, three objectives were developed, namely, (1) to develop realistic mathematical models to describe the most common fault mechanisms and the impact on dynamic response of bearings elements, (2) to develop a robust probabilistic framework to account for the many sources of uncertainty on the bearing operational parameters and material properties, and (3) to develop advanced strategies of condition monitoring and prognostics of bearings elements. (AU)