Abstract:

In the area of e-Maintenance the analysis of condition data plays an important role in order to determine the health condition of machines. The results of such an analysis are used to decide whether maintenance actions for a machine have to be scheduled or not. Predictive Maintenance is the next step as it offers the possibility to prognosticate the remaining time until a maintenance action of a machine has to be scheduled. Unfortunately, current solutions are only suitable for very specific use cases like reliability predictions based on vibration monitoring. Furthermore, they do not consider the fact that a machine may deteriorate non-uniformly, depending on external impacts (e.g., the work piece material in a milling machine, the changing fruit acid concentration in a bottling plant). These two problems are addressed in this paper. Therefore, concepts for a generic determination of reliability indicators and the handling of aging variability are presented.