Dynamic reliability assessment and prognostics with monitored data for multiple dependent degradation components

The traditional risk assessment and prognostics take into account data from a population of similar equipment, and, thus, the derived models give always a somewhat average performance of an equipment, not capturing the specific behavior of the individual equipment of interest. By taking into account the monitored data on the individual equipment, the risk assessment and prognostics can be made dynamic and the results improved. In this paper, we consider the dynamic risk assessment and prognostics of multiple, dependent, degrading components. As practical reference, the system considered is composed of two dependent components: a pump, whose degradation is modelled by a multi-state piecewise deterministic Markov process and a valve, whose continuous degradation process is described by a physical model. The monitored data are supposed to be noisy. As far as the authors know, this is the first time that the continuous degradation process, dependencies among components and observation noise are considered in a dynamic risk assessment and prognostics framework. The results on a case study show the effectiveness of the proposed method.