Exponential degradation model for Remaining Useful Life estimation of electrolytic capacitors

Abstract

Nowadays, Remaining Useful Life (RUL) is becoming an important and critical parameter in RAMS (Reliability, Availability, Maintainability and Safety) engineering. In the prognostic research field, much attention is usually given to mechanical device which are characterized by rapid aging processes. However, it is also fundamental to study prognostic and health management of electric and electronic equipment. In this context, electrolytic capacitors have a critical role in reliability of power electronics. Thus, in this work, an exponential degradation model has been used in order to implement a remaining useful life prediction technique for capacitors. A publicly available prognostic dataset composed by six capacitors tested under accelerated conditions has been taken into account. The procedure has been implemented using five capacitors for training and the remaining one for validation. The results of the implemented algorithm allow to estimate with a great advance when the capacitor under test will reach a predetermined failure threshold.