Uncertainty Analysis in Prognostics for Optimal Maintenance

Simone Turrin, Subanatarajan Subbiah

Abstract

Reliable prognostics have become indispensable to provide predictive service solutions. In fact, by assessing the residual useful life of a product, a condition-based maintenance strategy can be adopted. Consequently, the appropriate service activity can be assigned and scheduled only if and when required. Hence cost and time related to unnecessary preventive service tasks is saved. Prognostics algorithms are usually affected by different types of uncertainties stemming, for example, by randomness and lack of knowledge in the degradation behaviour, the inherent failure mechanism and the usage of the product by the customer. Such uncertainties can have potentially large effects on the determination of the residual useful life and, consequently, on the planning and scheduling of the subsequent service activity. This paper is focused on the customer usage as a main source of uncertainty. A statistical and a fuzzy approach are presented. The application case is given by the determination of the electrical overhaul for a generator circuit breaker (GCB). Nevertheless, the suggested approaches can be generally used for any type of products.

Download

IMEKO-TC10-2013-035.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings