Failure Prediction from Condition Monitoring of Complex Systems

• Activity Number: 370
• Type: Invited
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Risk Analysis
• Author(s): Bo Henry Lindqvist*+ and Gunnhild Hardersen Presthus
• Companies: Norwegian University of Science and Technology and Norwegian University of Science and Technology
• Keywords: Wiener process ; condition based maintenance ; change point estimation

Abstract

We consider a technical system subjected to condition monitoring by a marker process Y(t). Failure of the system is closely connected to the event that the process Y(t) crosses a certain critical threshold. The monitored process Y(t) depends probabilistically on the state of a latent process S(t), representing the underlying technical condition of the system. The goal is to, based on the observation of the process Y(t), estimate the distribution of the first passage time of Y(t) of the critical threshold. The process Y(t) is modeled as a piecewise Wiener process with change points determined by the latent process S(t). A Bayesian approach involving Markov Chain Monte Carlo simulations is used for estimation.