From monitoring data to remaining useful life : an evolving approach including uncertainty.

International audienceAlthough prognostic activity is nowadays recognized as a key feature in maintenance strategies, real prognostic systems are scarce in industry. That can be explained from different aspects, one of them being the lack of knowledge on the monitored system that impedes the development of classical dependability analysis (based on statistical data for example). Within this frame, the general purpose of the work is to propose a prognostic system that starts from monitoring data and goes through provisional reliability and remaining useful life by characterizing the uncertainty following from the degradation process. More precisely, the paper emphasizes on the development of an evolving neuro-fuzzy predictor that, not only "gives" an approximation of the degradation of an equipment but also associates to it a confidence measure