Prognostics of composite structures utilizing structural health monitoring data fusion

Abstract

A new structural health monitoring (SHM) data fusion methodology is proposed in order to produce features with strong prognostic capability. The Non-Homogenous Hidden Semi Markov model (NHHSMM) is utilized to estimate the remaining useful life (RUL) of composite structures using conventional as well as fused SHM data. The proposed data fusion methodology and NHHSMM are applied to open hole carbon/epoxy specimens under fatigue loading. During the specimens’ lifetime, acoustic emission (AE) and digital image correlation (DIC) techniques were used in order to capture the inherently dynamic and multi-scale character of damage. This work investigates the prognostic performance of the fused data in comparison to the performance of the conventional AE and DIC data. Prognostic performance metrics were used and as it was found fused data has potential to provide better prognostics.