Online remaining useful life prognosis for composite materials based on acoustic emission and strain data

Abstract

The procedure of fatigue damage accumulation in composite structures, is a complex phenomenon due to the multiphase nature of composites, the variation of inherent manufacturing defects, the randomness of loads, the stochastic activation of different damage mechanisms and an incomplete knowledge about the physics behind the evolution and interaction of damage mechanisms. In order to develop a robust model, which will reliably estimate the remaining useful life of the structure, we hypothesize that the damage accumulation process is of a stochastic nature and we propose a framework that combines a stochastic model with structural health monitoring data. In this study, the Non-Homogenous Hidden Semi Markov model and acoustic emission and strain data extracted from fatigue tests of open-hole quasi-isotropic carbon fibre reinforced polymers are selected. The damage accumulation is a hidden process and the goal of the framework is to correlate the structural health monitoring data with the parameters of the model and identify the hidden relationship. In conclusion, the remaining useful life estimations of each structural health monitoring technique are compared with the actual remaining useful life and based on performance metrics it was found that the strain data provided more accurate predictions.