This study focused on the in-situ fatigue damage assessment of open-hole carbon/epoxy coupons using Acoustic Emission (AE) and Digital Image Correlation (DIC) techniques. Constant amplitude fatigue tests were performed and the main objective was to investigate the damage process, the degradation process of the fatigue modules and to identify features, derived from the experimental data, that can be used as sensitive-to-damage indexes. To this end, the two experimental techniques were reviewed as potential online monitoring tools for the fatigue damage assessment.@en

The procedure of damage accumulation in composite materials, especially during fatigue loading, is a complex phenomenon which depends on a number of parameters such as ply orientation, material properties, geometrical non-linearities etc. Towards condition based health monitoring and decision making, the need not only for diagnostic but also for prognostic tools rises and draws increasing attention the last few years. The damage process is in general hidden and manifests itself through in-situ structural health monitoring (SHM) data. Due to the hidden nature of the damage accumulation, non-homogenous hidden Semi Markov process (NHHSMP) seems to be a suitable candidate for describing adequately the aforementioned system’s degradation in time. Its non-homogeneous aspect takes into account the system’s ageing. Moreover, the sojourn times in each state are assumed to be generally distributed, not necessarily exponentially distributed, which is a more realistic assumption for real world engineering systems. The SHM observations are coming from acoustic emission (AE) data recorded throughout constant amplitude fatigue testing of open-hole carbon/epoxy coupons. The scatter of the cycles to failure reported is quite large, an expected result of the stochasticity in the material properties and material inhomogeneities. A maximum likelihood approach for the estimation of the model parameters is followed and useful diagnostic and prognostic measures such as the coupon's current degradation level as well as measures the coupon's remaining useful life (RUL) are proposed for the monitoring of structural integrity of composite materials.@en

A prognostic framework is proposed in order to estimate the remaining useful life of composite materials under fatigue loading based on acoustic emission data and a sophisticated Non Homogenous Hidden Semi Markov Model. Bayesian neural networks are also utilized as an alternative machine learning technique for the non-linear regression task. A comparison between the two algorithms operation, input, output and performance highlights their ability to tackle the prognostic task.@en

The present study utilizes a state-of-the-art stochastic modeling with structural health monitoring (SHM) data derived from strain measurements, in order to assess the remaining useful life (RUL) online in composite materials under fatigue loading. Non-Homogenous Hidden Semi Markov model (NHHSMM) is a suitable candidate with a rich mathematical structure capable of describing the composite’s multi-state damage evolution in time. The proposed model uses as input SHM data in the form of strain measurements obtained from the Digital Image Correlation (DIC) technique to a coupon-level constant amplitude fatigue test campaign. The obtained from the stochastic model RUL estimations are compared with the actual RUL and the effectiveness of the prognosis is discussed. @en

An innovative prognostic data-driven framework is proposed to deal with the real-time estimation of the remaining useful life of composite materials under fatigue loading based on acoustic emission data and a sophisticated multi-state degradation Non Homogeneous Hidden Semi Markov Model (NHHSMM). The acoustic emission data pre-processing to extract damage sensitive health indicators and the maximum likelihood estimation of the model parameters from the training set are discussed in detail. In parallel, a Bayesian version of a well-established machine learning technique i.e. neural networks, is utilized to approach the remaining useful life estimation as a non-linear regression task. A comparison between the two algorithms training, operation, input-output and performance, highlights their ability to offer reliable remaining useful life estimates conditional on health monitoring data from composite structures under service loading. Both approaches result in very good estimations of the mean remaining useful life of unseen data. NHHSMM is concluded as the preferable option as it provides much less volatile predictions and more importantly is characterized by confidence intervals which shorten as more data come into play, an essential trait of a robust prognostic algorithm. @en