Fatigue damage diagnostics and prognostics of composites utilizing structural health monitoring data and stochastic processes

Abstract

The procedure of damage accumulation in composites, especially during fatigue loading, is a complex phenomenon of stochastic nature which depends on a number of parameters such as type and frequency of loading, stacking sequence, material properties, and so on. Toward condition-based health monitoring and decision making, the need for not only diagnostic but also prognostic tools rises and draws increasing attention in the last few years. To this direction, we model the damage evolution in composites as a doubly stochastic hidden Markov process that manifests itself via structural health monitoring observations, that is, acoustic emission data. The damage process is modeled via an extension of the classic hidden Markov models to account for nonhomogeneity, that is, age dependence in state transitions. The observations come from acoustic emission data recorded throughout fatigue testing of open-hole carbon–epoxy coupons. A procedure that utilizes multiple observation sequences from a training dataset and estimates in a maximum likelihood sense the optimal model parameters is presented and applied in unseen data via a cross-validation rationale. Diagnostics of the most likely health state determination, average degradation level, and prognostics of the remaining useful life are among the capabilities of the presented stochastic model.

Bayesian Estimation of Remaining Useful Life for Wind Turbine Blades

“…For example, Peng et al [9] used in situ stiffness measurements to estimate RUL of an open hole specimen using a Bayesian approach, and Eleftheroglou et al [10] used acoustic emission data to update a semi-Markov model.…”

Structural health monitoring data fusion for in-situ life prognosis of composite structures

“…Based on the available training SHM features the parameters θ of the mathematical model, that is utilized to provide with the RUL predictions, are estimated (Section 2.3). In the present study we utilize a stochastic multi-state degradation model, such as NHHSMM, based on its successful implementation in our previous investigations [13,14]. Three different NHHSMMs are trained after using features extracted from the training set of AE, DIC and fused AE & DIC.…”

“…In this paper, the feature extraction process is based on monotonicity since a feature that is sensitive to the degradation process is desirable to have a monotonic trend [25,26,28]. Prognosability is excluded from the present feature extraction process since NHHSMM dictates that the last observation of the monitoring data must be unique and common for all the degradation histories [13,29,30]. The feature extraction process does not take into account the influence of trendability since the target of this work is to identify monotonicity's influence in prognostics.…”

“…Notched carbon/epoxy composite specimens under fatigue loading were used. Eleftheroglou and Loutas [13] and Eleftheroglou et al [14] proposed the use of a multi-state degradation model, i.e. the Non-Homogenous Hidden Semi Markov Model (NHHSMM), for the in-situ prognostics of open hole carbon/ epoxy specimens under fatigue loading.…”

“…the Non-Homogenous Hidden Semi Markov Model (NHHSMM), for the in-situ prognostics of open hole carbon/ epoxy specimens under fatigue loading. They used AE [13] or strain measurements [14] to estimate the parameters of the NHHSMM and successfully used it to obtain RUL estimates in unseen data with uncertainty quantification. In Loutas et al [15] two data-driven prognostic models, i.e.…”