Health-Informed Uncertainty Quantifications via Bayesian Filters with Markov Chain Monte Carlo Simulations for Fatigue Critical Rotorcraft Components

Abstract

This paper presents the applications of Bayesian-filters (BF) with Markov chain Monte Carlo (MCMC) simulations for probabilistic lifing assessment of aircraft fatigue critical components. Uncertainties in damage growth parameters are updated with new information obtained through structuralhealth monitoring (SHM) systems and the remaining useful life (RUL) are predicted. State transition function representing virtual damage growth of a component and measurement function representing the SHM measurements of the component are defined. State transition function is described by a typical Paris equation for fatigue crack propagation. In the equation, the initial crack size and crack growth rate are updated by incoming SHM measurements. Measurement functions are assumed in this study which describe the relationship between the damage features derived from SHM signals and the damage sizes. Damage tolerance (DT) and risk-based remaining useful life of fatigue critical structural components are determined at various reliability levels. The variabilities of RUL are also quantified for various magnitudes of random measurement noise and various measurement frequencies. It is found that the variability of the RUL is proportional to that of the measurement noise. In addition, more frequent measurements will result in less variability in RUL.