The health monitoring and damage prognosis of aerospace hotspots is important for reducing maintenance costs and increasing inservice capacity of aging aircraft. One of the leading causes of structural failure in aerospace vehicles is fatigue damage. Based on the physical mechanism of damage nucleation and growth, a physics-based multiscale model is considered for fatigue damage assessment in metallic aircraft structures. A guided-wave based sensing approach is utilised to enable effective damage detection in a common structural hotspot: a lug joint. Finite element analysis is carried out with piezoelectric wafers bonded to the host structure and the simulated sensor signals are analysed. A damage classification strategy is developed, which integrates physically motivated time-frequency approaches with advanced stochastic modelling techniques. In particular, a variational Bayesian learning scheme is used to estimate the optimal model complexity automatically from the data, adapting the classifier for real-time use. Classification performance is studied as a function of signal-to-noise ratio and results are reported for the detection of fatigue crack damage in the lug joint. An adaptive hybrid prognosis model is proposed, which estimates the residual useful life of structural hotspots using damage condition information obtained in real-time.
ASJC Scopus subject areas
- Aerospace Engineering