This work focuses on an unsupervised, data driven statistical approach to detect and monitor fatigue crack growth in lug
joint samples using surface mounted piezoelectric sensors. Early and faithful detection of fatigue cracks in a lug joint can
guide in taking preventive measures, thus avoiding any possible fatal structural failure. The on-line damage state at any
given fatigue cycle is estimated using a damage index approach as the dynamical properties of a structure change with
the initiation of a new crack or the growth of an existing crack. Using the measurements performed on an intact lug joint
as baseline, damage indices are evaluated from the frequency response of the lug joint with an unknown damage state.
As the damage indices are evaluated, a Bayesian analysis is committed and a statistical metric is evaluated to identify
damage state(say crack length).