Random Vibration Based Robust Damage Detection for a Composite Aerostructure Under Assembly-Induced Uncertainty

The problem of random vibration based robust damage detection for a composite aerostructure under assembly-induced uncertainty is considered. The focus is on the exploration of the performance limits and critical assessment of two unsupervised Statistical Time Series (STS) type robust methods: a Multiple Model (MM) based method and a Principal Component Analysis (PCA) based counterpart. Three progressive damage scenarios are considered, the effects of which on the structural dynamics are ‘minor’ and almost completely ‘masked’ by assembly-induced uncertainty. The assessment, based on hundreds of experiments, suggests that damage as small as 10% reduction in the tightening torque of a single bolt is detectable in a reliable way. Both methods achieve remarkably good detection performance characterized by 100% correct detection for false alarm rates of 3.5% or higher, with the MM type method exhibiting a slight edge in performance over its PCA based counterpart! Overall the unsupervised STS type robust methods are shown to effectively detect ‘minor’ damage in the presence of significant assembly-induced uncertainty!