Random vibration based robust structural health monitoring for a composite wing-shaped structure under temperature uncertainty: an exploratory study

The problem of random vibration strain based early-stage damage detection and characterization on the FEM of a composite wing-shaped panel under Temperature uncertainty is explored within a Multiple Model (MM) framework. In this context three robust data-based damage diagnosis methods, all utilizing a single sensor, are explored. A parametric AutoRegressive model based (the MM-AR method), a Principal Component Analysis enhanced version (the MM-PCA-AR method), and a non-parametric Power Spectral Density based (the MM-PSD method). Three damage types are considered: Delamination, interface debonding, and through-the-thickness indentation. The results of the study based on hundreds of Monte Carlo Simulations indicate effective damage detection, primarily by the MM-AR and MM-PCA-AR methods, with a 96.1% correct detection rate for a 5% false alarm, and very good damage characterization, mainly by the MM-PCA-AR method, with an overall accuracy reaching 95.4%.