Damage identification on a flat panel based on combined modal curvature indices

In this study, the experimental identification of damage to a metallic panel was performed us-ing methods based on modal curvature and modal strain energy analysis. The techniques considered require setting a reference (intact) configuration to which the damaged structure is compared. The type of damage an-alyzed during the tests consisted of thickness reduction on a limited portion of the plate. The calculation of modal strain energy requires the identification of structural modes based on accelerometer data obtained using the Roving Hammer Technique (RHT). Although the exploited damage identification techniques have been deeply analyzed and compared in previous studies on 1D structures, their application to experimental data rel-ative to a more complex test case presents some challenges in terms of numerical evaluation of strain energy integrals and inclusion of additional points for the calculation of curvatures. A critical aspect considered in this study was the uncertainty caused by the propagation of errors relative to the identification of modal shapes. To lower uncertainties related to the predicted damage location, and thus to find a better compromise between the false positives and false negatives, a macro-index based on ensemble learning is defined and ap-plied for the first time to 2D structures.