Vibration-based damage identi cation in beam-like composite laminates by using arti cial neural networks

This paper investigates the effectiveness of the combination of global (changes in natural frequencies) and local (curvature mode shapes) vibration-based analysis data as input for artiŽ cial neural networks (ANNs) for location and severity prediction of damage in Ž bre-reinforced plastic laminates. A Ž nite element analysis tool has been used to obtain the dynamic characteristics of intact and damaged cantilever composite beams for the Ž rst three natural modes. Different damage scenarios have been introduced by reducing the local stiffness of the selected elements at different locations along the Ž nite element model of the beam structure. After performing the sensitivity analyses aimed at Ž nding the necessary parameters for the damage detection, different input–output sets have been introduced to various ANNs. In order to check the robustness of the input used in the analysis, random noise has been generated numerically and added to noise-free data during the training of the ANNs. F inally, trained feedforward back-propagation ANNs have been tested using new damage cases and checks have been made for severity and location prediction of the damage.