Analysis of cumulative damage from double impacts in composites using neural networks

Due to their mechanical properties and high strength-to-density ratio, composite materials have become one of the most important resources in the development of the industrial and aerospace industries. Despite that, low velocity impacts (LVI) determine the onset of a particular kind of damage, which is undetectable by inspections, the Barely Visible Impact Damage (BVID). To address the problem, structural health monitoring (SHM) techniques have been developed to provide continuous monitoring of structural integrity. This thesis provides the classic passive SHM approach using a dataset generated through a determined number of Abaqus FEA simulating a double-impact scenario on a composite panel. In order to vary the set of data, four different impactor’s radii have been considered. In the end, a 1D CNN-LSTM hybrid neural network was used to provide the predictions about the damage.