Artificial neural networks for impact force reconstruction on composite plates

Abstract

Impacts are one of the main causes of damage in composite panels. The determination of the impact location and the reconstruction of impact force are necessary to evaluate the health of the structure. These data may be measured indirectly from the measurements of responses of sensors located on the system subjected to the impact. In this study, a composite panel model developed in Abaqus/CAE is first validated and then numerical simulations based on the model are used to obtain data for several impacts, characterized by different impact locations, different impactor velocities and masses. Subsequently, these data are used to model the complex nonlinear behavior of the composite laminate by a nonlinear system identification approach. This is based on the use of artificial neural networks, which are employed to accurately reconstruct the impact forces and the impact locations.