Comparison of Artificial Neural Networks with Response Surface Models in Characterizing the Impact Damage Resistance of Sandwich Airframe Structures

In the development of a damage tolerance plan for composite airframe structures, the way to characterize the impact damage of sandwich composites under different levels of impact events and material property is crucial. The aim of the present research is to investigate the influence of material configuration and impact parameters on damage resistance responses of composite sandwich structures comprised of carbon-epoxy woven fabric facesheets and Nomex honeycomb cores. Two methods, artificial neural network and classic response surface methodology were used to predict the relationship between the impact damage response and its dependent parameters. The results obtained through artificial neural networks were compared with those through response surface methodology.