Wednesday  |  Monongahela  |  09:40 AM–10:00 AM

#13847, Machine Learning Based Material Models for Engineering Materials Subjected to Random Loading Paths

The main aim of this study is to develop experimental-data analysis methodologies to characterize the mechanical behavior of lightweight materials. The machine learning approaches are herein applied with the purpose of capturing the experimentally measured mechanical behavior and predict the response of material and structures subjected to complex random loading paths. Complex loading path experiments are performed under tension- torsion- compression random combined loading. The main objective of this study is to translate the obtained machine learning algorithms into suitable user subroutines to be employed in conjunction with commercial computer-aided engineering software and validate their effectiveness against real loading case scenarios. In this way, data-based machine learning driven models obtained from arbitrary complex mechanical loading experiments are predicted to allow for computationally efficient prediction of the deformation and failure of engineering components during real loading case scenarios. This results in effective optimization of their morphology and, consequently, a significant reduction in weight, particularly important for aerospace applications.

Burcu Tasdemir   University of Oxford
Vito Tagarielli   Imperial College London
Antonio Pellegrino   University of Oxford