Mechanical performance of zirconia-silica bilayer coating on aluminum alloys with varying porosities: Deep learning and microstructure-based FEM

•A novel multiscale modeling framework is proposed to predict the effective mechanical properties of heterogeneous materials.•Generative adversarial networks are used to generate microstructures with different porosities.•Elastic properties of silica and zirconia are greatly reduced when porosities exceed 5% and 3%, respectively.