Hybrid approach to predict the effective properties of heterogeneous materials using artificial neural networks and micromechanical models

In this article, an investigation was carried out to verify hybrid models capabilities to predict the effective properties of heterogeneous materials. A hybrid model is developed by combining artificial neural networks and micromechanical modeling. The homogenization approach used in this study is mainly based on Eshelby's inclusion problem. The model, once trained on an Eshelby's tensors database, showed an excellent predictive capabilities of...