Machine learning assisted prediction of mechanical properties of graphene/aluminium nanocomposite based on molecular dynamics simulation

Abstract

Predicting mechanical properties of graphene-reinforced metal matrix nanocomposites (GRMMNCs) usually requires atomistic simulations that are computationally expensive without scalability or micromechanics-based models (such as widely used Halpin-Tsai model) that may lead to considerable errors in many cases. This paper first combines molecular dynamic (MD) simulation and machine learning (ML) techniques to predict the mechanical properties of graphene reinforced aluminium (Gr/Al) nanocomposites and use them to modify Halpin-Tsai model. Extensive MD results for Young’s modulus and ultimate tensile strength of Gr/Al nanocomposites are obtained, with the intricate effects of graphene’s volume fraction, alignment angle, chirality and environment temperature being taken into account. After training and optimization based on MD data, ML models are developed with the capability in estimating Young’s modulus and ultimate tensile strength. The micromechanics based Halpin-Tsai model is then modified by using the Young’s modulus predicted by both MD and ML models such that the Young’s modulus can be readily determined with significantly improved accuracy from an explicit relationship that is very easy to use in the analysis and engineering design of Gr/Al nanocomposite structures.