Machine Learning (ML), a subset of Artificial Intelligence has been widely applied in various domains, but it has only just begun to be employed in the field of engineering. In the present investigation, various ML algorithms and artificial neural network (ANN) structures are used for the first time to predict the mechanical properties of pristine, boron-doped, and nitrogen-doped graphene while also taking into account the effects of various types of vacancy defects.
Fracture strain, Ultimate Tensile Strength (UTS), and Young's modulus are all predicted. ML technique reduces the computational cost and time required to find out mechanical properties of these materials. The training dataset for the ML models is developed using Molecular Dynamics (MD) simulations. It was shown that defects and doping both had an adverse effect on mechanical characteristics. While ANN, LASSO, and LASSO Lars have all performed quite well at predicting these features, pipeline polynomial regression has performed best across all datasets. New insights on the research of mechanical characteristics utilizing cutting-edge computational techniques are provided by the discoveries in this research.