Graphene exhibits a unique combination of mechanical, thermal and electrical properties due to the strong and anisotropic bonding, enabling a wide range of novel thermal management and electronic applications. However, it is extremely challenging and costly to investigate graphene solely depending on experimental tests. Atomistic simulation plays an essential role in material system analysis and design and is specifically powerful in characterizing low dimensional materials. However, successful applications of atomistic simulation highly depend on the fidelity and availability of force field potentials for describing the interatomic interactions. Significant discrepancies exist between the simplified empirical potentials and the reference data, and among the empirical potentials themselves. To address the challenge, a new artificial neural network potential is developed for graphene to enable the characterization of the interested properties using molecular dynamics simulations, which is expected to accelerate the discovery and design of novel graphene enabled functional materials.