AI-assisted prediction of particle impact deformation simulated by Material Point Method

A hybrid AI framework has been adopted to link the impact deformation of elastic-perfectly plastic particles with their material properties and impact velocity. Initially, Material Point Method (MPM) is employed to simulate the impact of an elastic-perfectly plastic particle with a rigid wall, covering an extensive range of material properties and impact velocities. The simulation results are then integrated into the AI framework to establish the relationship between the input and output parameters. Consequently, dimensionless equations are derived to predict the equivalent plastic strain and deformation extent based on the material properties and impact velocity of the particle, showing a strong agreement with the MPM results. The identified equations reveal that both the equivalent plastic strain and deformation extent depend on and can be determined from the yield strength of the material as well as the fraction of the incident kinetic energy that is spent on inducing plastic deformation. The validity of the equations is verified by comparing the MPM and predicted values of the equivalent plastic strain and deformation extent for cases with material properties and impact velocities beyond the initial dataset used for developing the equations. The equation identified by the framework for prediction of the deformation extent