Physics-informed machine learning towards predictive modelling and physical mechanisms decoupling of crater formation in Ti6Al4V machining

Crater wear is a major limitation of machining efficiency and quality for difficult-to-cut Ti6Al4V alloys. In this study, a physics-informed neural networks (PINNs) model is proposed for predictive modelling of crater formation, where multiple physical laws are derived as the loss function to modulate the training process. Thermomechanical loads at the tool-chip interface are obtained by experiments and an analytical model. The proposed model shows good agreements with experimental data, which also constructs the link between the microscopic and macroscopic perspectives for understanding crater formation, as well as providing an approach for quantitative analyses of wear mechanism composition.