A Machine Learning Model to Predict Tensile Properties of Annealed Ti6Al4V Parts Prepared by Selective Laser Melting
In the present work, a machine learning method (artificial neural network) is used to develop a data-driven model to identify the correlations between annealing process parameters, microstructures and mechanical properties of Ti6Al4V parts. A database was first established by collecting data published reports on annealing treatment of SLM Ti6Al4V from 2006 to 2020, using Google Scholar, Science Direct, Research Gate, and Springer.
The model shows a promising performance in predicting yield strength (YS), ultimate tensile strength (UTS) and elongation (EL), as indicated by the relatively high accuracy (> 90.56%). The model also confirms that the UTS and YS are sensitive to annealing temperature and annealing time. The long exposing time and high heating temperature result in extensive grain growth and finally leads to low final YS and UTS. A Hall-Petch relationship is established to describe the relationship of YS and grain size for heat treated Ti6Al4V parts prepared by SLM process. The kHP for the Hall-Petch relationship is significantly larger than previous study is this area. The prediction of strain to failure shows lower accuracy compared with that of YS and UTS. The differences among the printing processes by different research groups may have resulted in large variations of structural defects and thus led to significant scattering of the elongation data.