Predicting Strength of Selective Laser Melting 3D Printed A1Si10Mg Alloy Parts by Machine Learning Models

Materials and components are manufactured concurrently in a single process in metals additive manufacturing (AM), where metal layers are fabricated on top of each other in the near-final topology necessary for the end-use product. The adoption of modern machine learning algorithms to simulate these degrees of freedom can speed up and lower the cost of understanding metals AM. However, the poor strength of the produced items is determined by the flaws created during laser powder bed fusion. The influence of Laser Power, Scan Speed, Scan Space, and Island Size on the tensile strength of A1Si10Mg alloy, developed by a selective laser melted process, was investigated using machine learning models. Under varied settings, six machine learning models were compared: Deep Learning, Decision Tree, Bagging, Linear Regression, Ridge Regression, and Random Forest. The two top models were found to be Deep Learning and Decision Tree, with prediction levels of 99 percent and 89 percent, respectively. The Laser Power was also discovered to be one of the most influential parameters, and it should be kept on the higher side for improved tensile strength.