Abstract Scope |
The electron beam melting (EBM), one of powder bed fusion additive manufacturing technologies, is complex process, and various parameters have a huge influence on the performance of the formed part. In order to expand the use of EBM technology in the materials industry, one of the problems is the generation of internal defects (pores, un-melted powder, and so on) during process. In this research, we applied several machine learning algorithms for process optimization of the EBM process of a carbon steel, and we compared accuracy of each algorithms. We determined a quantitative criteria for classifying surface quality based on surface roughness, and we have revealed that different surface quality (uneven, even, and porous) includes different type of internal defects. Even surface samples have the highest density and hardness. In addition, six kinds of machine learning algorithms have been investigated. Among them, the SVM model has the highest accuracy. |