Machine Learning Assisted Prediction of the Manufacturability of Laser-Based Powder Bed Fusion Process

Abstract

Laser-based powder bed fusion (LPBF) process is a type of additive manufacturing process which is able to produce complex metal geometries. The fast development of laser-based powder bed fusion process offers new opportunities to the industries. Comparing to the conventional manufacturing process, LPBF offers more freedom on the shape complexity and hierarchical complexity. Even though the LPBF process has many advantages, there are still many constraints on LPBF. At the current stage, LPBF process still has a very high threshold for industrial application. It requires designers to have extensive knowledge of LPBF process to make the design manufacturable. The need for the automatic manufacturability analysis in the early design stage is essential. In this paper, a novel approach on analyzing the manufacturability of LPBF process is introduced. The machine learning model is developed to predict the manufacturability of LPBF. The unique dataset is established as the training examples. The proposed method achieves very competitive accuracy on analyzing the manufacturability of LBPF. The limitation and future work will be discussed in the end.