AI driven process monitoring for 3D printing technologies

Abstract

The Direct energy Deposit (DED) method is a technique of additive manufacturing (AM) that deposits the required material from an origin powder or wire material stock onto a base material. The machine does so by using high powered beams in the form of lasers, electron beams, electric arc, or plasma to continuously melt the feedstock material to form a tiny melt pool and does so in single layers. While there are numerous advantages that DED has over its other counterparts, one key issue that is restricting its widespread application is due to its inconsistent print quality. The attribution of inconsistent print quality is due to many factors, such as inconsistent machine speeds, thermal stress from the quick heating and cooling cycles, and localized heat accumulation. Current measures to quality control the print is to use different visual equipment to observe and adjust the printing arm speed and laser power all from a central computer. In this project, data processing will be applied on cloud point datasets and visual analysis done to identify key weaknesses in DED printing processes. Results are discussed and lastly, limitations and future work are discussed.