A Machine-Learning-Based Approach to Critical Geometrical Feature Identification and Segmentation in Additive Manufacturing

Additive manufacturing (AM) processes offer a good opportunity to manufacture three- dimensional objects using various materials. However, many of the processes, notably laser Powder bed fusion, face limitations in specific geometrical features due their physical constraints, such as thermal conductivity surrounding medium, internal stresses, and warpage or weight part being manufactured. This work investigates use machine learning algorithms order identify hard-to-manufacture features. The segmentation these from main body permits application different strategies improve overall manufacturability. After selecting that are particularly problematic during powder fusion stainless steel, an algorithm is trained simple geometries, which identification on new parts with success rate 88%, showing potential this approach.