Machine Learning Based in Process Monitoring of Powder Spreading Abnormalities in Laser Powder Bed Fusion

Additive manufacturing is a revolutionary technique that offers unmatched advantages due to its ability to create complex parts with design flexibility, catering to a wide array of applications. It comprises of a variety of methods and Laser Powder Bed Fusion (LPBF) is one of the metal based techniques, in which a laser is used to melt and fuse a thin layer of powder spread on a build platform by a device known as recoater blade/roller. It has several advantages such as producing parts with near-net shape, intricate contours and reduced production times. However, it is prone to defects such as porosity, lack of fusion, warping etc. and powder spreading abnormalities like recoater interference, layer distortion, powder overfeed/underfeed etc. These defects and powder spreading abnormalities are formed owing to the influence of various process variables and printing environment and could lead to the wastage of resources coupled with detrimental impact on the final component’s properties. As a result, it is imperative to control and prevent these defects, in order to obtain the desired performance in the product and this study primarily focuses on the application of a Machine Learning (ML) algorithm for in process monitoring of powder spreading within the LPBF process, using the layerwise images of real samples, for early detection of powder spreading abnormalities such as recoater interference, layer distortion and streaking.