High precision 3D contour detection for laser powder bed fusion in-process layerwise monitoring using active contours driven by multiple energy

Abstract

The Laser Powder Bed Fusion (LPBF) process utilizes a layered manufacturing approach, offering significant advantages in the fabrication of complex metal components. A key challenge in the production of high-precision, high-performance parts is the online monitoring of the solidified region in each layer of the powder bed, as well as identifying geometric defects. In this study, a novel approach for contour extraction is proposed, integrating multiple energy terms, including grayscale, entropy, phase difference, and photometric stereo, to drive the evolution of active contours. Various combinations of these energy terms are compared, and the effectiveness of each is validated. The results show that, compared to traditional active contour methods that consider only grayscale energy, the proposed method reduces the mean squared errors by 84.4 %, 61.7 %, and 95.8 % for ring contours, diamond ring contours, and center distances, respectively. These improvements contribute to the development of a quality monitoring and process parameter feedback system based on 3D geometric deviations in the solidified region, thus enhancing the precision and repeatability of LPBF processes.