In-Situ Dimensional and Macro Surface Defect Mitigation in Wire-Fed Additive manufacturing Processes

Additive Manufacturing (AM) technologies have reached promising heights in the recent years while offering several advantages over traditional manufacturing methods, especially in the case of aerospace parts and in-space manufacturing. Whereas the advancements in wire feed metal AM are promising for recently ramped up LEO and deep space efforts of NASA (Artemis, OSAM) and private space agencies, the technology still lacks build reliability which would have ensured a defect free part with correct form-factor the first time without needing post-build quality assurance or in the worst case, part rejection. Thus, it is significant for a wire feed metal AM system to have the ability of in-situ, layer-by-layer build quality assurance by maintaining partsrsquo; dimensional accuracy and defect-free metal deposition. This Phase I effort presents a novel approach to solve the problem by targeting to detect, identify and correct optically detectable weld surface defects such as porosity, hot-cracking and lack of fusion. The first innovation is a camera vision-based layer-by-layer weld-surface defect identification system for wire feed AM process. The camera is mounted on an articulated robot for autonomous scanning of every single layer of deposited metal. A Convolutional Neural Network type Deep Learning-based software is utilized to aid defect identification and sending an output to the robot for a ldquo;Gordquo;/rdquo;No-Gordquo; decision over continuing deposit of subsequent layers. The second innovation is an automated laser displacement sensor for real-time measurement of part dimensions and auto-correction of robot path to compensate for discrepancies in metal deposition that can potentially result in parts with dimensional defects. The proposed innovations are targeted toward filling the gaps of real time dimensional and weld-surface defect detection, identification and correction for wire-feed metal AM processes.