Towards intelligent defect detection in metal powder bed fusion: A review of in situ monitoring, data pre-processing, and machine learning

Metal powder bed fusion (PBF) is a pivotal additive manufacturing (AM) technique for producing metallic parts. However, it is plagued by defects such as porosity, cracks, and warping, which compromise the quality of the final product. In response, there is a growing interest in leveraging in situ monitoring, data pre-processing, and machine learning (ML) techniques for defect detection and prediction in the metal PBF process. This review provides a comprehensive analysis of current advancements in these areas. Specifically, we highlight the emerging trend of data pre-processing that serves as a bridge between in situ monitoring and ML. By addressing challenges such as background noise, data loss, and large volumes of data, pre-processing of in situ monitoring data plays a crucial role in improving the accuracy of defect detection and prediction in the metal PBF process. We also discuss notable methodologies, technologies, and trends in the field, offering insights into the current challenges and potential prospects for advancing in situ monitoring, data pre-processing, and ML techniques for defect investigation in metal PBF printed components.