Uhlmann, EckartEckartUhlmannReis de Ascencao, GustavoGustavoReis de AscencaoHesse, BernhardBernhardHesseSuuronen, Jussi PetteriJussi PetteriSuuronenDomingos, David CarlosDavid CarlosDomingosZhuang, JianlinJianlinZhuang2025-04-082025-04-082025https://publica.fraunhofer.de/handle/publica/48631910.1007/978-3-031-77429-4_832-s2.0-85218098330The sustainability of AM is strongly related to the ability of producing functional parts using minimal resources. In-situ monitoring saves energy and material by providing information to support process stops in case of anomalous events detected in critical regions. In this work, AlSi10Mg samples with different defect conditions were PBF-LB/M manufactured under the observation of photodiodes in three wavelength ranges. The samples were scanned through synchrotron-based micro-CT, and a defect score was attributed to every point in a non-destructive manner. The results were registered to the emissions acquired during the PBF-LB/M process. An exploratory data analysis has shown that the mean and standard deviation of the emissions of defect points do not differ from the ones found for points labelled as normal by micro-CT. A neural network was trained to infer the points’ quality based on their emissions. The maximum F1-score was 4.79%, suggesting that the photodiodes in the studied set-up are not suitable for the in-situ identification of local porosities in PBF-LB/M parts.entrueAdditive Manufacturing (AM)Computed Tomography (CT)Laser-based Powder Bed Fusion of Metals (PBF-LB/M)Process Monitoringconference paper