Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Melt Pool Monitoring using Fuzzy Based Anomaly Detection in Laser Beam Melting

Vortrag gehalten auf der Metal Additive Manufacturing Conference (MAMC 2019), 25.-27. November, Örebro, Sweden
: Boos, Eugen; Schwarzenberger, Michael; Jaretzki, Martin; Wiemer, Hajo; Ihlenfeldt, Steffen

2019, 12 pp.
Metal Additive Manufacturing Conference (MAMC) <2019, Örebro>
Conference Paper
Fraunhofer IWU ()
additive manufacturing; laser beam melting; process monitoring; QM-Meltpool 3D; uncertainty analysis; fuzzy; anomaly detection; feature engineering
Anfrage beim Institut / Available on request from the institute

With the introduction and further development of additive manufacturing (AM) processes and topology optimization algorithms, components are designed to meet high performance objectives. However, the quality standards required for the high-performance market are not entirely given yet. Known but hard to determin-istically quantiable inuences have a considerable eect on the manufacturing process and consequently the microstructure and the mechanical properties of the components. Therefore, process monitoring systems are implemented for a better evaluation of the process itself and the quality of the component. Due to the challenging nature of the AM process, data uncertainties within the collected monitoring data are common. To achieve the high quality criteria not only an improved understanding of the present uncertain inuences is crucial, but also correspondingly the correct mapping of the collected monitoring data itself. This paper presents an approach to monitor and evaluate the microstructure of a Laser Beam Melting (LBM) printed component. It aims to estimate the quality of the component by determining printing anomalies within the solid structure, using the fuzzy set theory for feature engineering. The anomaly detection exploits the collected melt pool data generated by image analysis with optical signals of micro-sections for the fuzzy based anomaly detection. The proposed concept is proven with nested printing anomalies.