Machine learning applied to Eddy Current data collected during LPBF manufacturing
In the frame of a partnership with the University of Applied Sciences and Arts Western Switzerland (HES-SO), a paper has been recently published on the application of machine learning to detect process degradation during the LPBF process monitored with Eddy Current sensors. The empirical approach is achieved by setting up a trained AI algorithm for the in-situ detection of porosity defects generated during the part fabrication. Comparison between predicted and experimental outcomes shows that the method allows the detection of porosity layer by layer with a mean absolute error of 0.1% using convolutional neural networks. Thanks to Prof. Haifa Sallem and Prof. Hatem Ghorbel and their teams for this work. Contact us if you want to know more about this work.
https://link.springer.com/chapter/10.1007/978-3-031-47784-3_18