(109c) Computational Fluid Dynamics-Based in-Situ Sensor Analytics of Direct Metal Laser Sintering Process Using Machine Learning

The wide utilization of the AM method lies in its advantages including rapid prototyping, faster manufacturing, reduced time and operating cost, and also the possibility to create almost any geometry. For complicated, high yield strength and high precision metal part construction, laser powder bed fusion (LPBF) has been extensively researched and prototyped [1]. To identify errors in real-time to avoid detrimental damage to the machine and material and energy waste, in-situ sensor monitoring technologies are developed to record the manufacturing information [2]. Nevertheless, the overwhelming amount of image based data produced by the optical sensors poses challenges in the data analysis and storage [3].

To address the aforementioned problem, we develop an automated data-flow that integrates the simulation model and machine learning network for real-time process simulation and sensor data analytics. In particular, the simulation models, such as thermal history analysis, are tailored to mimic the manufacturing platform characteristic and to incorporate potential process disturbances. From the simulation result, the heat map images are reconstructed with corresponding classified disturbances categories, according to the industrial sensor technologies. Advanced machine learning algorithms like convolutional neural networks (CNN) are designed and trained with transfer learning to achieve a robust yet computationally acceptable in-situ process analysis.

[1] Frazier, W.E., 2014. Metal additive manufacturing: a review. Journal of Materials Engineering andPerformance 23, 1917–1928.

[2] Everton, S.K., Hirsch, M., Stravroulakis, P., Leach, R.K., Clare, A.T., 2016. Review of in-situ process monitoring and in-situ metrology for metal additive manufacturing. Materials & Design 95.

[3] Scime, L., Beuth, J., 2019. Using machine learning to identify in-situ melt pool signatures indica-tive of flaw formation in a laser powder bed fusion additive manufacturing process. AdditiveManufacturing 25, 151–165.