Title: Machine-Learning for Automated Fiber Placement for Manufacturing Efficiency and Process Optimization

Authors: Waruna Seneviratne, John Tomblin, Upul Palliyaguru

DOI: 10.33599/nasampe/s.21.0491

Abstract: In order to meet aggressive demand, aircraft manufacturing processes must undergo significant technology advancements and future manufacturing engineers must be equipped with advanced hybrid, scalable, flexible, and extensible tools to adapt to growing complexities. Global aircraft manufacturers are aggressively seeking methods for advancing manufacturing technologies through automation and innovative materials/processes that increase manufacturing rates and efficiency. With the advancement of sensor technologies and manipulators, industrial robots are now capable of performing non-routine complex functions such as labor-intensive advanced composite layup that typically require meticulous, trained technicians. Automated fiber placement (AFP) has the potential to significantly decrease lead-time with increased material yield and production rates due to fewer interruptions and improved consistency. With the use of advanced sensors, process simulation software, and in-process inspection systems, labor-intensive nondestructive inspection for quality assurance can be automated for minimizing interruptions and to significantly improve part quality. In-process inspection systems equipped with advanced sensors is deployed for automatically identifying manufacturing defects and feed digital information into machine learning algorithms to take corrective actions on subsequent manufacturing runs to improve part quality. This approach, which develops a digital manufacturing twin for supporting sustainment activities, also fits well into the Factory of the Future concept and will aid in increasing production rates of commercial and defense aircraft.

References: [1] C. P. C. P. Tom Cooper. John Smiley, Global Fleet & MRO Market Forecast Commentary, Oliver Wyman, 2018. [2] X. Z. S. R. J. S. Kaiming He, "Deep Residual Learning for Image Recognition," Microsoft Research, 2015. [3] D. A. D. E. C. S. S. R. ,. C.-Y. F. A. C. B. Wei Liu, "SSD: Single Shot MultiBox Detector," UNC Chapel Hill 2, 2016.

Conference: SAMPE NEXUS 2021

Publication Date: 2021/06/29

SKU: TP21-0000000491

Pages: 15

Price: FREE