Automated Testing and Characterization of Additive Manufacturing (ATCAM)
Open Access
- 2 August 2021
- journal article
- research article
- Published by Springer Nature in Journal of Materials Engineering and Performance
- Vol. 30 (9) , 6862-6873
Abstract
The sensitivity of additive manufacturing (AM) to the variability of feedstock quality, machine calibration, and accuracy drives the need for frequent characterization of fabricated objects for a robust material process. The constant testing is fiscally and logistically intensive, often requiring coupons that are manufactured and tested in independent facilities. As a step toward integrating testing and characterization into the AM process while reducing cost, we propose the automated testing and characterization of AM (ATCAM). ATCAM is configured for fused deposition modeling (FDM) and introduces the concept of dynamic coupons to generate large quantities of basic AM samples. An in situ actuator is printed on the build surface to deploy coupons through impact, which is sensed by a load cell system utilizing machine learning (ML) to correlate AM data. We test ATCAM’s ability to distinguish the quality of three PLA feedstock at differing price points by generating and comparing 3000 dynamic coupons in 10 repetitions of 100 coupon cycles per material. ATCAM correlated the quality of each feedstock and visualized fatigue of in situ actuators over each testing cycle. Three ML algorithms were then compared, with Gradient Boost regression demonstrating a 71% correlation of dynamic coupons to their parent feedstock and provided confidence for the quality of AM data ATCAM generates.Keywords
This publication has 62 references indexed in Scilit:
- Progress on ‘pico’ air vehiclesThe International Journal of Robotics Research, 2012
- Proactive Assembly Oriented Design Approach Based on the Deployment of Functional RequirementsJournal of Computing and Information Science in Engineering, 2011
- RepRap – the replicating rapid prototyperRobotica, 2011
- RepRap: The Replicating Rapid Prototyper: Maximizing Customizability by Breeding the Means of Production2009
- Computational Tools for Probing Interactions in Multiple Linear Regression, Multilevel Modeling, and Latent Curve AnalysisJournal of Educational and Behavioral Statistics, 2006
- Evolutionary Robotics: A New Scientific Tool for Studying CognitionArtificial Life, 2005
- Classical and Modern Regression With Applications.Journal of the American Statistical Association, 1988
- A Test for Normality of Observations and Regression ResidualsInternational Statistical Review, 1987
- Outliers and influential data points in regression analysis.Psychological Bulletin, 1984
- The Assumptions of the Linear Regression ModelTransactions of the Institute of British Geographers, 1971