Arman Sabbaghi, the leading researcher and assistant professor of statistics at Purdue University, explained, “We’re really taking a giant leap and working on the future of manufacturing.”
“We have developed automated machine learning technology to help improve additive manufacturing. This kind of innovation is heading on the path to essentially allowing anyone to be a manufacturer.”
Catching the deviants
In this research Sabbagh was joined by Raquel Ferreira and Kevin Amstutz, graduate students at Purdue, Qiang Huang from the University of Southern California, and Tirthankar Dasgupta of Rutgers University.
Their research establishes a predictive methodology based on the Bayesian probability theory. The algorithms measure the deviations in the shape of a 3D printed model to improve the quality of the next print.
Funded by the National Science Foundation, a U.S. government funding agency, the findings of the study were published in two separate papers. The first paper developed a method to predict shape deviations in stereolithography 3D printing process. It states “… if an initial deviation model speciﬁcation, and associated compensation plan, does not suﬃciently reduce deviation, the manufactured shape is not a loss because it still contains useful information for updating the model and compensation plan.”
“Diﬀerent types of discrepancy measures can be applied to the new shape as in the ﬁrst step of our methodology, so as to directly learn from it and previous (distinct) shapes how to reﬁne the deviation model. Such a closed loop dynamic addresses the goal of low-cost and high-conﬁdence quality control for 3D printing.”
Following this, the latter paper shows how the Bayesian methodology can be used to ensure transferability from one process to another, i.e. from SLA/DLP to metal 3D printing or FDM/FFF.
Quality control in AM
One of the primary benefits of 3D printing is that it can manufacture complex shapes that are difficult to achieve using traditional manufacturing methods. However, one of the frontiers to be crossed by additive manufacturing is efficient quality control which minimizes print runs and costs of manufacturing.
Technology companies and researchers have attempted to tackle the quality assurance problem in various ways, such as advanced simulation software, a combination of hardware and software and machine learning methods, such as the one founded by the Purdue researchers.
Arman Sabbaghi, said, “This has applications for many industries, such as aerospace, where exact geometric dimensions are crucial to ensure reliability and safety.”
“We use machine learning technology to quickly correct computer-aided design models and produce parts with improved geometric accuracy.”
The research papers discussed in this article were titled, “Bayesian Model Building From Small Samples of Disparate Data for Capturing In-Plane Deviation in Additive Manufacturing“, published in Technometrics and “Model Transfer Across Additive Manufacturing Processes via Mean Effect Equivalence of Lurking Variables“, published in The Annals of Applied Statistics.
3D Printing Awards 2019 are near. Nominate Now!
Looking for a job in the industry? Then visit our 3D Printing Jobs board.
Featured image shows the main entrance of Purdue University. Photo via Matthew Thomas/Purdue University.