A Data-Driven Approach for Selecting Critical Process Parameters in Material Extrusion Additive Manufacturing

F. Pourkamali-Anaraki, A. Peterson, R. Jensen
University of Massachusetts Lowell,
United States

Keywords: additive manufacturing, fused filament fabrication, process parameters, machine learning


Material extrusion is an additive manufacturing methodology that holds great potential to reduce costs associated with creating complex structures for various industries. However, a significant challenge is to uncover the influences of numerous processing parameters, such as environmental and printing conditions, on resulting structures and their properties. In this work, we address this issue by employing a statistical method that allows us to automatically identify a fraction of relevant parameters that describe processing-structure-property relationships. Notably, we discern influential parameters using LASSO (least absolute shrinkage and selection operator), a popular technique in machine learning that performs regression analysis and variable selection simultaneously. The parameter space dimension reduction method will improve the effectiveness of the design of experiments (DOE), thus facilitating screening and optimization procedures in additive manufacturing. Our case study stems from a recent work that designed an informatics workflow for collecting data from each stage of the fused filament fabrication (FFF) printing process. Our results show that printer selection is crucial for understanding processing-structure-property relationships in extrusion material additive manufacturing.