Year
2026
Season
Spring
Paper Type
Master's Thesis
College
College of Computing, Engineering & Construction
Degree Name
Master of Science in Mechanical Engineering (MSME)
Department
Engineering
NACO controlled Corporate Body
University of North Florida. School of Engineering
Committee Chairperson
Dr. Jutima Simsiriwong
Second Advisor
Dr. Gokan May
Rights Statement
http://rightsstatements.org/vocab/InC/1.0/
Third Advisor
Dr. Chady Ghnatios
Department Chair
Dr. Alan Harris
College Dean
Dr. William Klostermeyer
Abstract
The work presented in this thesis evaluates the surface roughness and geometric accuracy of fused deposition modeling (FDM)-produced parts utilizing machine learning. Through a Taguchi L54 design, part parameters of infill pattern, infill density, body thickness, raster angle, part orientation, and annealing were evaluated for two materials, Nylon 12 (polyamide-12), and ULTEM 1010 (polyetherimide). Surface roughness and geometric accuracy were quantified using an optical microscopy system that captured numeric and visual data at two locations on each printed part. The influences of printing parameters on surface roughness were modeled using three machine learning algorithms: Artificial Neural Network (ANN), Random Forest (RF), and Gaussian Process Regression (GPR). Models trained on Nylon 12 outperformed those trained on ULTEM 1010, with the best performance achieved by the RF model (R2 = 0.644). ANOVA identified orientation, body thickness, and measurement site as the most influential factors. During surface roughness characterization, full-site microscope images were collected and manually processed in MakeSense.AI to delineate surface boundaries. Each labeled region was then scaled using a known pixel-to-millimeter conversion to calculate the printed surface area. The measured areas were compared with the corresponding true areas defined in the CAD model, and four geometric accuracy metrics were computed: actual surface area, area percentage difference, intersection over union, and Hausdorff distance. These metrics were used to evaluate the influence of printing parameters on geometric accuracy using the three established algorithms. The RF model achieved the highest performance (R2 = 0.5617). Through ANOVA, material, measurement site, orientation, and annealing were identified as the most influential parameters on geometric accuracy. The performances of the machine learning models reveal key factors that impact both surface roughness and geometric accuracy, offering insights that can guide future research and improve part quality in industrial applications.
Suggested Citation
Pressly, Addison D., "Enhancing manufacturing quality and sustainability through Machine Learning" (2026). UNF Graduate Theses and Dissertations. 1417.
https://digitalcommons.unf.edu/etd/1417
Included in
Biomechanical Engineering Commons, Manufacturing Commons, Other Mechanical Engineering Commons
Accessibility Statement
This item was created or digitized before April 24, 2027, or is a reproduction of legacy material created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the Library provides accessible versions of archival materials by request. If you are experiencing difficulty accessing the information on the site due to a disability, please submit a request through the following form for assistance.