Title: Deep learning for defect identification in 3D printing with fused filament fabrication

Authors: Deepak Malekar; Shreyas Aniyambeth; Tuğrul Özel

Addresses: Rutgers University, Manufacturing and Automation Research Laboratory (MARLAB), Department of Industrial and Systems Engineering, Piscataway, New Jersey, 08854, USA ' Rutgers University, Manufacturing and Automation Research Laboratory (MARLAB), Department of Industrial and Systems Engineering, Piscataway, New Jersey, 08854, USA ' Rutgers University, Manufacturing and Automation Research Laboratory (MARLAB), Department of Industrial and Systems Engineering, Piscataway, New Jersey, 08854, USA

Abstract: Fused filament fabrication (FFF) is one of the most utilised three-dimensional printing technologies. The process involves concurrently feeding and melting a thermoplastic filament material into liquefier and extruding through a nozzle in a computer-controlled print head for continuously depositing filament lines to form layers of a 3D geometry. While the deposited molten material quickly solidifies and shrinks, deposited segments and layers should fuse uniformly and bond together seamlessly. However, the FFF process suffers from several issues such as irregular sizes of deposited filament lines, a high rate of printing errors and defects, and poor surface finish. These problems and defects could cause significant deformity and generate barriers for full industrial adoption of FFF in manufacturing. A method for defect identification has been proposed using deep learning. This paper first presents a brief literature review and then a deep learning model with experiments to identify defects as the printing process occurs.

Keywords: additive manufacturing; smart manufacturing; FFF; fused filament fabrication; polymer; monitoring; sensors; measurement; quality; defects; machine learning; deep neural network.

DOI: 10.1504/IJMMS.2023.133395

International Journal of Mechatronics and Manufacturing Systems, 2023 Vol.16 No.2/3, pp.243 - 260

Received: 06 Jan 2023
Accepted: 01 Jun 2023
Published online: 14 Sep 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article