Real-time Defect Detection in Ceramic Additive Manufacturing Using Deep Convolutional Neural Networks
Abstract
Although ceramic additive manufacturing technologies have greatly advanced over the past decade, they are still far from perfect. Even small printing defects can have a severe impact on the mechanical properties of additively manufactured ceramic parts due to their low fracture toughness. The objective of this study was to detect these defects during the printing process for a new ceramic additive manufacturing technique, called ceramic on-demand extrusion (CODE). Images were collected, pre-processed, augmented, and separated based on their distinctive classes (i.e., type of defects). Deep convolutional neural networks were developed to identify multiple types of defects in real-time.
Fields of Study & Subjects
Additive Manufacturing Materials and Processes
Additive Manufacturing and 3D Printing Technologies
Machine Learning in Materials Science
Publication details
- Journal / source
- Procedia CIRP
- Publisher
- Elsevier BV
- Volume
- 138
- Pages
- 602–607
- Publication date
- 1 January 2026
- Language
- EN