Abstract
This research focuses on developing a Convolutional Neural Network (CNN)-based machine learning algorithm for detecting anomalies in Fused Deposition Modelling (FDM) 3D printers. To develop the CNN algorithm, the training data was obtained from the experiments. Therefore, process parameters such as; material flow, printing speed, and vibration of the print bed were adjusted to create infill pattern defects such as over-extrusion and layer shift (misalignment). A high-resolution camera was used to take the images during the printing. The developed CNN-based machine learning algorithm is capable of detecting and distinguishing types of defects providing a systematic quality assessment of a product. The main advantage is that it can be used to detect different types of defects related to infill patterns, and it can reduce the calibration error. Therefore, the application of machine learning in defect detection in 3D printing illustrates the prospects of integrating machine learning algorithms into manufacturing processes.