Detection of Spaghetti and Stringing Failure in 3D Printing

Three-dimensional (3D) printing technology is getting more accessible today and many products can be self-printed using this method. However, the occurrence of printing errors or defects like spaghetti and stringing remains a significant challenge within the 3D printing industry. Spaghetti defects cause incomplete prints, rendering them unusable, while stringing defects introduce surface imperfections to the printed object that can still be fixed. The printing errors, which usually occur during the printing process, can lead to waste in time, material, and cost. Therefore, a failure detection system specifically tailored for Fused Deposition Modelling (FDM) 3D printing is proposed, incorporating computer vision techniques and machine learning models. The AlexNet-SVM (Support Vector Machine) model is employed for spaghetti and stringing defects classification. By training the AlexNet-SVM model with a total of 1058 images, 87.1% accuracy was achieved by the AlexNet-SVM model to classify spaghetti defect class, stringing defect class and no defect class. With the implementation of the model into the built prototype, an overall accuracy of 86% was achieved. The accuracies obtained proved the efficiency of the proposed detection system which will help to reduce wastage in the 3D printing industry.