Deep Learning-Based 3D Printer Fault Detection
Mark Verana, Cosmas Ifeanyi Nwakanma, Jae Min Lee, Dong‐Seong Kim
Abstract
The development of intelligent manufacturing and 3D printers is rapidly engaging in the industry. However, 3D printers are challenged by occasional anomalies due to leading to failure in 3D performance. In this work, a fault diagnosis based on a convolutional neural network (CNN) for 3D printers is proposed. We have leveraged an online repository of a set of data streams collected from working 3D printers. The CNN was used to process, detect and classify anomalies in 3D printing with appreciable accuracy. The proposed CNN outperformed the support vector machine (SVM), and artificial neural network (ANN) by 5.1% and 25.7%, respectively.
Topics & Concepts
Convolutional neural networkComputer scienceSupport vector machineArtificial neural networkArtificial intelligenceDeep learningProcess (computing)3D printingFault (geology)Set (abstract data type)Pattern recognition (psychology)Data set3d printerEngineeringOperating systemProgramming languageSeismologyGeologyMechanical engineeringIndustrial Vision Systems and Defect DetectionAnomaly Detection Techniques and ApplicationsAdvanced Neural Network Applications