Fusing convolutional generative adversarial encoders for 3D printer fault detection with only normal condition signals

•A fault detection model can be built only using normal condition signals measured in the 3D printers.•The 1D convolutional generative adversarial encoder can build a deep feature space oriented to the detection task.•In comparison with the single sensor approach, fusing velocity and angle signals i...

Full description

Saved in:
Bibliographic Details
Published in:Mechanical systems and signal processing 2021-01, Vol.147, p.107108, Article 107108
Main Authors: Li, Chuan, Cabrera, Diego, Sancho, Fernando, Sánchez, René-Vinicio, Cerrada, Mariela, Long, Jianyu, Valente de Oliveira, José
Format: Article
Language:English
Subjects:
3D printers
Adversarial learning
Artificial neural networks
Coders
Condition-based maintenance
Convolutional neural networks
Data collection
Fault detection
Mechanical systems
Printers
Support vector machines
Three dimensional printing
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A fault detection model can be built only using normal condition signals measured in the 3D printers.•The 1D convolutional generative adversarial encoder can build a deep feature space oriented to the detection task.•In comparison with the single sensor approach, fusing velocity and angle signals improve the fault detection rate.•An isolation forest-based or SVM-based discriminative function can be used to characterize the deep feature space. Collecting data from mechanical systems in abnormal conditions is expensive and time consuming. Consequently, fault detection approaches based on classical supervised learning working with both normal and abnormal data are not applicable in some condition-based maintenance tasks. To address this problem, this paper proposes Fusing Convolutional Generative Adversarial Encoders (fCGAE) method to create fault detection models from only normal data. Firstly, to obtain an adequate deep feature space, encoder models based on 1D convolutional neural networks are created. Then, these encoders are optimized in an unsupervised way through Bidirectional Generative Adversarial Networks. Finally, the multi-channel features collected from the system are merged with One-Class Support Vector Machine. fCGAE is applied to fault detection in 3D printers, where experimental results in two fault detection cases show excellent generalization capabilities and better performance compared to peer methods.
ISSN:0888-3270
1096-1216
DOI:10.1016/j.ymssp.2020.107108