Deep prototypical networks based domain adaptation for fault diagnosis
- PDF / 1,125,566 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 92 Downloads / 209 Views
Deep prototypical networks based domain adaptation for fault diagnosis Huanjie Wang1,2 · Xiwei Bai1,2 · Jie Tan1
· Jiechao Yang1,2
Received: 13 March 2020 / Accepted: 29 October 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Due to the existence of domain shifts, the distributions of data acquired from different machines show significant discrepancies in industrial applications, which leads to performance degradation of traditional machine learning methods. In this paper, we propose a novel method that combines supervised domain adaptation with prototype learning for fault diagnosis. The proposed method consists of two modules, i.e., feature learning and condition recognition. The module of feature learning applies the Siamese architecture based on one-dimensional convolutional neural networks to learn a domain invariant subspace, which reduces the inter-domain discrepancy of distributions. The module of condition recognition applies a prototypical layer to learn the prototypes of each class. Then the classification task is simplified to find the nearest class prototype. Compared with existing intelligent fault diagnosis methods, this proposed method can be extended to address the problem when the classes from the source and target domains are partially overlapped. The model must generalize to unknown classes in the source domain, given only a few samples of each new target class. The effectiveness of the proposed method is verified using two bearing datasets. The model quickly converges with high classification accuracy using a few labeled target samples in training, even one per class can be effective. Keywords Bearing · Fault diagnosis · Domain adaptation · Prototype learning
Introduction Rolling element bearings are precision components in rotating machines, which are widely used in industrial, automotive, aerospace and marine applications (Ai 2013). With the development of advanced manufacturing technology, various sensors (e.g., temperature, vibration, displacement) have been utilized to monitor the condition of the bearing. The vibration signals that contain machine health information have proven to be effective for fault diagnosis and prognosis of bearings. This has promoted a great deal of work on vibration analysis over the last few decades (Yu et al. 2006; Sreejith et al. 2008; Wen et al. 2017b; Gao et al. 2019; Chen et al. 2020). Yu et al. (2006) proposed a method based on empirical mode decomposition (EMD) energy entropy for bearing
B
Jie Tan [email protected]
1
Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
2
School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China
fault diagnosis. EMD is applied to decompose the raw vibration signals and obtain intrinsic mode functions (IMFs). Then the extracted energy features of the IMFs are taken as input to the artificial neural network to distinguish normal bearing. The method proposed by Sreejith et al. (2008) extracts the Normal negative log-li
Data Loading...
