Abstract
The bearing is a component of the support shaft that guides the rotational movement of the shaft, widely used in the mechanical industry and also called a mechanical joint. In bearing fault diagnosis, the accuracy much depends on the feature extraction, which always needs a lot of training samples and classification in the commonly used methods. Neural networks are good at latent feature extraction and fault classification, however, they have problems with instability and over-fitting, and more labeled samples must be trained. Switchable normalization and semi-supervised learning are introduced to solve the above obstacles in this paper, which proposes a novel bearing fault diagnosis method based on switchable normalization semi-supervised generative adversarial networks (SN-SSGAN) with 1-dimensional representation of vibration signals as input. Experimental results showed that the proposed method has a desirable 99.93% classification accuracy in the case of less labeled data from the public data set of West Reserve University, which is better than the state-of-the-art methods.
Highlights
In the past, fault diagnosis of the bearing component has mainly relied on listening sticks to contact the bearing component, which is extremely demanding for the technician who is responsible for hearing the difference between the norm and failure
We propose a semi-supervised generative adversarial network based on switchable normalization (SN-SSGAN) to learn the latent features of the raw bearing vibration signal and to distinguish the fault categories, which consists of a G and a D
The architecture of the D is based on the architecture of CNN in WDCNN [29] and deletes the pooling layer, which means the difference between SN-SSGAN and WDCNN lies in the following two points: (1) SN-SSGAN is based on switchable normalization, but WDCNN is based on batch normalization, and (2) SN-SSGAN is an adversarial network, that is, SN-SSGAN has one more G than WDCNN
Summary
Fault diagnosis of the bearing component has mainly relied on listening sticks to contact the bearing component, which is extremely demanding for the technician who is responsible for hearing the difference between the norm and failure. Zhang proposed a supervised bearing fault diagnosis method based on convolution. Li proposed a bearing fault diagnosis method based on a fully-connected winner-take-all auto encoder, which used unsupervised learning to get bearing features by using an auto encoder, supervised fine tuning to optimize the model, and a Softmax classifier was used to classify faults, because the loss of the auto encoder is based on the re-built error, which make images more or less fuzzy. Following the idea of reducing the number of training samples while maintaining ideal performance, we consider semi-supervised generative adversarial networks based switchable normalization. We propose a semi-supervised generative adversarial network based on switchable normalization (SN-SSGAN) to learn the latent features of the raw bearing vibration signal and to distinguish the fault categories, which consists of a G and a D.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
