Transfer learning based on improved stacked autoencoder for bearing fault diagnosis

Luo, Shuyang; Huang, Xu‐Feng; Wang, Yanzhi; Luo, Rongmin; Zhou, Qi

doi:10.1016/j.knosys.2022.109846

Cited by 61 publications

(14 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are still some limitations to be explored in future work: (1) the redundancy in the iteration of MVMD has a significant impact on its efficiency. Therefore, further research is necessary to investigate this issue thoroughly [49]; (2) this paper only uses an open-source dataset to evaluate the proposed method, the proposed method will be further tested on laboratory datasets with gearbox or pump failure data [15]; (3) in engineering applications, the rotary machines often operate under normal conditions with few faults. Thus, it is hard to obtain enough training data of health conditions from engineering scenarios to support the training of 3D-HFN.…”

Section: Discussionmentioning

confidence: 99%

“…In addition to shallow ML methods, deep learning (DL) approaches have been widely used in prognostics and health management [11]. Typical DLbased diagnostic algorithms include convolutional neural networks (CNN) [12], recurrent neural networks [13], generative adversarial networks [14], stacked autoencoders [15], domain adaptation network [16], self-attention network [17] etc. These DL-based methods can automatically extract the implicit features from raw data, which can obtain better diagnostic performance than those shallow ML approaches using hand-crafted features.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Three-dimensional hybrid fusion networks for current-based bearing fault diagnosis

Huang,

Xie,

et al. 2023

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Intelligent fault diagnosis (IFD) techniques commonly use vibration-based measurements to perform health monitoring of critical rotating components in industrial systems. However, these vibration-based approaches may be limited in cost-sensitive applications, because the installation of vibration sensors is inconvenient and vibration sensors are expensive. Considering the difficulties of IFD using only current-related information from the motor current signal (MCS), this paper proposes a three-dimensional hybrid-fusion neural network (3D-HFN) that can automatically perform both data- and feature-level fusion of multi-phase current signals for MCS-based IFD of the rolling bearing. The 3D-HFN consists of the multivariate variational mode decomposition (MVMD) and an improved three-dimensional convolution neural network (3D-CNN). Firstly, MVMD is proposed to process multi-phase current signals, which adaptively acquire several intrinsic mode functions with mode-alignment properties. Subsequently, signal-to-image conversion and 3D stacking methods are used to construct 3D-like data in the current-phase dimension, which can fully preserve the interaction relationship between different phases using data-level fusion. Finally, an improved 3D-CNN with multiscale feature fusion and the smooth maximum unit is proposed to learn the 3D-like data and identify different health conditions for the rolling bearing. An open-source dataset with composite bearing faults is used to validate the merits of the proposed method. Experimental results show that the proposed approach has achieved more reliable diagnosis performance than other hand-crafted or 2D/3D-CNN-based algorithms in MCS-based IFD of the rolling bearing.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Three-dimensional hybrid fusion networks for current-based bearing fault diagnosis

Huang,

Xie,

et al. 2023

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The goal of transfer learning is to apply knowledge or models learned on a domain or task to a different but related domain or problem [9]. The existing methods of transfer learning can be classified as Knowledge Transfer [10,11] and Domain Adaptation [12][13][14][15]. The method of Knowledge Transfer transferring the knowledge learned from the source domain model to the target domain, is more suitable for edge-cloud collaboration, and can rapidly adapt to the diagnostic requirements of an individual with a minimum cost.…”

Section: Transfer Learningmentioning

confidence: 99%

Edge-cloud collaborative transfer learning based on stacked sparse autoencoder with data regularization

Li,

Zhao,

Cui

et al. 2024

Third International Conference on Advanced Manufacturing Technology and Electronic Information (AMTEI 2023)

View full text Add to dashboard Cite

Edge-cloud collaboration provides a better solution for condition monitoring, which can reduce response time while maintaining computational efficiency. In practical condition monitoring scenarios, the individual differences among equipment often decrease the accuracy of diagnostic models. To tackle this problem, a transfer learning method based on stacked sparse autoencoder is proposed, which employs a data regularization strategy to improve feature extraction ability. The fault diagnosis model trained in the cloud transfers its model parameters and structure to the edge side. By a finetuning process with a small amount of data, and the model is further updated for condition monitoring of the individual machine. The experimental results show that the proposed KT-SAE method has improved transfer accuracy compared to other related transfer learning methods.

show abstract

“…Shao et al [23] transformed the raw signals into time-frequency maps by wavelet transform and fine-tuned the AlexNet model pre-trained on ImageNet to improve the recognition accuracy of bearings. Luo et al [24] developed a transfer diagnosis model using improved stacked autoencoder, combined with convolution shortcut and domain fusion strategy to realize fault detection of bearings. To further solve the distribution differences in cross-domain fault detection, domain adaptation (DA) provides a viable idea by searching for domain invariant features and domain discriminant classifiers.…”

Section: Introductionmentioning

confidence: 99%

Few-shot condition diagnosis of rolling bearing using adversarial transfer network with class aggregation-guided

Tian,

Zhen,

Sun

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

For the challenge of fault identification under limited labeled data in engineering applications,a novel adversarial transfer network with class aggregation-guided (ATN-CA) is proposed for few-shot condition diagnosis of bearings. The ATN-CA can focus on the discrepancy features of bearings by the proposed local discrepancy feature representation, avoiding that the features extracted by a single neural network may omit important fault information. Moreover, the further proposed class aggregation-guided strategy uses the semantic information of signals to guide the dynamic adaptation of marginal and conditional distributions of source and target data, which shortens the distribution distance of the same category in different domains, thus completing the transfer diagnosis. By comparing with other current methods on the artificial and real bearing fault datasets,results show this proposed method has the highest test precision and the smallest accuracy deviation in bearing transfer diagnosis.

show abstract

Transfer learning based on improved stacked autoencoder for bearing fault diagnosis

Cited by 61 publications

References 43 publications

Three-dimensional hybrid fusion networks for current-based bearing fault diagnosis

Three-dimensional hybrid fusion networks for current-based bearing fault diagnosis

Edge-cloud collaborative transfer learning based on stacked sparse autoencoder with data regularization

Few-shot condition diagnosis of rolling bearing using adversarial transfer network with class aggregation-guided

Contact Info

Product

Resources

About