Deep imbalanced domain adaptation for transfer learning fault diagnosis of bearings under multiple working conditions

Ding, Yifei; Jia, Minping; Zhuang, Jichao; Cao, Yudong; Zhao, Xiaoli; Lee, Chi-Guhn

doi:10.1016/j.ress.2022.108890

Cited by 86 publications

(26 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While convolutional neural networks (CNNs) have gained popularity, their ability to process one-dimensional time-series signals, such as those emitted by rotating machinery, remains suboptimal (Hu and Wang, 2019; Zhao and Chen, 2022). Long short-term memory (LSTM) networks have emerged as a robust mechanism for modeling spatial and temporal signals simultaneously, thereby improving the classification and predictive accuracy of time-series signals (Gu et al , 2020; Kong et al , 2022; Ding et al , 2023).…”

Section: Introductionmentioning

confidence: 99%

Interactive spatiotemporal LSTM approach for enhanced industrial fault diagnosis

Zhang,

Huang,

et al. 2024

ILT

View full text Add to dashboard Cite

Purpose Rotating machinery is a crucial component of large equipment, and detecting faults in it accurately is critical for reliable operation. Although fault diagnosis methods based on deep learning have been significantly developed, the existing methods model spatial and temporal features separately and then weigh them, resulting in the decoupling of spatiotemporal features. Design/methodology/approach The authors propose a spatiotemporal long short-term memory (ST-LSTM) method for fault diagnosis of rotating machinery. The authors collected vibration signals from real rolling bearing and gearing test rigs for verification. Findings Through these two experiments, the authors demonstrate that machine learning methods still have advantages on small-scale data sets, but our proposed method exhibits a significant advantage due to the simultaneous modeling of the time domain and space domain. These results indicate the potential of the interactive spatiotemporal modeling method for fault diagnosis of rotating machinery. Originality/value The authors propose a ST-LSTM method for fault diagnosis of rotating machinery. The authors collected vibration signals from real rolling bearing and gearing test rigs for verification.

show abstract

Section: Introductionmentioning

confidence: 99%

Interactive spatiotemporal LSTM approach for enhanced industrial fault diagnosis

Zhang,

Huang,

et al. 2024

ILT

View full text Add to dashboard Cite

show abstract

“…Luo et al [33] proposed an enhanced deep subdomain adaptation network, achieving transfer learning for bearing fault diagnosis under different rotational speed conditions. Ding et al [34], focusing on the issue of feature shift under different speeds, introduced a novel deep imbalanced domain adaptation framework to tackle the transfer learning problem in bearing fault diagnosis across different speeds. While transfer learning for bearing fault diagnosis has attracted attention among scholars, further research is needed to better understand how to narrow the gap between the source and target domains.…”

Section: Introductionmentioning

confidence: 99%

Transfer learning for bearing fault diagnosis: adaptive batch normalization and combined optimization method

Li,

Su,

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Bearings are crucial components in rotating machinery equipment. Bearing fault diagnosis plays a significant role in the maintenance of mechanical equipment. In practical industrial settings, equipment conditions often vary continuously, making it challenging to collect data for all operating conditions for bearing fault diagnosis. This paper proposes a transfer learning approach for bearing fault diagnosis based on Adaptive Batch Normalization (AdaBN) and a combined optimization algorithm. Initially, a ResNet neural network is trained using source domain data. Subsequently, the trained model is transferred to the target domain, where AdaBN is applied to mitigate domain shift issues. Furthermore, a combined optimization algorithm is employed during model training to enhance fault diagnosis accuracy. Experimental validation is conducted using bearing data from the CWRU dataset and NEFU dataset. Comparison shows that AdaBN and the combined optimization algorithm improve bearing fault diagnosis accuracy effectively. On the NEFU dataset, the diagnostic accuracy exceeds 95%.

show abstract

“…The second method is the Cost-Sensitive (CS) learning method, which can solve the problem of data imbalance in classification duties and has been successfully used for various traditional DL engineering tasks [27]- [30]. The third method is solving the problem of poor identification accuracy regarding equipment status caused by the scarcity of available data in the actual engineering environment via Transfer Learning (TL) [31]- [33]. However, based on oversampling data processing methods balance the datasets just by increasing the number of fault samples, the sample information is not increased, which may lead to over-fitting of the model [34].…”

Section: Introductionmentioning

confidence: 99%

Fault Diagnosis Method for Imbalanced Data of Rotating Machinery Based on Time Domain Signal Prediction and SC-ResNeSt

et al. 2023

View full text Add to dashboard Cite

In an actual engineering environment, some rotating machines are usually in normal operation, but their time in a fault state is very short, which leads to a serious imbalance in the fault diagnosis datasets for rotating machinery, and gives the traditional network model the shortcomings of poor stability and low accuracy in practical engineering applications. To solve this problem, we propose a fault diagnosis method based on the combination of a new Dual-stage Attention-based Recurrent Neural Network (DA-RNN) and depth residual dispersion self-calibration convolution network (SC-ResNeSt). Firstly, a novel DA-RNN network with a gated cycle unit (GRU) as a coding-decoding unit was designed, and the network was used to predict and expand the scarce fault signals. Secondly, to make full use of the time domain information of vibration signals, a new image coding method, namely, Gram Angle Product Field (GAPF), was proposed. Then, because the traditional convolution layer lacks a dynamic receptive field to extract more representative features, self-calibrated convolution modules were introduced on the basis of the distraction network (ResNeSt), and a new network model, SC-ResNeSt, was established. Finally, the expanded vibration signal is converted into GAPF, which is used as the input for the SC-ResNeSt network to classify the fault types. To check the performance of the model, the Case Western Reserve University rolling bearing dataset and planetary gearbox dataset were used for testing. Ultimately, good results were obtained in a prediction experiment for bearing fault samples and a subsequent fault diagnosis experiment for bearings and gears, which verified the feasibility and practicability of the model.

show abstract

Deep imbalanced domain adaptation for transfer learning fault diagnosis of bearings under multiple working conditions

Cited by 86 publications

References 36 publications

Interactive spatiotemporal LSTM approach for enhanced industrial fault diagnosis

Interactive spatiotemporal LSTM approach for enhanced industrial fault diagnosis

Transfer learning for bearing fault diagnosis: adaptive batch normalization and combined optimization method

Fault Diagnosis Method for Imbalanced Data of Rotating Machinery Based on Time Domain Signal Prediction and SC-ResNeSt

Contact Info

Product

Resources

About