Multiscale dynamic fusion prototypical cluster network for fault diagnosis of planetary gearbox under few labeled samples

Li, Biao; Tang, Baoping; Deng, Lei; Yu, Xiaoxia

doi:10.1016/j.compind.2020.103331

Cited by 38 publications

(7 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, prototype learning has also made progress in fault diagnosis tasks using a few labeled samples. For example, Li et al [6] proposed a multi-scale dynamic fusion prototypical cluster to compensate for the weak ability of a ProNet to extract mechanical vibration signals from a few labeled samples. Yu et al [31] proposed a hybrid attention structure to improve the ProNet to fully learn weak features in a few labeled fault samples.…”

Section: Backbone Networkmentioning

confidence: 99%

“…discriminative fault features from high-dimensional data [3][4][5], which do not require accurate physical models and rich experience in signal analyses, have recently become mainstream methods for wind gearbox fault diagnosis [6,7]. Deep strong-supervised learning algorithms, such as deep auto-encoders [8,9], deep belief networks [10], convolutional neural networks (CNNs) [11,12], recurrent neural networks (RNNs) [13], and residual networks (ResNets) [14,15], have been widely applied to the task of fault diagnosis for WT gearboxes and have achieved high diagnostic accuracy.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Twin prototype networks with noisy label self-correction for fault diagnosis of wind turbine gearboxes

Liu¹,

Tang²,

Li³

et al. 2022

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Deep strong-supervised learning-based methods have been widely used and have made significant progress in intelligent fault diagnosis for wind turbine (WT) gearboxes. The superior performance of such methods relies on high-quality labels. However, correctly labeling the data is challenging because of the complexity of fault vibration signals and fault modes in real industrial scenarios, resulting in noisy labels in datasets, which significantly restricts the application of strong-supervised fault diagnosis models. In this study, a method based on twin prototype networks with noisy label self-correction was proposed to address fault diagnosis for WT gearboxes with noisy labels. This method introduced a collaborative learning architecture to improve the confirmation bias in the self-training of individual networks and to slow the speed of learning noisy-labeled samples. Simultaneously, the loss distribution of the samples from each network was modeled using the Gaussian mixture model to dynamically identify mislabeled samples in the training dataset. Finally, a collaborative relabeling prototype refinement module was designed to optimize the prototype learning process and enable self-correction of noisy labels. The experiments demonstrated the effectiveness and superiority of the proposed method.

show abstract

Section: Backbone Networkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Twin prototype networks with noisy label self-correction for fault diagnosis of wind turbine gearboxes

Liu¹,

Tang²,

Li³

et al. 2022

Meas. Sci. Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Therefore, the application of metricbased meta-learning approach to bearing fault identification can obtain remarkable results. Li et al [26] presented a cluster method to refine the prototype. The problem of small sample bearing fault identification was effectively addressed by Wang et al [27] through employing a fusion of Siamese net and prototypical net.…”

Section: Introductionmentioning

confidence: 99%

Improved metric-based meta learning with attention mechanism for few-shot cross-domain train bearing fault diagnosis

Zhong,

He,

Lao

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Traction motor bearings, as a crucial component of subway trains, play a pivotal role in ensuring the safety of train operations. Therefore, intelligent diagnosis of train bearings holds significant importance. However, due to the complex and dynamic nature of bearing conditions coupled with limited fault data availability, traditional diagnostic methods fail to yield satisfactory results. To address this issue, we propose an improved metrics-based meta-learning (IMML) approach for accurate few-shot cross-domain fault diagnosis of train bearings. Firstly, we introduce a 1D-signal channel attention mechanism that effectively extracts latent features and enhances recognition accuracy. Secondly, by incorporating the Adabound algorithm into our model framework, we further enhance its classification performance. Finally, through several case studies, we demonstrate the effectiveness of our proposed method in comparison to other approaches within similar settings.

show abstract

“…Few-shot learning is a kind of meta-learning [ 28 ]. In recent years, scholars have achieved many results in the field of meta-learning, mainly including initialization-based models, such as Model-agnostic meta-learning [ 29 , 30 ], and metric-based models, such as Siamese networks [ 23 , 31 , 32 , 33 ], matching networks [ 34 , 35 ], prototypical networks [ 36 , 37 , 38 , 39 ], etc. All of these models have good cross-domain performance and high accuracy.…”

Section: Introductionmentioning

confidence: 99%

“…Wang H et al proposed Deep Prototypical Networks, which combine the advantages of the prototypical network and the Siamese network, using the Siamese structure to extract features, and then use the prototype learning method to map this to the feature space [ 37 ]. Li B et al use multi-scale dynamic fusion to extract features and improve the clustering method of the prototypical network for intelligent diagnosis of planetary gearbox [ 36 ]. The above studies have achieved good results, but they all use only the time domain data of the raw signal, whereas the frequency domain information is more sensitive to some faults [ 38 ].…”

Section: Introductionmentioning

confidence: 99%

Wavelet-Prototypical Network Based on Fusion of Time and Frequency Domain for Fault Diagnosis

Wang

Chen

Liu

et al. 2021

Sensors

View full text Add to dashboard Cite

Neural networks for fault diagnosis need enough samples for training, but in practical applications, there are often insufficient samples. In order to solve this problem, we propose a wavelet-prototypical network based on fusion of time and frequency domain (WPNF). The time domain and frequency domain information of the vibration signal can be sent to the model simultaneously to expand the characteristics of the data, a parallel two-channel convolutional structure is proposed to process the information of the signal. After that, a wavelet layer is designed to further extract features. Finally, a prototypical layer is applied to train this network. Experimental results show that the proposed method can accurately identify new classes that have never been used during the training phase when the number of samples in each class is very small, and it is far better than other traditional machine learning models in few-shot scenarios.

show abstract

Multiscale dynamic fusion prototypical cluster network for fault diagnosis of planetary gearbox under few labeled samples

Cited by 38 publications

References 24 publications

Twin prototype networks with noisy label self-correction for fault diagnosis of wind turbine gearboxes

Twin prototype networks with noisy label self-correction for fault diagnosis of wind turbine gearboxes

Improved metric-based meta learning with attention mechanism for few-shot cross-domain train bearing fault diagnosis

Wavelet-Prototypical Network Based on Fusion of Time and Frequency Domain for Fault Diagnosis

Contact Info

Product

Resources

About