Prior knowledge-based residuals shrinkage prototype networks for cross-domain fault diagnosis

Abstract

In engineering practice, device failure samples are limited in the case of unexpected catastrophic faults, thereby limiting the application of deep learning in fault diagnosis. In this study, we propose a prior knowledge-based residual shrinkage prototype network (PK-RSPN) to resolve the fault diagnosis challenges under limited labeled samples. First, our method combines general supervised learning and metric meta-learning to extract prior knowledge from the labeled source data by utilizing a denoised residual shrinkage network (RSN). Further, the knowledge extracted from the supervised learning is used for prototype metric training to achieve a better feature representation for the fault diagnosis. Finally, our approach outperforms a series of baseline methods in the few-shot cross-domain diagnostic task on the gearbox and bearing datasets. A diagnosis accuracy of more than 95% has been achieved in a variety of working conditions for diagnostic tasks, which is far higher than the existing basic method.

Semi-supervised adaptive anti-noise meta-learning for few-shot industrial gearbox fault diagnosis

“…ProNet [36] uses distance measures to determine the similarity between classification tasks and sample prototypes. In the training of the meta-task, the support set samples are used to generate feature vectors through the feature extractor f φ (•), where φ represents the model parameters.…”

“…To better evaluate the effectiveness of our proposed method in few-shot tasks, we compare the proposed ASMN with several basic methods: traditional supervised learning: CNN, ResNet [38], deep neural network based on domain adaptation in fault diagnosis (DAFD) [39], meta-learning ProNet [36], semi-supervised SSMN [33]. Table 3 lists the details of the comparison methods.…”

A hierarchical transformer-based adaptive metric and joint-learning network for few-shot rolling bearing fault diagnosis

“…Fortunately, the introduction of meta-learning has simultaneously tackled the challenges of data scarcity and cross-domain, leading to satisfactory results in handling FSFD tasks [11]. Current applications in fault diagnosis are mainly in the areas of parameter optimization [12,13], similarity metrics [14][15][16], and modelbased [17] methods. Among them, the metric-based methods are widely adopted because of their stronger generalization capability and portability.…”