Meta-Learning With Learnable Margin Triplet Loss for Few-Shot Fault Diagnosis

The application of deep learning (DL) based methods for machinery and equipment fault diagnosis has yielded significant results. However, the success of DL is contingent upon the availability of large amounts of labelled data, which is often limited due to the scarcity and value of fault samples. Even neural networks with advanced structures face challenges in performing effective diagnosis with few samples. To address this issue, we develop a novel meta-learning with learnable margin triplet loss (MLLMTL) for few-shot fault diagnosis. The method constructs meta-learning tasks by blending multi-source domain data. Cross-platform and variable operating conditions meta-tasks can better enhance the learning capability of the model. Furthermore, a novel learnable margin triplet loss function (LMTL) is proposed in this paper. Firstly, conventional logical margins are transformed into learnable physical margins and the relationship between positive and negative samples are transformed into an independent relationship with the margins. Secondly, we force the similarity of dissimilar anchor features to maintain a large margin by using the feature distance between any two anchors as a loss term. In comparison to the conventional triplet loss function, the LMTL has the capacity to facilitate a more nuanced differentiation of distinct fault features within intricate scenarios. Results on two different diagnostic tasks demonstrate that both proposed methods outperform several state-of-the-art few-shot fault diagnosis methods.