Search papers, labs, and topics across Lattice.
This paper introduces Spectral Aliasing Pretext (SAP), a self-supervised learning approach that addresses the challenge of limited labeled data in machinery fault diagnosis by leveraging spectral aliasing in vibration signals. By undersampling signals to create a folded spectrum and training a Transformer to reconstruct the original spectrum, the method enables the model to learn frequency-domain invariants associated with mechanical faults. Experiments demonstrate that SAP achieves high classification performance with minimal labeled data, outperforming traditional fully supervised training methods in terms of stability and reliability.
Self-supervised learning can outperform fully supervised methods in machinery fault diagnosis, achieving high accuracy with significantly less labeled data.
Deep learning is a new way for machinery fault diagnosis but requires extensive labeled data, a scarce resource in industrial settings. We propose Spectral Aliasing Pretext (SAP), a self-supervised learning method that pretrains models on unlabeled vibration data by exploiting spectral aliasing. We deliberately undersample signals to create folded spectrum, then train a Transformer to reconstruct the original unfolded spectrum. This pretext task forces the model to learn frequency-domain invariants characteristic of mechanical faults, without potentially destructive augmentations. Experiments on the CWRU dataset show that SAP learns stable and highly discriminative representations. In a linear probing setting, SAP quickly achieves very high classification performance with only a small fraction of labeled data and low variance. In contrast, full fine-tuning, including fully supervised training, does not lead to more stable or better results. Overall, these findings suggest that SAP combined with linear probing can be more effective and reliable than fully supervised training for fault diagnosis with limited labeled data.
