This paper proposes a highly efficient framework, the Frequency-domain Circulant Attention Vision Transformer (FC-ViT), for robust rotating machinery monitoring, and demonstrates superior noise immunity and cross-load generalization.
Abstract
Deep learning models often struggle with complex noise and variable loads in industrial fault diagnosis. This paper proposes a highly efficient framework, the Frequency-domain Circulant Attention Vision Transformer (FC-ViT), for robust rotating machinery monitoring. Raw vibration signals are first converted into 2D time-frequency representations via Continuous Wavelet Transform (CWT). FC-ViT then utilizes an improved Frequency-domain Circulant Attention mechanism, which achieves a log-linear computational complexity of $O(N \log N)$ , to isolate fault-related impulses from heavy stochastic and non-Gaussian impulsive noise. Validated on the Case Western Reserve University (CWRU) and Paderborn University (PU) datasets, FC-ViT achieves 100% accuracy under noise-free conditions. At an extreme −5 dB SNR, it maintains 89.4%–98.3% accuracy, significantly outperforming both state-of-the-art 1D diagnostic networks and generic 2D vision models (e.g., Swin-Transformer, ConvNeXt, and CA-DeiT). These results demonstrate superior noise immunity and cross-load generalization. Furthermore, comprehensive hardware deployment evaluations confirm its ultra-low inference latency and minimal memory footprint, providing a practical and real-time solution for real-world industrial condition monitoring.
Introduction Rolling bearing fault diagnosis is essential for predictive maintenance because bearing failures can cause unexpected downtime, increased maintenance costs, and safety risks. Conventional signal-domain and image-based approaches may not adequately capture both local frequency characteristics and global structural relationships among fault patterns. Methods This study proposes a frequency-aware signal image representation and graph transformer learning (FSI-GTL) framework for bearing fault diagnosis. Raw vibration signals are transformed using short-time Fourier transform (STFT), wavelet packet transform (WPT), empirical mode decomposition (EMD), and cepstral analysis. The resulting representations are adaptively fused, and discriminative frequency-domain features are used to construct graph representations. A graph neural network (GNN) captures local spectral topology, while a Transformer encoder models long-range dependencies through self-attention. Results The proposed framework achieved an average classification accuracy of 99.42% on the CWRU dataset. Cross-dataset evaluation on the Paderborn University bearing dataset achieved an average accuracy of 98.71%, demonstrating strong generalization across datasets. The framework also showed high precision, recall, and F1-score across the evaluated bearing fault categories. Discussion The results demonstrate that integrating frequency-aware signal representations, graph-based structural learning, and Transformer-based global contextual modeling can effectively improve bearing fault discrimination. The proposed FSI-GTL framework provides a promising approach for intelligent condition monitoring and predictive maintenance.
Sandhya Mitta, P. Nedunchezhian· Frontiers in Artificial Inte...· 0 citations
A hybrid deep learning architecture is proposed for robust vibration-based fault diagnosis in industrial machinery by jointly modeling time-domain, frequency-domain, and temporal dynamics, enabling coherent cross-domain interaction and robust fault characterization.
Canan Taştimur· Information Technology and C...· 0 citations
Experiments on the CWRU bearing dataset across seven signal-to-noise ratio levels demonstrate that the proposed method achieves 99.75% accuracy under clean conditions and maintains 92.50% at -4 dB SNR, representing a 7.25 percentage-point improvement over the single-domain baseline.
A novel diagnosis method integrating FFT-VMD feature extraction with a Bi-TCN-Bi-GRU neural network, which maintains an exceptional diagnostic accuracy even under severe background noise.
The proposed MSFormer incorporates a parallel multi-scale Convolutional Neural Network architecture and hierarchical Transformer modules to comprehensively process 1D vibration signals to provide a powerful and precise intelligent solution for mechanical fault diagnosis.
Shu Guo, Jin Li, Tian-Ci Zhang· Machines· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
