In practical industrial applications, rolling bearing fault samples are often scarce and costly to annotate, making small-sample fault diagnosis challenging. To address this issue, this paper proposes a frequency-aware time–frequency representation learning framework, named Multi-branch Frequency Enhancement Temporal Attention Network (MFETA-Net). In the proposed framework, dual-channel vibration signals are first transformed into time–frequency representations using the Short-Time Fourier Transform (STFT). Then, a multi-branch frequency enhancement encoder is used to extract local frequency-band patterns, cross-band correlations, and frequency variation features. A temporal-frequency dependency modeling mechanism preserves the correspondence between temporal positions and frequency distributions during sequential modeling, while a temporal attention aggregation module emphasizes diagnostically important regions. Extensive experiments on the CWRU and HUST bearing datasets show that MFETA-Net achieves accuracies of 77.26% and 79.60% under the smallest training setting, respectively, indicating its capability to learn discriminative fault representations from limited labeled samples. Ablation studies further verify the effectiveness of each proposed module, while noise experiments confirm the robustness of the proposed framework under controlled noisy conditions.
Automatic Identification System (AIS)-based vessel trajectory prediction is essential for maritime traffic management and navigation safety. Existing deep learning methods typically model vessel motion within a unified temporal representation space, which may entangle long-term navigation trends with local maneuvering behaviors. However, vessel trajectories inherently exhibit heterogeneous dynamics, including steady route evolution and non-stationary maneuver perturbations. To address this issue, this paper proposes MD-EDTCNFormer, a motion-decoupled dual-stream framework for vessel trajectory prediction. A Global Navigation Dynamics Encoder is designed to capture dominant route-level temporal evolution from raw AIS sequences, while a Residual Maneuver Dynamics Encoder explicitly models maneuver-related local perturbations through state transition residual representations. In addition, a state-adaptive motion aggregation mechanism is introduced to dynamically balance global navigation dependencies and local maneuver-aware dynamics under different navigation states. Depthwise separable temporal convolution and efficient channel attention are further integrated to suppress redundant temporal-channel coupling and emphasize dynamically dominant motion cues. Experiments on a real-world AIS dataset from the Zhoushan coastal area demonstrate the effectiveness of the proposed framework under coastal traffic conditions, and show improvements in prediction accuracy and trajectory stability compared with representative baseline methods.
Chiming Wang, Dongke Zheng, Yiying Zhou et al.· Journal of Marine Science an...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
