Open access
Oct 2026
Learnable Band-Selective Envelope-Spectrum Residual Fusion for Bearing Fault Diagnosis Under Matched Noise Conditions
This study proposes a learnable band-selective envelope fusion WD-CNN (LBEF-WDCNN), where a raw-waveform branch extracts temporal features, whereas an auxiliary branch applies four differentiable Gaussian masks, reconstructs the corresponding spectrally selected waveforms.
Shi-Qian Wu, Wei-Min Zhang, Yu-Rui Sang et al.
· Machines · 0 citations