Table 15 Comparison of different image encoders on CWRU and Ottawa datasets (%).

From: TFDFNet: a dual-branch fault diagnosis model for bearings under noisy and complex industrial environments

Image Encoder

CWRU

Ottawa

CNN (3-layer)

93.82

86.45

ResNet18

94.34

87.12

Swin Transformer (ours)

96.57

89.73

Back to article page