Deep Adaptive Adversarial Network-Based Method for Mechanical Fault Diagnosis under Different Working Conditions

Table 3

Diagnosis results in Experiment 3.

Method

G ⟶ H

I ⟶ H

J ⟶ H

H ⟶ G

I ⟶ G

J ⟶ G

G ⟶ I

H ⟶ I

J ⟶ I

G ⟶ J

H ⟶ J

I ⟶ J

SAE

90.24 ± 1.35%

91.01 ± 1.04%

85.04 ± 2.84%

92.41 ± 1.54%

89.95 ± 0.85%

81.15 ± 2.08%

92.68 ± 0.78%

88.47 ± 0.92%

91.04 ± 0.21%

92.87 ± 0.63%

91.63 ± 1.03%

90.25 ± 1.65%

TCA

96.02 ± 0.84%

96.04 ± 0.75%

88.97 ± 2.43%

96.24 ± 0.69%

95.10 ± 1.02%

84.36 ± 3.46%

95.42 ± 0.96%

96.16 ± 0.74%

92.47 ± 2.04%

94.63 ± 1.12%

95.84 ± 0.98%

94.95 ± 1.32%

DANN

98.75 ± 0.24%

98.86 ± 0.08%

97.58 ± 0.77%

99.12 ± 0.02%

99.08 ± 0.04%

97.24 ± 0.11%

99.67 ± 0.02%

99.55 ± 0.01%

99.88 ± 0.01%

99.25 ± 0.12%

98.14 ± 0.35%

99.24 ± 0.02%

MK-MMD

100%

100%

96.95 ± 1.02%

100%

100%

95.86 ± 1.63%

100%

100%

100%

99.75 ± 0.25%

99.87 ± 0.06%

100%

SFDA

100%

99.95 ± 0.02%

98.62 ± 0.16%

100%

99.96 ± 0.01%

98.70 ± 0.12%

100%

100%

100%

99.98 ± 0.01%

100%

100%

DAAN

100%

100%

98.71 ± 0.08%

100%

100%

98.82 ± 0.17%

100%

100%

100%

99.92 ± 0.03%

100%

100%