Translation Invariance-Based Deep Learning for Rotating Machinery Diagnosis

Table 1

Performance comparison of the developed DCWT-SSAE method and wavelet-SSAE trained by different percentages of samples (unit: %).

	1%	2%	3%	5%	10%	25%	40%
Wavelet-SSAE	58.85	79.59	86.68	96.41	98.38	99.95	99.96
Wavelet-SAE	42.83	61.80	71.62	84.68	94.00	98.29	98.44
DCWT-SSAE	64.97	81.85	90.64	98.07	99.71	99.99	100