Fault Diagnosis of Gearbox in Multiple Conditions Based on Fine-Grained Classification CNN Algorithm

Table 4

Data description of CNN-FG.

Working condition	Health state	Sample number	Test/train sample proportion	Health state label	Fine-grained class label	Coarse class label
Condition 1	Normal	600	20%/80%	0	0	1
	Fault 1	600	20%/80%	1	1
	Fault 2	600	20%/80%	2	2
	Fault 3	600	20%/80%	3	3
	Fault 4	600	20%/80%	4	4
Condition 2	Normal	600	20%/80%	0	5	2
	Fault 1	600	20%/80%	1	6
	Fault 2	600	20%/80%	2	7
	Fault 3	600	20%/80%	3	8
	Fault 4	600	20%/80%	4	9
Condition 3	Normal	600	20%/80%	0	10	3
	Fault 1	600	20%/80%	1	11
	Fault 2	600	20%/80%	2	12
	Fault 3	600	20%/80%	3	13
	Fault 4	600	20%/80%	4	14
Condition 4	Normal	600	20%/80%	0	15	4
	Fault 1	600	20%/80%	1	16
	Fault 2	600	20%/80%	2	17
	Fault 3	600	20%/80%	3	18
	Fault 4	600	20%/80%	4	19