A New Generative Neural Network for Bearing Fault Diagnosis with Imbalanced Data

Table 7

Performance of different deep learning methods in sixteen-classes diagnosis.

Sixteen		Precision	Recall	F1-score	AUC
GCNN	Case 2	0.746	0.690	0.717	0.993
	Case 3	0.761	0.723	0.742	0.985
	Case 4	0.765	0.709	0.736	0.973
	Case 5	0.801	0.745	0.772	0.969
	Case 6	0.705	0.576	0.634	0.935
	Case 7	0.672	0.516	0.584	0.933
WCNN	Case 2	0.878	0.869	0.873	0.979
	Case 3	0.834	0.831	0.832	0.978
	Case 4	0.796	0.783	0.789	0.957
	Case 5	0.815	0.815	0.815	0.944
	Case 6	0.774	0.765	0.769	0.917
	Case 7	0.757	0.723	0.739	0.910
NCNN	Case 2	0.757	0.676	0.714	0.975
	Case 3	0.797	0.708	0.749	0.978
	Case 4	0.679	0.570	0.619	0.953
	Case 5	0.645	0.545	0.591	0.921
	Case 6	0.684	0.535	0.601	0.892
	Case 7	0.655	0.477	0.552	0.875
FNN	Case 2	0.449	0.303	0.362	0.810
	Case 3	0.489	0.335	0.397	0.833
	Case 4	0.375	0.248	0.298	0.756
	Case 5	0.370	0.250	0.299	0.776
	Case 6	0.381	0.381	0.381	0.761
	Case 7	0.500	0.310	0.382	0.792