A Multimodel Decision Fusion Method Based on DCNN-IDST for Fault Diagnosis of Rolling Bearing

Table 6

The actual outputs of the three network models are compared with the expected results.

Models	Actual outputs				Expect results	Diagnosis results
BPNN-FL1	0.000000	0.960499	0.000000	0.000045	0,1,0,0	Inner ring fault
	0.000000	0.986047	0.000002	0.000483	0,1,0,0	Inner ring fault
	0.000026	0.414734	0.408441	0.068637	0,1,0,0	Uncertainty
	0.000000	0.982463	0.004014	0.030370	0,1,0,0	Inner ring fault
	0.000000	0.925475	0.000000	0.000172	0,1,0,0	Inner ring fault
	0.000000	0.725717	0.000043	0.003560	0,1,0,0	Inner ring fault
RBFNN-FL1	−0.349173	0.283555	0.650220	0.415404	0,1,0,0	Uncertainty
	−0.048801	1.123021	0.128835	−0.203042	0,1,0,0	Inner ring fault
	0.221095	0.833929	0.189633	−0.244645	0,1,0,0	Inner ring fault
	−0.441047	0.937030	0.526606	−0.022593	0,1,0,0	Inner ring fault
	−0.041985	1.174574	−0.290296	0.157707	0,1,0,0	Inner ring fault
	0.312547	1.065893	−0.560675	0.182236	0,1,0,0	Inner ring fault
DCNN-FL1	0.000067	0.552400	0.446572	0.000961	0,1,0,0	Uncertainty
	0.000000	0.989853	0.000146	0.000001	0,1,0,0	Inner ring fault
	0.000000	0.940449	0.059551	0.000000	0,1,0,0	Inner ring fault
	0.000000	0.981931	0.018045	0.000025	0,1,0,0	Inner ring fault
	0.000000	0.976631	0.000069	0.000000	0,1,0,0	Inner ring fault
	0.000000	0.973404	0.026596	0.000001	0,1,0,0	Inner ring fault