Bearing Fault Diagnosis Method Based on Multidomain Heterogeneous Information Entropy Fusion and Model Self-Optimisation

Table 9

Comparison of diagnostic performances of different models.

Methods	Feature sets	CWRU test accuracy (%)	CWRU test time (s)	IMS test accuracy (%)	IMS test time (s)
	CWT	97.42 ± 0.22	0.27	96.75 ± 0.15	0.24
ST-CNN	GADF	94.64 ± 1.33	0.42	95.41 ± 0.72	0.33
	FD	94.79 ± 0.81	0.31	93.05 ± 0.28	0.27
	CWT	93.78 ± 1.54	0.68	92.66 ± 2.41	0.53
DBN	GADF	92.43 ± 1.80	0.91	92.11 ± 2.79	0.94
	FD	93.33 ± 2.04	0.85	92.47 ± 0.68	0.77
	CWT	94.14 ± 1.47	0.33	90.42 ± 1.87	0.29
CNNEPDNN	GADF	92.70 ± 1.66	0.71	92.02 ± 0.28	0.64
	FD	92.16 ± 0.95	0.44	92.69 ± 0.77	0.45
SVM	Time	89.47 ± 1.17	0.16	90.22 ± 0.46	0.07
	Frequency	90.35 ± 0.75	0.14	90.64 ± 0.70	0.16
PSO-LSSVM	Time	94.50 ± 0.13	0.57	94.32 ± 0.58	0.44
	Frequency	94.80 ± 0.34	0.70	94.45 ± 0.14	0.26
BP	Time	89.66 ± 1.21	0.21	88.27 ± 2.63	0.10
	Frequency	91.34 ± 0.25	0.24	91.85 ± 1.20	0.25
Our method	CWT + GADF	99.88 ± 0.10	0.30	99.60 ± 0.29	0.21