Research on Intelligent Fault Diagnosis of Rolling Bearing Based on Improved ShufflenetV2-LSTM

Table 3

Results of ablation experiments with data from CWRU.

Model	Factor				Training accuracy (%)	Test accuracy (%)	Time (min)
	LSTM	Optimizer	Dropout	Activation function
ShufflenetV2	0	0	0	0	92.19	88.25	942
	1	0	0	0	98.44	95.63	957
	0	1	0	0	100	94	944
	0	0	1	0	91.88	90.75	2039
	0	0	0	1	94.44	89.37	1010
This study	1	1	1	1	100	97.38	923