A Hybrid Deep Learning Prediction Method of Remaining Useful Life for Rolling Bearings Using Multiscale Stacking Deep Residual Shrinkage Network

Table 7

Comparison of evaluation metric results.

Testing bearing		MSDRSN			BiLSTM			MSCNN-FC
		MSE	MAE	R^2	MSE	MAE	R^2	MSE	MAE	R^2
1-3	2 db	0.005	0.042	0.954	0.102	0.243	0.418	0.047	0.186	0.537
	4 db	0.003	0.028	0.975	0.078	0.141	0.568	0.050	0.126	0.503
	6 db	0.004	0.054	0.956	0.115	0.548	0.432	0.053	0.122	0.477
	8 db	0.004	0.036	0.962	0.256	0.106	0.566	0.035	0.068	0.659
2-3	2 db	0.004	0.012	0.625	0.015	0.275	0.578	0.007	0.720	0.551
	4 db	0.005	0.016	0.503	0.032	0.349	0.498	0.008	0.135	0.256
	6 db	0.003	0.014	0.653	0.048	0.217	0.651	0.008	0.057	0.146
	8 db	0.003	0.014	0.657	0.041	0.136	0.557	0.009	0.060	0.093
3-3	2 db	0.003	0.026	0.971	0.033	0.159	0.667	0.015	0.076	0.851
	4 db	0.004	0.054	0.964	0.006	0.047	0.941	0.011	0.063	0.892
	6 db	0.002	0.043	0.976	0.046	0.189	0.537	0.025	0.144	0.744
	8 db	0.004	0.054	0.963	0.024	0.102	0.759	0.018	0.077	0.811

The bold values in the table are the values of the proposed method, which is convenient and clearer to compare with the values of other methods.