联邦协作框架下的跨工况半监督剩余使用寿命预测

李琦媛,程鑫,马文清,张开淦,夏唐斌,奚立峰

Federated collaborative framework-based semi-supervised remaining useful life prediction under cross-operating conditions

Qiyuan LI,Xin CHENG,Wenqing MA,Kaigan ZHANG,Tangbin XIA,Lifeng XI

表 3 SAFCM与其他先进方法的域适应预测性能比较

Tab.3 Comparison of domain adaptation prediction performance of SAFCM and other state-of-the-art methods

方法 RMSE1/cycles RMSE2/cycles Score1 Score2 FD002 FD003 FD004 FD001 FD003 FD004 FD002 FD003 FD004 FD001 FD003 FD004 DANN[24] 31.21 24.07 27.23 22.71 22.41 24.18 12 415.32 4 005.81 9 016.18 2 289.48 3 700.22 5 349.81 CORAL[25] 39.97 42.01 38.35 35.47 36.40 37.26 37 600.07 46 826.09 28 640.64 9 321.05 13 314.34 19 782.51 CADA[26] 33.43 17.90 29.37 17.42 20.55 24.89 19 996.12 1 694.28 11 581.88 822.35 3 008.26 5 511.68 MCDA[27] 42.19 39.37 34.57 37.96 36.44 37.96 40 147.26 25 641.77 5 382.38 11 059.26 7 686.89 17 670.28 SAFCM 20.39 17.89 21.81 17.73 21.27 22.47 4 386.66 1 445.38 4 506.43 863.15 2 626.98 4 372.02 方法 RMSE3/cycles RMSE4/cycles Score3 Score4 FD001 FD002 FD004 FD001 FD002 FD003 FD001 FD002 FD004 FD001 FD002 FD003 DANN[24] 22.26 28.86 25.48 19.51 22.85 22.12 5 082.15 7 540.77 9 888.57 1 000.34 6 214.72 3 324.13 CORAL[25] 35.63 40.10 38.58 35.73 38.35 35.14 9 446.81 35 069.45 29 449.80 10 294.36 36 487.91 9 525.98 CADA[26] 20.25 28.69 30.32 25.51 23.42 20.34 1 341.10 11 864.96 12 018.76 2 780.70 7 204.24 3 323.91 MCDA[27] 35.18 40.87 40.02 39.33 40.16 37.08 5 334.81 42 708.90 45 589.35 17 057.24 52 331.20 11 804.43 SAFCM 18.06 21.70 22.64 20.10 21.43 19.54 1 086.96 5 133.46 7 608.82 1 350.91 3 622.50 1 745.91