Research Article
Remaining Useful Life Prediction of Milling Tool Based on Pyramid CNN
Table 6
Performance estimation result of the testing dataset for eight state-of-the-art models.
| Methods | Scores | RMSE | C1 | C2 | C3 | C4 | C1 | C2 | C3 | C4 |
| DCNN [35] | 5.572 ± 10.297 | 4.835 ± 10.112 | 5.214 ± 10.822 | 5.713 ± 10.640 | 44.694 ± 2.558 | 45.206 ± 2.353 | 44.967 ± 2.301 | 44.515 ± 2.140 | SVR [34] | 16.997 ± 9.125 | 16.935 ± 9.002 | 17.279 ± 8.910 | 17.529 ± 8.812 | 39.012 ± 1.213 | 39.245 ± 1.221 | 38.914 ± 1.208 | 38.839 ± 1.194 | DBN [38] | 38.114 ± 5.198 | 38.202 ± 5.109 | 38.883 ± 4.950 | 39.163 ± 4.792 | 30.891 ± 1.315 | 31.048 ± 1.382 | 31.008 ± 1.213 | 30.427 ± 1.290 | RDN [36] | 31.341 ± 9.914 | 31.840 ± 9.757 | 32.019 ± 9.857 | 32.498 ± 9.309 | 29.890 ± 1.811 | 29.784 ± 1.887 | 29.560 ± 1.891 | 29.216 ± 1.921 | MSCNN [37] | 40.152 ± 7.486 | 40.274 ± 7.397 | 39.914 ± 8.152 | 40.889 ± 7.109 | 29.407 ± 1.712 | 28.903 ± 1.664 | 29.729 ± 1.677 | 28.597 ± 1.402 | CLSTM [24] | 45.331 ± 4.980 | 45.519 ± 4.832 | 45.712 ± 4.962 | 45.870 ± 4.954 | 28.245 ± 1.290 | 28.552 ± 1.285 | 28.771 ± 1.297 | 28.245 ± 1.314 | MSCAN [39] | 32.497 ± 2.244 | 32.575 ± 2.225 | 32.573 ± 2.294 | 32.814 ± 2.106 | 28.612 ± 1.104 | 28.504 ± 1.123 | 28.482 ± 1.109 | 28.374 ± 1.011 | RF [34] | 40.126 ± 2.268 | 40.125 ± 2.740 | 40.219 ± 2.442 | 40.370 ± 2.589 | 21.441 ± 1.071 | 21.372 ± 1.096 | 21.457 ± 1.101 | 21.371 ± 1.002 | Proposed | 53.962 ± 1.813 | 52.217 ± 1.872 | 52.771 ± 1.617 | 54.248 ± 1.712 | 19.374 ± 0.923 | 19.139 ± 0.853 | 19.424 ± 0.721 | 19.051 ± 0.804 |
|
|
The bold values express that the proposed method has the best performance.
|