Research Article
Deep Multiscale Soft-Threshold Support Vector Data Description for Enhanced Heavy-Duty Gas Turbine Generator Sets’ Anomaly Detection
Table 4
Comparative performance of DMS-SVDD and other anomaly detection algorithms in ablation experiments.
| | Accurate | Recall | F1 score | Smoothness |
| Deep-SVDD | 0.8249 | 0.8050 | 0.8920 | 0.00617 | Deep-SVDD + S | 0.8584 | 0.8422 | 0.9144 | 0.00402 | Deep-SVDD + M | 0.9930 | 0.9922 | 0.9961 | 0.00515 | OCSVM | 0.6788 | 0.6876 | 0.7936 | 0.00634 | AE | 0.7696 | 0.7567 | 0.8550 | 0.00493 | DMS-SVDD | 0.9985 | 0.9989 | 0.9992 | 0.00300 |
|
|