Research Article
Multidomain Feature Fusion for Varying Speed Bearing Diagnosis Using Broad Learning System
Table 5
Classification comparison of different combinations.
| Models | Training dataset (km/h) | Testing dataset (km/h) | Average accuracy (%) |
| BLS | 30 | 50 | 94.12 | 30 | 100 | 79.24 | 50 | 30 | 98.03 | 50 | 100 | 85.63 | 100 | 30 | 74.94 | 100 | 50 | 91.32 |
| ANN | 30 | 50 | 87.33 | 30 | 100 | 72.44 | 50 | 30 | 80.02 | 50 | 100 | 71.16 | 100 | 30 | 79.46 | 100 | 50 | 90.68 |
| ELM | 30 | 50 | 82.64 | 30 | 100 | 68.65 | 50 | 30 | 95.25 | 50 | 100 | 81.46 | 100 | 30 | 78.12 | 100 | 50 | 70.18 |
| SVM | 30 | 50 | 78.91 | 30 | 100 | 80.14 | 50 | 30 | 94.33 | 50 | 100 | 94.42 | 100 | 30 | 80.95 | 100 | 50 | 91.70 |
| LR | 30 | 50 | 83.05 | 30 | 100 | 64.28 | 50 | 30 | 95.51 | 50 | 100 | 87.98 | 100 | 30 | 63.35 | 100 | 50 | 83.37 |
|
|