Research Article
Fault Diagnosis of Bearings with Adjusted Vibration Spectrum Images
Table 8
The classification accuracy for multiclass fault severity detection.
| Dataset | n | Testing accuracy (%) | Load 0 | Load 1 | Load 2 | Load 3 |
| G1 | 1 | 98.54 | 93.97 | 91.17 | 89.48 | 3 | 99.87 | 98.04 | 96.58 | 94.33 | 5 | 100 | 98.79 | 97.81 | 95.56 | 10 | 100 | 99.24 | 99.48 | 97.98 |
| G2 | 1 | 98.44 | 100 | 99.92 | 99.92 | 3 | 99.61 | 100 | 99.97 | 100 | 5 | 99.63 | 100 | 100 | 100 | 10 | 99.98 | 100 | 100 | 100 |
| G3 | 1 | 80.48 | 99.69 | 99.96 | 100 | 3 | 86.11 | 99.78 | 100 | 100 | 5 | 88.97 | 99.89 | 100 | 100 | 10 | 92.72 | 99.95 | 100 | 100 |
| G4 | 1 | 84.21 | 98.14 | 99.76 | 100 | 3 | 83.24 | 99.00 | 100 | 100 | 5 | 84.19 | 99.45 | 100 | 100 | 10 | 84.08 | 99.70 | 100 | 100 |
| G5 | 1 | 99.56 | 100 | 99.98 | 100 | 3 | 100 | 100 | 100 | 100 | 5 | 100 | 100 | 100 | 100 | 10 | 100 | 100 | 100 | 100 |
|
|
n is the training samples per class under each of the specified load conditions.
|