Research Article
KPCA and AE Based Local-Global Feature Extraction Method for Vibration Signals of Rotating Machinery
Table 2
Parameter configuration with mean classification accuracy higher than 95%.
| Ns | Number of spectrum line | Vs | 1 | 2 | 3 | Dimension of global feature | Mean accuracy (%) | Dimension of global feature | Mean accuracy (%) | Dimension of global feature | Mean accuracy (%) |
| 2 | 512 | 2 | — | 4 | — | 6 | 95.56 | 4 | 256 | 4 | — | 8 | 96.67 | 12 | 96.67 | 8 | 128 | 8 | 97.78 | 16 | 96.67 | 24 | 96.67 | 16 | 64 | 16 | 97.78 | 32 | 97.78 | 48 | 95.56 | 32 | 32 | 32 | 96.67 | 64 | 98.89 | 96 | — | 64 | 16 | 64 | 96.67 | 128 | 96.67 | 192 | — | 128 | 8 | 128 | 96.67 | 256 | — | 384 | — |
|
|