Research Article
Improvement of Adaptive GAs and Back Propagation ANNs Performance in Condition Diagnosis of Multiple Bearing System Using Grey Relational Analysis
Table 6
Comparison of performance between the AGAs-BPNNs and GRA-AGAs-BPNNs.
| Number of dominant features selected | BPNNs topology | Training | Validation | Testing | CPU time |
| 10 features | 30-30-30-30-16 | 99.3% | 91.7% | 92.4% | 2905.0 | 1 feature | 3-3-3-3-16 | 41.5% | 38.8% | 36.2% | 992.2 | 3 features | 9-9-9-9-16 | 83.6% | 80.4% | 80.4% | 1116.8 | 5 features | 15-15-15-15-16 | 99.4% | 97.1% | 96.7% | 1279.3 | 7 features | 21-21-21-21-16 | 100.0% | 94.6% | 92.5% | 1468.2 | 1 feature (the 10th feature) | 3-3-3-3-16 | 28.7% | 26.7% | 26.3% | 982.6 | 2 features (the 1st and the 10th features) | 6-6-6-6-16 | 59.9% | 52.9% | 55.4% | 1074.8 |
|
|