Research Article
Multisignal VGG19 Network with Transposed Convolution for Rotating Machinery Fault Diagnosis Based on Deep Transfer Learning
Table 3
Classification results for the gearbox dataset.
| | | Fault diagnosis method | Time window | Channel | Bearing (20-0) | Bearing (30-2) | Gearset (20-0) (%) | Gearset (30-2) (%) |
| | [35] | Premodel (VGG16) | 1024 | 1 | 98.90% | 98.84% | 98.70 | 98.07 |
| | [36] | SAE-DNN | 20000 | 1 | 87.50% | 92.10% | 92.70 | 91.90 | | GRU | 20000 | 1 | 91.20% | 92.40% | 93.80 | 90.70 | | BiGRU | 20000 | 1 | 93.00% | 93.60% | 93.80 | 90.70 | | LFGRU | 20000 | 1 | 93.20% | 94.00% | 94.80 | 95.80 |
| | [38] | DSR | 1024 | 1 | | | 99.59 | 99.78 | | SACNN | 1024 | 1 | | | 99.86 | 99.88 |
| | | CNN trained from scratch | 512 | 1 | 98.08% | 98.08% | 97.80 | 97.80 | | DNN | 512 | 1 | 97.56% | 97.56% | 93.68 | 93.68 | | Premodel (VGG19) | 512 | 8 | 99.78% | 99.78% | 99.99 | 99.99 |
|
|
