Research Article
Bearing Fault Diagnosis Based on Frequency Subbands Feature Extraction and Multibranch One-Dimension Convolutional Neural Network
Table 8
Accuracy comparison of different models across loads.
| | Training domain | FFT-SVM (%) | FFT-MLP (%) | FFT-DNN (%) | WDCNN (%) | TICNN (%) | Ensemble TICNN (%) | The proposed model (%) |
| | 1 ⟶ 2 | 68.6 | 82.1 | 82.2 | 99.2 | 99.1 | 99.5 | 99.4 | | 1 ⟶ 3 | 60.0 | 85.6 | 82.6 | 91.0 | 90.7 | 91.1 | 97.1 | | 2 ⟶ 1 | 73.2 | 71.5 | 72.3 | 95.1 | 97.4 | 97.6 | 96.1 | | 2 ⟶ 3 | 67.6 | 82.4 | 77.0 | 91.5 | 98.8 | 99.4 | 98.6 | | 3 ⟶ 1 | 68.4 | 81.8 | 76.9 | 78.1 | 89.2 | 90.2 | 90.2 | | 3 ⟶ 2 | 62.0 | 79.0 | 77.3 | 85.1 | 97.6 | 98.7 | 97.3 | | AVG | 66.6 | 80.4 | 78.1 | 90.0 | 95.5 | 96.1 | 96.4 |
|
|
