Research Article
A Damage Classification Approach for Structural Health Monitoring Using Machine Learning
Table 3
Behavior of machines with five scores per sensor (specimen 1, four sensors).
| Machine type | UND | DMG1 | DMG2 | DMG3 | DMG4 | DMG5 | DMG6 |
| Complex Tree | 90% | 99% | 18% | 99% | 99% | 97% | 100% |
| Medium Tree | 90% | 99% | 18% | 99% | 99% | 97% | 100% |
| Simple Tree | 90% | 99% | 0% | 100% | 0% | 97% | 100% |
| Linear SVM | 97% | 100% | 100% | 99% | 99% | 99% | 100% |
| Quadratic SVM | 97% | 100% | 100% | 99% | 99% | 99% | 100% |
| Cubic SVM | 97% | 100% | 100% | 99% | 99% | 99% | 100% |
| Fine Gaussian SVM | 100% | 9% | 8% | 28% | 8% | 30% | 56% |
| Medium Gaussian SVM | 99% | 100% | 98% | 99% | 99% | 98% | 100% |
| Coarse Gaussian SVM | 98% | 100% | 100% | 100% | 99% | 100% | 100% |
| Fine KNN | 97% | 100% | 100% | 100% | 99% | 100% | 100% |
| Medium KNN | 97% | 100% | 100% | 100% | 99% | 100% | 100% |
| Coarse KNN | 93% | 100% | 100% | 99% | 97% | 100% | 100% |
| Cosine KNN | 96% | 100% | 100% | 100% | 99% | 100% | 100% |
| Cubic KNN | 95% | 100% | 100% | 100% | 99% | 99% | 100% |
| Weighted KNN | 97% | 100% | 100% | 100% | 99% | 100% | 100% |
| Boosted Trees | 90% | 100% | 0% | 100% | 0% | 100% | 100% |
| Bagged Trees | 99% | 100% | 100% | 100% | 100% | 100% | 100% |
| Subspace Discriminant | 98% | 100% | 100% | 100% | 99% | 100% | 100% |
| Subspace KNN | 98% | 100% | 100% | 100% | 99% | 100% | 100% |
| Rusboosted Trees | 90% | 100% | 0% | 0% | 0% | 0% | 0% |
|
|