Open Access
Review
Table 6
Performance comparison of feature extraction techniques in nuclear pump fault diagnosis evolution.
| Technology type | Feature extraction method | Adaptability to dynamic operating conditions (%) | Computational complexity (FLOPs) | Diagnostic accuracy (%) |
|---|---|---|---|---|
| Traditional signal processing | Manual design (power spectrum) | Poor (Δ-24.7) | Low (1.1 × 106) | 89.4 [5] |
| Adaptive algorithms | Algorithm optimization (envelope demodulation) | Moderate (Δ-9.2) | Moderate (3.5 × 106) | 93.6 [9] |
| Multi-source Fusion | Multimodal feature fusion | Good (Δ-6.3) | Moderate-High (5.8 × 106) | 96.8 [15] |
| Deep learning | Automatic learning (attention mechanism) | Excellent (Δ-2.1) | High (12.8 × 106) | 99.5 [11] |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.