Table 4 Comparison of average F1 scores among different methods.
From: Non-intrusive load monitoring based on time-enhanced multidimensional feature visualization
Ref. | Characteristic | Training model | setResult (%) | |||
|---|---|---|---|---|---|---|
PLAID | WHITED | REDD | HRAD | |||
V-I + harmonic | SE-ResNet | 96 | – | 96 | – | |
V-I | CGAN | 95.3 | 98.31 | – | – | |
V-I | VGG16 + SVM | 98.2 | – | – | – | |
Color-encoded V-I | BOA + CNN | 87 | – | – | – | |
V-I | CNN + Siamese + BP | 99.2 | – | – | – | |
WVI + MTF + I-GAF | EN-SE-RECNN | 97.44 | 95.43 | – | 96.84 | |
Time-frequency features | DOSL | 95.78 | – | – | – | |
Proposed | Three-dimensional space-time color V-I | ECA-ResNet | 97.3 | 98.8 | – | – |
