Table 5 Performance of PV power forecasting models on testing set (Dataset #2).

From: A novel hybrid model integrating CEEMDAN decomposition, dispersion entropy and LSTM for photovoltaic power forecasting and anomaly detection

Model

RMSE

MAE

SSE

nRMSE

nMAE

R²

GRU

2.287

2.122

30,121

9.47%

8.78%

0.837

LSTM

1.839

1.286

19,481

7.61%

5.32%

0.894

Transformer

1.473

1.166

12,501

6.10%

4.83%

0.932

DLinear

1.606

1.201

14,852

4.97%

4.97%

0.919

CEEMDAN-LSTM

1.374

1.048

10,866

5.68%

4.34%

0.941

CEEMDAN-DispEn-LSTM

1.120

0.781

7235.6

4.63%

3.23%

0.961

  1. The bold values are the optimal values among all methods.
Back to article page