Comparison and Optimization of Neural Networks and Network Ensembles for Gap Filling of Wind Energy Data

Table 15

Summary of results for the modeled energy output of turbine WE 1 using 13 input variables from the site Olpe.

Destination variable	Input variables	Network type and topology	RMSE (kWh)	Correlation (𝑟𝑆)
𝑊WE1	𝑊(OL1); average instantaneous power (OL 1); Minimum instantaneous power (OL 1); Average wind direction (OL 1); Maximum wind speed (OL 1); 𝑊(OL2); Daily hours of operation (OL 2); Minimum instantaneous power (OL 2); Average instantaneous power (OL 2); Maximum instantaneous power (OL 2); Average wind speed (OL 2); 𝑊(OL3); Average wind speed (OL 3)	Ensemble; 2 × RBF (15-92-1); (15-101-1)	477	0.97
		RBF (15-95-1)	494	0.97
		RBF (15-93-1)	486	0.97