Energy Informatics

Table 3 RMSE of best-performing models from each of the 10 training runs

From: Estimating time-delayed variables using transformer-based soft sensors

	PV-Data		Inv-Data
	Average	Best	Average	Best
MLP	0.0250 ± 0.0013	0.0234	0.0201 ± 0.0014	0.0186
LSTM	0.0240 ± 0.0016	0.0228	0.0173 ± 0.0013	0.0151
Transformer	0.0260 ± 0.0013	0.0241	0.0215 ± 0.0025	0.0191
(pretrained) Transformer	0.0287 ± 0.0010	0.0273	0.0246 ± 0.0029	0.0208

Bold values indicate the best result in comparison to the different model types

Back to article page