A Comparison of Hour-Ahead Solar Irradiance Forecasting Models Based on LSTM Network

Table 10

Root mean square error (RMSE, W/m²) and normalized root mean square error (nRMSE, %) of different models in sunny, cloudy, and rainy weather.

Model	Sunny (June 27, 2016)		Cloudy (June 8, 2016)		Rainy (May 17, 2016)
	RMSE	nRMSE	RMSE	nRMSE	RMSE	nRMSE
BPNN	38.0148	7.1308	186.9453	47.3051	42.6140	74.164
RNN	49.4941	9.2841	188.9789	47.8197	37.2885	64.8956
Random forest	29.4791	5.5297	147.7844	37.3957	43.9405	76.4725
SVM	66.8935	12.5478	171.4332	43.3799	61.1678	106.4545
General LSTM	21.8439	4.0975	183.1270	46.3389	29.0062	50.4815
The proposed LSTM-MLP	25.2291	4.7325	141.8265	35.8881	26.6292	46.3446