A novel technique for forecasting the optimal production of a wind generator installed at any geographical point located within a very large area

Journal of Electrical Systems and Information Technology

Table 6 Performance comparison

Refs.	Forecast type	Method	Forecast horizon	Obtained accuracy
Refs.	Forecast type	Method	Forecast horizon	RMSE	MAPE (%)	R²
Yatiyana et al. [58]	Wind speed	ARIMA	6 h	–	4.90	–
Yatiyana et al. [58]	Wind direction			–	15.60	–
Çevik et al. [59]	Wind speed	SWD-ANFIS, ANN and Ensemble methods	3 h	0.3842 m/s	–	–
Liu et al. [60]	Wind speed	ICEEMDAN-MMODA	3 h	0.35 m/s	5.02
Liu et al. [56]	Wind speed	Jaya-SVM	3 h	1.01 m/s	15.34	0.85
Duan et al. [57]	Wind speed	ICEEMDAN-GRU-ARIMA	15 min	0.09 m/s	2.34
Neshat et al. [61]	Wind speed	ED-HGNDO-BiLSTM	1 h	1.57 m/s		0.91
Yun Wang et al. [36]	Wind speed	CNN	1 h	0.73	9.84	0.96
Yun Wang et al. [36]	Wind speed	LSTM	1 h	0.77	10.51	0.95
Sheng-Xiang and Lin [62]	Wind speed	HTD- MOBBSA- Seq2Seq	1 h	1.58 m/s
Our proposed model	Wind speed	LSTM	12 h	0.34 m/s	8.22	0.93
Our proposed model	Wind direction	LSTM		7.8 rad	6.09	0.70