An Experimental Investigation of FNN Model for Wind Speed Forecasting Using EEMD and CS

Table 2

The more the step of forecast, the lower the accuracy.

	CS-FNN model
	Errors	WTG25	WTG26	WTG27	WTG28	Mean
Jan.	One-step ahead forecasting errors
	MAE (m/s)	0.60	0.63	0.79	0.62	0.66
	MAPE (%)	5.46	6.97	7.34	6.12	6.47
	RMSE (m/s)	0.80	0.84	1.22	0.83	0.92
	Two-step ahead forecasting errors
	MAE (m/s)	0.73	0.79	1.30	0.75	0.89
	MAPE (%)	6.80	9.15	11.23	7.40	8.65
	RMSE (m/s)	0.98	1.05	2.97	0.98	1.50
	Three-step ahead forecasting errors
	MAE (m/s)	0.79	0.75	1.07	0.80	0.85
	MAPE (%)	7.41	8.83	10.54	7.96	8.68
	RMSE (m/s)	1.07	0.99	1.51	1.07	1.16
	Four-step ahead forecasting errors
	MAE (m/s)	0.86	0.87	1.15	0.85	0.93
	MAPE (%)	8.19	11.05	11.53	8.66	9.86
	RMSE (m/s)	1.13	1.13	1.85	1.15	1.32
	Five-step ahead forecasting errors
	MAE (m/s)	0.92	0.89	1.22	0.93	0.99
	MAPE (%)	8.79	11.29	12.53	9.49	10.52
	RMSE (m/s)	1.21	1.18	1.73	1.24	1.34
	Six-step ahead forecasting errors
	MAE (m/s)	1.01	0.99	1.41	0.97	1.09
	MAPE (%)	9.56	12.20	14.01	10.12	11.47
	RMSE (m/s)	1.32	1.29	2.62	1.31	1.64