Short-Term Wind Speed Forecasting Using Decomposition-Based Neural Networks Combining Abnormal Detection Method

Table 1

Forecasting errors of different models.

Sample	HEN1			HEN2			NAR			53H-NAR			EEMD-NAR			Persist			ARIMA
	RMSE	MAE	MAPE	RMSE	MAE	MAPE	RMSE	MAE	MAPE	RMSE	MAE	MAPE	RMSE	MAE	MAPE	RMSE	MAE	MAPE	RMSE	MAE	MEPE
Site 1
Jan.	0.53	0.36	9.49%	0.54	0.36	9.68%	0.79	0.56	14.44%	0.79	0.56	15.94%	0.54	0.37	10.45%	0.79	0.56	15.51%	0.78	0.54	15.60%
Apr.	0.63	0.47	5.03%	0.62	0.47	5.02%	1.10	0.78	8.24%	1.30	0.87	9.09%	0.62	0.47	5.66%	1.17	0.80	8.50%	1.14	0.78	8.35%
Jul.	0.95	0.70	15.40%	0.83	0.61	14.46%	1.44	1.04	22.04%	1.52	1.05	20.49%	0.87	0.65	15.20%	1.45	0.96	19.96%	1.49	1.00	20.53%
Oct.	0.56	0.41	5.18%	0.57	0.42	5.22%	1.21	0.82	11.40%	1.19	0.81	10.91%	0.59	0.43	5.87%	1.00	0.70	9.42%	1.02	0.72	9.43%
Site 2
Jan.	0.57	0.46	20.35%	0.55	0.37	17.20%	0.71	0.52	24.65%	0.77	0.55	24.52%	0.55	0.38	17.99%	0.78	0.57	26.22%	0.77	0.57	25.36%
Apr.	1.50	1.11	16.00%	1.38	1.01	15.82%	1.68	1.20	19.43%	1.68	1.16	18.24%	1.60	1.11	16.79%	1.68	1.19	18.61%	1.70	1.22	19.00%
Jul.	0.73	0.55	5.96%	0.60	0.41	3.86%	0.88	0.62	5.95%	1.05	0.70	6.66%	0.61	0.43	4.49%	0.93	0.65	6.18%	0.93	0.65	6.21%
Oct.	0.74	0.56	5.42%	0.59	0.42	3.63%	2.14	0.86	7.05%	0.93	0.66	5.51%	0.58	0.43	4.20%	0.96	0.66	5.66%	0.94	0.67	5.70%
Site 3
Jan.	0.45	0.31	9.07%	0.35	0.26	7.48%	0.58	0.43	12.58%	0.57	0.41	12.38%	0.36	0.27	8.41%	0.59	0.43	12.44%	0.56	0.41	11.69%
Apr.	1.38	0.90	10.56%	0.99	0.75	9.81%	1.73	1.22	15.03%	1.68	1.14	14.21%	1.01	0.77	10.81%	1.84	1.29	15.57%	1.88	1.32	15.90%
Jul.	0.66	0.44	6.56%	0.54	0.38	5.57%	0.75	0.56	7.78%	0.71	0.51	7.16%	0.55	0.40	6.08%	0.78	0.54	7.80%	0.78	0.56	8.02%
Oct.	0.74	0.51	7.02%	0.68	0.48	6.71%	1.04	0.74	10.14%	0.89	0.67	9.46%	0.75	0.52	7.72%	0.94	0.69	9.58%	0.92	0.68	9.45%
Mean	0.78	0.56	9.67%	0.69	0.50	8.71%	1.17	0.78	13.23%	1.09	0.76	12.88%	0.72	0.52	9.47%	1.08	0.75	12.95%	1.08	0.76	12.94%