A Novel Hybrid Model for Short-Term Wind Speed Forecasting Based on Twice Decomposition, PSR, and IMVO-ELM

Table 3

The average evaluation metrics of forecasting by different methods with 20 times’ independent calculations.

Datasets	Methods	Evaluation metrics of forecasting
		MAPE (%)	MAE (m/s)	RMSE (m/s)
A	IMVO-ELM	25.919	0.508	0.647
	EMD--IMVO-ELM	13.326	0.320	0.399
	CEEMDAN—IMVO-ELM	10.586	0.237	0.308
	ICEEMDAN--IMVO-ELM	9.314	0.207	0.281
	EMD-cc-PSR-IMVO-ELM	11.849	0.307	0.391
	CEEMDAN-cc-PSR-IMVO-ELM	9.765	0.226	0.297
	ICEEMDAN-cc-PSR-IMVO-ELM	8.081	0.184	0.252
	The proposed method	6.682	0.162	0.225
B	IMVO-ELM	12.069	1.361	1.685
	EMD--IMVO-ELM	8.345	0.976	1.182
	CEEMDAN—IMVO-ELM	7.420	0.868	1.066
	ICEEMDAN--IMVO-ELM	6.821	0.785	0.944
	EMD-cc-PSR-IMVO-ELM	7.415	0.874	1.104
	CEEMDAN-cc-PSR-IMVO-ELM	6.446	0.746	0.891
	ICEEMDAN-cc-PSR-IMVO-ELM	5.830	0.673	0.812
	The proposed method	5.392	0.625	0.767
C	IMVO-ELM	9.033	0.814	1.069
	EMD--IMVO-ELM	5.227	0.465	0.585
	CEEMDAN—IMVO-ELM	5.521	0.493	0.641
	ICEEMDAN--IMVO-ELM	5.013	0.444	0.589
	EMD-cc-PSR-IMVO-ELM	5.071	0.448	0.572
	CEEMDAN-cc-PSR-IMVO-ELM	4.419	0.397	0.522
	ICEEMDAN-cc-PSR-IMVO-ELM	4.013	0.361	0.477
	The proposed method	3.446	0.308	0.419
D	IMVO-ELM	12.717	1.173	1.574
	EMD--IMVO-ELM	8.592	0.799	1.021
	CEEMDAN—IMVO-ELM	8.536	0.775	0.944
	ICEEMDAN--IMVO-ELM	7.800	0.711	0.875
	EMD-cc-PSR-IMVO-ELM	8.334	0.778	1.003
	CEEMDAN-cc-PSR-IMVO-ELM	6.946	0.646	0.808
	ICEEMDAN-cc-PSR-IMVO-ELM	5.832	0.541	0.699
	The proposed method	5.435	0.509	0.668