Robust Fault Detection of Wind Energy Conversion Systems Based on Dynamic Neural Networks

Table 3

The training and testing results for the network structure selection to model 𝛽₁.

Network structure	𝐾	Training			Testing
		𝐽	𝑓FPE	𝑓AIC	𝐽	𝑓FPE	𝑓AIC
N^22,3,1(1)	25	0.2447	0.2583	−1.3536	0.2510	0.2574	−1.3573
N^22,4,1(1)	33	0.1023	0.1099	−2.2083	0.1471	0.1520	−1.8836
N^22,5,1(1)	41	0.0394	0.0431	−3.1451	0.0821	0.0855	−2.4588
N^22,6,1(1)	49	0.0422	0.0465	−3.0673	0.0964	0.1012	−2.2902
N^22,7,1(1)	57	0.0509	0.0561	−2.8799	0.1150	0.1208	−2.1138
N^22,3,1(1)	31	0.0304	0.0325	−3.4261	0.0793	0.0818	−2.5035
N^22,4,1(1)	41	0.0781	0.0854	−2.4609	0.1078	0.1123	−2.1865
N^22,5,1(1)	51	0.0471	0.0526	−2.9450	0.1037	0.1091	−2.2153
N^22,6,1(1)	61	0.0348	0.0384	−3.2601	0.0873	0.0917	−2.3894
N^22,7,1(1)	71	0.0457	0.0527	−2.9437	0.0782	0.0936	−2.3686
N^32,3,2,1(2-2)	50	0.0641	0.0714	−2.6390	0.0952	0.1001	−2.3018
N^32,4,2,1(2-2)	61	0.0373	0.0426	−3.1566	0.0849	0.0902	−2.4053
N^32,4,3,1(2-2)	73	0.0282	0.0325	−3.4103	0.1012	0.1089	−2.2177