Data-Driven Photovoltaic System Modeling Based on Nonlinear System Identification

Table 2

Best fit with different input and output nonlinearity.

Hammerstein-Wiener model
Input nonlinearity	Output nonlinearity	Final predicted error (FPE)	Loss function	Best fit (%)
Piecewise linear	Piecewise linear	6.723	6.723	80.25
Piecewise linear	Sigmoid network	5.036	5.036	86.11
Piecewise linear	Saturation	1.751	1.751	92.83
Piecewise linear	Dead zone	1.023	1.023	92.24
Piecewise linear	Wavelet network	2.083	2.082	88.11
Sigmoid network	Piecewise linear	1.265	1.265	77.94
Sigmoid network	Sigmoid network	3.547	3.547	81.05
Sigmoid network	Saturation	2.016	2.016	89.45
Sigmoid network	Dead zone	4.952	4.951	85.37
Sigmoid network	Wavelet network	0.1694	0.1693	63.09
Saturation	Piecewise linear	1.262	1.262	92.22
Saturation	Sigmoid network	1.199	1.199	92.44
Saturation	Saturation	FE	FE	FE
Saturation	Dead zone	FE	FE	FE
Saturation	Wavelet network	0.7606	0.7601	85.83
Dead zone	Piecewise linear	0.3974	0.3974	93.4
Dead zone	Sigmoid network	1.464	1.464	93.98
Dead zone	Saturation	FE	FE	FE
Dead zone	Dead zone	FE	FE	FE
Dead zone	Wavelet network	0.2972	0.2972	90.86
Wavelet network	Piecewise linear	1.717	1.716	91.42
Wavelet network	Sigmoid network	2.038	2.037	89.61
Wavelet network	Saturation	131.5	131.5	−0.7279
Wavelet network	Dead zone	2.054	2.054	90.09
Wavelet network	Wavelet network	0.3643	0.364	93.81

FE: failed estimation.