Prediction of fatty methyl esters and physical properties of soybean oil/biodiesel blends from near and mid-infrared spectra using the data fusion strategy

The data fusion strategy was compared to the conventional sensor methodology to evaluate the refractive index, the relative density, and the total fatty acid methyl esters in soybean oil/biodiesel blend samples using near infrared and mid-infrared spectral data coupled with linear and non-linear regression methods. The refractive index and the relative density values were obtained using a refractometer and densimeter system. The fatty acid methyl ester values were acquired by gas chromatography using the standard method (EN-14103:2011). The data fusion approach was applied successfully and showed the lowest root standard error of prediction (RSEP), at least for the refractive index and fatty acid methyl esters in relation to other strategies.