Uso da espectroscopia NIR e calibração multivariada para prospecção de oleaginosas quanto as suas características de óleo e proteína

Abstract

The potential application of NIR spectroscopy was studied to prospect oil seeds as to its characteristics of oil and protein. Spectra in the region 400-2500 nm were obtained from seeds of peanut, sesame, cotton and castor oil. The reference measurements included: oil content, protein and fatty acids. Using the technique PLS calibration models were developed for genotypes of groundnut, sesame and castor. As for peanuts: oil content (R = 0.4 and RMSECV = 1.7%), water content (R = 0.7 and RMSECV = 0.2%), oleic acid (R = 0.7 and RMSECV = 1.1%), linoleic acid (R = 0.7 and RMSECV = 0.8%), palmitic acid (R = 0.8 and RMSECV = 0.1%) and stearic acid (R = 0.6 and RMSECV = 0.3%); Sesame: oil content (R = 0.7 and RMSECV = 1.7%), water content (R = 0.9 and RMSECV = 0.1%), total protein (R = RMSECV = 0.8 and 0.9%), oleic acid (R = 0.6 and RMSECV = 0.8%), linoleic acid (R = 0.8 and RMSECV = 0.8%), palmitic acid ( R = 0.9 and RMSECV = 0.1%), stearic acid (r = 0.6 and RMSECV = 0.1%) and linolenic acid (R = 0.8 and RMSECV = 0.01%) and; castor : ricinoleic acid (R = 0.90 and RMSECV = 0.57%), oleic acid (R = 0.94 and RMSECV = 0.18%), linoleic acid (R = 0.90 and RMSECV = 0.25% ), palmitic acid (R = 0.92 and RMSECV = 0.04%), stearic acid (R = 0.91 and RMSECV = 0.03%) and linolenic acid (R = 0.51 and = 0.01 RMSECV %). Techniques were applied to cotton selection of samples and variables (APS-MLR). The models PLS and APS-MLR were compared and this obtained better results: oil content (R = 2.1% e RMSEP = 0.7), water content (R = 0.9% and RMSEP = 0.5), linoleic acid (r = 2.0% and RMSEP = 0.1), oleic acid (R = 0.3% and RMSEP = 0.7), palmitic acid (R = 0.9% and RMSEP = 0.3) and stearic acid (R = 0.05 and RMSEP = 0.05%). The levels of oil, water, total protein and fatty acids were predicted in seed and oil by VIS- NIR modeling spectra and calibration by PLS and MLR-APS in oilseeds