Identification Of Edible Oil And Detection Of Its Adulteration Via Raman Spectroscopy

This paper made a qualitative identification of ordinary vegetable oil and inferior(edible)oil based on Raman spectroscopy. Raman spectra were acquired through the portable Raman spectrometer,then it corrected the baseline of the original Raman spectra through polynomial linearity curve fitting. Secondly, a wavenumber segment 700 cm-1~1800 cm-1with relative rich spectral information was chosen for polynomial smoothing filtering and normalization of the spectra. Finally,rapid identification and detection of adulteration of oil content of inferior(edible)oil is studied via portable optical raman scattering spectrum technology combined with chemometrics method(pls-da; fisher discriminant analysis; Random forest algorithm),ratio method andquantitative analysis method(i PLS/siPLS). Research results demonstrated that:Qualitative analysis results show that both ratio method and PLS-DA could identify inferior(edible) oil samples accurately. The qualified oil and inferior(edible) oil fatty acid relative degree of saturation to be characterize by y=I1654cm-1/1438 cm-1in the ratio method which found that there is a significant difference of the relative degree between qualified oil and inferior(edible) oil. A model for the identification of inferior(edible)oil in olive oil was established via Raman spectroscopy combined with fisher discriminant analysis and random forest algorithm as well as the characteristic band ratio method.Compare the effect of three methods can be seen that the characteristic band ratio method is the best, the random forest algorithm is the second, and the Fisher discriminant analysis is the most effective.Quantitative analysis results show that the model of quantitative analysis of olive oil mixed with inferior(edible) oil was established which found that the effect after SNV pretreatment of the siPLS model is the best, the optimal wavelength range is:1159 cm-1~1190cm-1、 1264cm-1~1295 cm-1、 1433 cm-1~1464 cm-1、 1627 cm-1~1658 cm-1,RMSEP=0.0503,r=0.9613 of training set; RMSEP=0.0485,r=0.9822 of prediction set.Moreover, the LLA(lower limit of application LLA) of IEO(inferior edibleoilsIEO)percentage in the mixed oil detected in the experiment is c[IEO]=0.5%. The theoretical LOD(limit of detection Lo D) calculated from table4-7 is about c[IEO]=0.475%.