A portable LED-induced fluorescence system for quantitative detection of different kinds of vegetable oil adulteration

This study developed a portable LED-induced fluorescence detection system for quantitative detection of various vegetable oil adulteration. Eight common vegetable oils and 14 different adulterated samples with adulteration concentrations ranging from 0 to 50% were prepared. Before quantitative analysis, different classification models were established for the determination of types of adulteration oil. The overall recognition accuracy was greater than 98%. Furthermore, with a simple calculation, the proposed normalized spectral ratio (NSR) preprocessing method was used to eliminate the light scattering effects in the raw fluorescence spectra. In addition, the competitive adaptive reweighted sampling (CARS) method was used to select characteristic wavelengths. The final oil adulteration quantitative analysis model was NSR_CARS+PLS. The range of correlation coefficient ( R p) and root-mean-square-error ( RMSEP ) for the prediction datasets were 0.9548-0.9974 and 1.0265%-5.0236%, respectively. • A LED-induced fluorescence system was developed to detect oil adulteration. • NSR method was proposed for elimination of light scattering effects. • Fourteen adulteration detection models including 8 kinds of oil were developed. • Root-mean-square-error for the prediction datasets were 1.0265%-5.0236%.