Rapid Identification of Vegetable Oil Species Using Low-Field Nuclear Magnetic Resonance

Abstract

The relaxation signals of rapeseed oil, soybean oil, peanut oil, sunflower oil and corn oil were investigated using low-field nuclear magnetic resonance (LF-NMR), and the correlation between the composition of vegetable oils and their NMR relaxation characteristics was analyzed. Furthermore, a classification model for vegetable oils was established based on the echo attenuation information of LF-NMR using principal component analysis-linear discriminant analysis (PCA-LDA), and the effects of discriminant functions and the number of principal components (PC) on the model’s performance were studied. The experimental results showed that the decreasing order of the attenuation rates of the echo curves was peanut oil > rapeseed oil > corn oil > soybean oil > sunflower oil. The types of vegetable oils had significant effects on relaxation properties including T2W, T23, S23 and Stotal (P < 0.05). There existed extremely significant correlations between T2W, T22, T23, S23 and Stotal and the contents of C18:1, C18:2, C20:0, monounsaturated fatty acid (MUFA) and polyunsaturated fatty acid (PUFA) (P < 0.01). The classification precision of the PCA-LDA model developed using linear discriminant function and 10 PCs for the training and prediction sets were 100.0% and 88.2%, respectively. It can be seen that it is feasible to identify vegetable oil species using LF-NMR. This study can provide a theoretical basis and technical support for quality and safety detection of different edible vegetable oils.

Keywords

• low-field nuclear magnetic resonance; vegetable oil; species identification; principal component analysis; linear discriminant analysis