A Rapid Screening Approach for Authentication of Olive Oil and Classification of Binary Blends of Olive Oils Using Low-Field Nuclear Magnetic Resonance Spectra and Support Vector Machine

Abstract

Due to the quality differentiation and commercial concerns, rapid authentication and addressing the adulterants in olive oil is of great importance. The feasibility of identifying pure olive oil as well as classifying the binary blends of olive oils according to the adulterants in olive oils using low-field NMR spectroscopy and support vector machine (SVM) have been investigated. Based on the characterization of low-field NMR profiles of six types of vegetable oil and the binary blends of olive oils with three types of seeds oils (corn, soybean, and sunflower seed oils), SVM was employed to build the authentication and classification models. The result indicated that the difference of oils and the type of blends can be monitored by low-field NMR profiles. SVM classification models for identifying pure olive oils from blended ones were developed and an 84.92% classification accuracy was acquired when the adulteration ratio is above 10%. For the classification of binary blends of olive oils according to the seed oils, two SVM classification strategies have been developed and compared, and the SVM model with a suspected range of 10%–30% could provide an acceptable classification result. LF-NMR could be a novel screening method for the authentication of olive oil.