| Aiming at the quality problems of seabuckthorn seed oil on market, the method was developed for rapid discerning of seabuckthorn seed oil by near-infrared spectroscopy. The transflectance spectra between 4 000 cm-1 and 6 000 cm-1 of two hundreds and thirty-four samples were collected including pure seabuckthorn seed oils, other vegetable oils and adulterated seabuckthorn seed oils. The chemometrics methods of soft independent modeling of class analogy(SIMCA), partial least squares discriminant analysis(PLS-DA) and support vector machine(SVM) were adopted to estabish discrimination models of the mentioned three types of oils respectively. And one hundred and seventeen independent samples were prepared for model validation. Satisfactory results were obtained by the three chemometrics methods. The discrimination accuracy of SVM was the best, reaching 100% for calibration and independent validation. The applications of near-infrared spectroscopy technology in discerning pure seabuckthorn seed oil and adulteration type were practical. It was feasible to discern seabuckthorn seed oil rapidly by near-infrared spectroscopy. |
