Abstract：In the present paper, the use of near infrared spectroscopy (NIR) as a rapid and cost-effective classification and quantification techniques for the authentication of virgin olive oil were preliminarily investigated. NIR spectra in the range of 12 000-3 700 cm-1 were recorded for pure virgin olive oil and virgin olive oil samples adulterated with varying concentrations of sesame oil, soybean oil and sunflower oil (5%-50% adulterations in the weight of virgin olive oil). The spectral range from 12 000 to 5 390 cm-1 was adopted to set up an analysis model. In order to handle these data efficiently，after pretreatment, firstly，principal component analysis (PCA) was used to compress thousands of spectral data into several variables and to describe the body of the spectra，and the analysis suggested that the cumulate reliabilities of the first six components was more than 99.999%. Then ANN-BP was chosen as further research method. The six components were secondly applied as ANN-BP inputs. The experiment took a total of 100 samples as original model examples and left 52 samples as unknown samples to predict. Finally, the results showed that the 52 test samples were discriminated accurately. And the calibration models of quantitative analysis were built using partial-least-square (PLS). The R values for PLS model are 98.77, 99.37 and 99.44 for sesame oil, soybean oil and sunflower oil respectively, the root mean standard errors of cross validation (RMSECV) are 1.3, 1.1 and 1.04 respectively. Overall, the near infrared spectroscopic method in the present paper played a good role in the discrimination and quantification, and offered a new approach to the rapid discrimination of pure and adulterated virgin olive oil.
Key words：Near infrared spectroscopy(NIR);Virgin olive oil;Discrimination and quantification;BP artificial neural network (BP-ANN);Partial least square(PLS)
