Abstract:

Near-infrared spectroscopy (NIRS) quantitative detection model of sesame oil adulterated with rapeseed oil was established by partial least squares (PLS). Mixed samples of sesame and rapeseed oil with different proportions were scanned using a Thermo-Nicolet Protege-460 FT-IR spectrometer and their near infrared diffuse reflectance spectra were collected in 4200–10000 cm-1 region. The reliability of the model established was verified by cross-validation and external test. The results showed that the optimum wavelength range was 4500–8745 cm-1 and the optimum spectra pre-treatment way was multiplicative signal correction (MSC)/first derivative/Norris derivative filter (3, 5). The predictive correlation coefficient of the PLS model was 0.9983, and the root mean square error (RMSEC) was 0.976. Meanwhile, the model was applied to predict the unknown samples, and it was found that in the addition range of 10%– 70% of rapeseed oil, the predicted values were accurate and reliable. Therefore, this PLS model based on NIRS can be used for rapid detection of rapeseed oil adulterated in sesame oil.

Key words: near infrared spectroscopy (NIRS), sesame oil, adulteration, rapeseed oil, partial least squares (PLS)