Abstract: The discrimination models of tea water adulterated with carmine pigment were constructed combined with the technology of near-infrared（NIR）spectroscopy and partial least squares discriminant analysis（PLS-DA）. Firstly, 38 unadulterated tea water samples, and 38 tea water samples adulterated with carmine pigment（0.1~10.0 μg/mL）were prepared respectively. The NIR spectra of all samples were collected and pretreated with de-noising method in the range of 4 000~15 000 cm^-1. Based on the de-noising pretreatment of spectral data, 7 different wave-number ranges were selected to construct PLS-DA models for discriminating adulterated tea water samples. It is pointed out that the wave-number interval selection has great influence on predictive ability of PLS-DA model. The results show that good discriminant model can be obtained in the range of 4 000~11 000 cm^-1. The classification accuracies of the constructed PLS-DA model were 100% and 96.15% for calibration set and prediction set respectively.

Key words: near-infrared spectroscopy, black tea, carmine, wave-number selection in model, partial least squares discriminant analysis