Study on discrimination of brands of chinese distilled spirit using near infrared transmission and reflectance spectra

A new method for the fast discrimination of brands of Chinese distilled spirit by means of near infrared transmission spectroscopy was developed. A Field Spec 3 spectroradiometer was used for collecting 60 sample transmission and reflectance data of the three brands of Chinese distilled spirit separately. Then principal component analysis (PCA) was used to process the spectral data after pretreatment. Using the transmission spectra, fifteen PCs were selected with the accumulative reliabilities of 96.466%; while using the reflectance spectra, five PCs were selected with the accumulative reliabilities of 99.867%. These selected PCs would be taken as the inputs of the three-layer back-propagation artificial neural network(BP-ANN), the three brands of Chinese distilled spirit acted as output variety, two discrimination model were obtained. Then the models were used to predict the sample in the validation set. The result showed that 1) a 100% recognition ration was achieved with the threshold predictive error ±0.1 using the transmission spectra; 2) a 60% recognition ration was achieved with the threshold predictive error ±0.1 using the reflectance spectra. It could be concluded that PCA combined with BP-ANN was better method for discrimination of brands of Chinese distilled spirit using the transmission spectra.