Application of visible/near infrared spectroscopy and chemometric calibrations for variety discrimination of instant milk teas

Visible and near infrared (Vis/NIR) spectroscopy combined with back propagation neural network (BPNN) and least squares-support vector machine (LS-SVM) was investigated to implement the fast discrimination of instant milk teas. Five brands of milk teas were obtained. The effective wavelengths (EWs) were selected according to x-loading weights and regression coefficients by partial least squares (PLS) analysis. A total of 18 EWs were selected as the inputs of BPNN and LS-SVM models with a comparison of principal components (PCs). The prediction precision and recognition ratio was 98.7% in validation set by both PC and EW models. The results indicated that the EWs reflected and represented the main characteristics of milk tea, and the variety discrimination was successfully implemented using Vis/NIR spectroscopy based on BPNN and LS-SVM. The selected EWs would be helpful for the development of portable instruments for commercial applications of variety and quality detection of milk teas.