Identification of liquors from the same brand based on ultraviolet, near-infrared and fluorescence spectroscopy combined with chemometrics.

Accurate identification of various liquors from the same brand is of great significance for safeguarding the rights and interests of consumers and the market economy. Here, the spectral properties of liquors were studied based on ultraviolet (UV), near-infrared (NIR) and multi-way fluorescence spectroscopy. Then these liquors were distinguished by integrating their spectral properties with the chemometrics such as Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), and Backpropagation Neural Networks (BPNN). To improve the accuracy, sensitivity, and specificity of the liquor identification, a four-way fluorescence spectrum data array was constructed by adding three acid-sensitive quantum dots (QDs) as an additional dimension. Combined with mid-level data fusion, this strategy can identify liquors from the same brand with the accuracy, sensitivity, and specificity of 99.17%, 99.15%, and 99.96%. In addition, an automated analysis platform based on MATLAB App Designer was developed to improve the efficiency of spectral data modeling.