Key Laboratory for Biorheological Science and Technology of Ministry of Education, Bioengineering College of Chongqing University, Chongqing 400044, PR China.
Key Laboratory for Biorheological Science and Technology of Ministry of Education, Bioengineering College of Chongqing University, Chongqing 400044, PR China; Chongqing University Three Gorges Hospital, Chongqing 404000, PR China.
Food Chem. 2023 Jan 30;400:134064. doi: 10.1016/j.foodchem.2022.134064. Epub 2022 Sep 5.
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.
准确识别同一品牌的各种酒类对于维护消费者权益和市场经济具有重要意义。本研究基于紫外（UV）、近红外（NIR）和多向荧光光谱法，研究了酒类的光谱特性。然后，通过将光谱特性与线性判别分析（LDA）、支持向量机（SVM）和反向传播神经网络（BPNN）等化学计量学方法相结合，对这些酒类进行了区分。为了提高酒类识别的准确性、灵敏度和特异性，通过添加三个酸敏量子点（QDs）作为附加维度，构建了四路荧光光谱数据阵列。结合中层数据融合，该策略可以以 99.17%、99.15%和 99.96%的准确率、灵敏度和特异性识别同一品牌的酒类。此外，还基于 MATLAB App Designer 开发了一个自动化分析平台，以提高光谱数据建模的效率。
