基于紫外-近红外融合光谱对“互助”青稞酒判别分析研究  

Qualitative identification of Chinese Huzhu Qingke liquor based on ultraviolet spectroscopy combined with near-infrared spectroscopy

作　　者：赵玉霞 王茹 张世芝 殷博 张明锦 ZHAO Yuxia;WANG Ru;ZHANG Shizhi;YIN Bo;ZHANG Mingjin(College of Chemistry and Chemical Engineering,Qinghai Normal University,Xining 810016,China;College of Chemistry and Chemical Engineering,Qinghai Minzu University,Xining 810016,China;Qinghai Key Laboratory of Advanced Technology and Application of Environmental Functional Materials,Xining 810016,China)

机构地区：[1]青海师范大学化学化工学院,青海西宁810016 [2]青海民族大学化学化工学院,青海西宁810016 [3]青海省环境功能材料先进技术与应用重点实验室,青海西宁810016

出　　处：《食品与发酵工业》2025年第4期301-308,共8页Food and Fermentation Industries

基　　金：国家自然科学基金项目(22363010);青海省自然科学基金(2022-ZJ-769)资助。

摘　　要：“互助”青稞酒是中国地理标志产品,对其快速、准确识别至关重要。该研究利用紫外光谱(ultraviolet spectrum,UV)、近红外光谱(near infrared spectrum,NIR)及UV-NIR融合光谱结合化学计量学方法对“互助”青稞酒进行判别分析。对上述三类光谱数据,分别考察4种光谱预处理方法和5种特征选择方法后确定最佳处理方法,并建立了“互助”青稞酒的偏最小二乘判别分析(partial least squares discriminant analysis,PLS-DA)模型。结果表明,融合光谱能够互补多元化学信息,提高分类模型性能,且中级数据融合光谱建立的模型性能优于低级数据融合光谱。其中用二阶导数预处理、用逐步投影算法筛选变量后得到的中级数据融合光谱建立的PLS-DA模型预测能力最佳,在训练集上的灵敏度、特异性、准确度和曲线下面积值分别为1、1、0.9773和1,在测试集上的灵敏度、特异性和曲线下面积值分别达到1、0.8276和0.9667。紫外-近红外中级数据融合光谱结合化学计量学方法可用于“互助”青稞酒的快速判别分析。Chinese Huzhu Qingke liquor is a geographical indication product in China,and its rapid and accurate identification is essential.In this study,ultraviolet spectroscopy(UV),near-infrared spectroscopy(NIR),and UV-NIR fusion spectroscopy combined with chemometrics were used to identify Huzhu Qingke liquor.For the above three types of spectral data,four spectral pre-processing methods and five feature selection methods were investigated to determine the optimal processing method,and a partial least squares discriminant analysis(PLS-DA)model of Huzhu Qingke liquor was established.Results show that the fusion spectra could complement each other and improve the performance of the classification model,and the model built by intermediate data fusion spectra was better than that built by low-level data fusion spectra.Among them,the PLS-DA model built with intermediate-level data fusion spectra obtained after preprocessing with second-order derivatives and screening variables with a stepwise projection algorithm had the best predictive ability,with sensitivity,specificity,accuracy,and AUC values of 1,1,0.9773,and 1 on the training set,and the sensitivity,specificity,and AUC values reached to 1,0.8276,and 0.9667 on the test set,respectively.UV-NIR intermediate data fusion spectroscopy combined with chemometrics can be used for the rapid discriminant analysis of Huzhu Qingke liquor.

关 键 词：中国“互助”青稞酒 中级数据融合 化学计量学 偏最小二乘判别分析 

分 类 号：TS262[轻工技术与工程—发酵工程] O657.33[轻工技术与工程—食品科学与工程]