紫外可见光谱结合化学计量学方法用以古井贡酒年份快速鉴别  

Rapid Identification of the Age of Gujing Gongjiu Liquor by Ultraviolet Visible Spectroscopy Coupled with Chemometric Methods

作　　者：张正勇[1,2] 宋超[1] 沙敏[1,2] 刘军[1,2] 王海燕[1,2] 

机构地区：[1]南京财经大学管理科学与工程学院,江苏南京210023 [2]江苏省质量安全工程研究院,江苏南京210023

出　　处：《酿酒科技》2016年第11期20-23,共4页Liquor-Making Science & Technology

基　　金：江苏省高校自然科学研究面上项目(16KJB150015);国家重大科学仪器设备开发专项(2013YQ090703);国家自然科学基金(61373058;71433006);科技部港澳台科技合作专项项目(2014DFM30080);质检公益性行业科研专项(201410173);南京财经大学青年学者支持计划(2015)基金

摘　　要：实验以古井贡酒8年、16年、26年年份酒为对象,研究了基于紫外可见光谱的白酒年份酒快速鉴别方法研发的可能性。通过采集实验样品的紫外可见光谱,构建了该品牌年份酒与紫外可见最大吸收峰(277 nm)强度关系图,古井贡酒8年酒紫外可见最大吸收峰强度围绕1.017±0.127范围波动,古井贡酒16年酒紫外可见最大吸收峰强度围绕1.855±0.410范围波动,古井贡酒26年酒紫外可见最大吸收峰强度围绕2.494±0.130范围波动,基于此可进行初步的年份酒判别分析。同时,利用古井贡酒年份酒紫外可见光谱全谱数据,运用核主成分分析方法,这是主成分分析法在非线性领域的推广,可有效避免数据冗余,提高特征提取效率,进而结合最近邻算法、稀释识别表示分类器,实现白酒年份酒快速、高效、智能鉴别。结果表明,在交叉验证实验条件下识别率可达93.75%。本方案的提出可为白酒年份酒品质保证提供一种简单、快速的鉴别方法。In the experiment, Gujing Gongjiu Liquor of 8, 16 and 26 years respectively was used as the research object, and the feasibility of us-ing ultraviolet visible spectroscopy coupled with chemometric methods for rapid identification of liquor age was studied. The UV-Vis spectra of the experimental samples were collected and then the relationship diagram between liquor age and the intensity of the maximum absorption peak （277 nm） of UV-Vis spectra had been constructed. The intensity of Gujing Gongjiu Liquor of 8 years was located in the range of 1.017 ± 0.127. The intensity of Gujing Gongjiu Liquor of 16 years and of 26 years was located in the range of 1.855 ± 0.410 and in the range of 2.494 ± 0.130, respectively. Accordingly, preliminary discriminate analysis of liquor age could be obtained based on the relationship diagram. Mean-while, total UV-Vis spectra data of Gujing Gongjiu Liquor of 8 years, 16 years and 26 years were investigated by using kernel principal compo-nent analysis （KPCA）, which was the generalized application in nonlinear field of principal component analysis （PCA） and could avoid data re-dundancy as well as improve the efficiency of feature extraction. Finally, rapid, efficient and intelligent identification of Gujing Gongjiu Liquor of different ages could be achieved by using KPCA combined with the nearest neighbor （NN） and sparse representation classification （SRC）. The experimental results displayed that the recognition rate was up to 93.75%. This study provided a simple and rapid method for the identifi-cation of Baijiu of different ages.

关 键 词：核主成分分析 模式识别 紫外可见光谱 年份酒 古井贡酒 

分 类 号：TS261.7[轻工技术与工程—发酵工程]