基于紫外光谱技术和PLS-DA算法烟用爆珠内液的判别分析  

Discrimination of different breakable capsules in cigarette by UV spectra with PLS-DA algorithm

作　　者：许嘉东 唐杰 朱玲 朱立军 谭广璐 彭忠 胡兴锋 Xu Jiadong;Tang Jie;Zhu Ling;Zhu Lijun;Tan Guanglu;Peng Zhong;Hu Xingfeng(Technology Center of China Tobacco Chongqing Industrial Co.,Ltd.,Chongqing 400060,China)

机构地区：[1]重庆中烟工业有限责任公司技术中心,重庆400060

出　　处：《分析仪器》2021年第3期106-111,共6页Analytical Instrumentation

基　　金：重庆中烟工业有限责任公司科研项目“烟用爆珠内液质量稳定性检测方法研究”(HX2017006)。

摘　　要：为了快速、准确的判别不同牌号烟用爆珠类型及其一致性,采用无水乙醇稀释爆珠内液,紫外可见光谱(UV)扫描,平滑后标准正态变量变换(SNV)预处理光谱,偏最小二乘判别分析(PLS-DA)算法建立了烟用爆珠的分类判别模型,并对模型进行了验证。结果表明:3个牌号烟用爆珠PLS模型的校正集分类变量的预测值与实测值的相关系数均≥0.9562,模型拟合性好,PLS-DA模型对于参与建模的310个烟用爆珠样本(校正集)和未参与建模的60个烟用爆珠样本(验证集)的预测准确率均为100%,说明烟用爆珠PLS-DA模型预测准确度高,判别效果好。基于PLS-DA模型的分类变量值建立的单值控制图可以对烟用爆珠内液的一致性进行快速判定。因此,紫外光谱技术结合PLS-DA算法可用于烟用爆珠内液的判别分析,且方法快速准确、经济环保。In order to discrimination the type and stability of different breakable capsules in cigarette,the inner contents of breakable capsules were dissolved in absolute ethyl alcohol,then the UV spectra of samples were collected and pre-processed by smoothing and standard normal variate transformation(SNV).Three discrimination models with PLS-DA algorithm on the basis of UV spectra were established and validated.The results showed that the correlation coefficient between the predicted values and the measured values of the three brands of PLS models’calibration sets were more than or equal to 0.9562,and the models were significantly fit well.All prediction accuracy of models for samples in calibration set and validation set were 100%,it showed that the models were credible,accurate,and effective.The stability of the breakable capsules could be effectively discriminated by the the single-value control chart based on the classification variable value of PLS-DA model.The method is rapid,accurate,economical,and environment-friendly.

关 键 词：烟用爆珠 紫外可见光谱 PLS-DA算法 模型 控制图 

分 类 号：TS426[农业科学—烟草工业] O657.32[理学—分析化学]