近红外光谱与化学计量学方法用于镉污染稻米的定性鉴别  

Near Infrared Spectroscopy Combining with Chemometrics for Qualitative Identification of Cadmium-Polluted Rice

作　　者：朱向荣[1,2] 李高阳[1,2] 黄绿红[1] 苏东林[1,2] 刘伟[1,2] 单杨 

机构地区：[1]湖南省农业科学院湖南省食品测试分析中心,长沙410125 [2]中南大学研究生院隆平分院,长沙410125

出　　处：《分析化学》2015年第4期599-603,共5页Chinese Journal of Analytical Chemistry

基　　金：科技部"十二五"国家科技支撑计划(No.2012BAK17B17);农业部农业科研杰出人才培养计划;湖南省科技重大专项(No.2011FJ1002-4)资助项目~~

摘　　要：采用近红外光谱漫反射模式结合化学计量学方法对稻米镉含量是否超标进行可行性鉴别分析。本研究收集了120个样本,测定其镉含量值(合格49个,不合格71个)。对光谱数据预处理方法优化,确定了平滑,一阶导数以及自归一化后的数据作为输入变量。采用竞争性自适应重加权算法筛选了45个关键变量,并对上述变量的光谱吸收带进行归属。比较了主成分分析-判别分析法、偏最小二乘识别分析、线性判别分析、K-最近邻法与簇类独立软模式法5种模式识别方法。确定采用偏最小二乘识别分析建模效果最好,模型训练集与预测集鉴别准确率分别达到98.8%与91.7%。结果表明,近红外光谱作为初筛方法可用于鉴别稻米中镉含量是否超标。Near-infrared（ NIR） diffuse reflectance spectroscopy and chemometrics method were used to discriminate cadmium-polluted rice. The samples set contained 120 spectra of qualified（ n = 49） and excessive（ n = 71） was collected and scanned. After optimization,a combination（ smoothing coupled with first derivative and mean centering） was utilized as a spectral pretreatment method. Competitive adaptive reweighed sampling（ CARS） was adapted to selected 45 key variables,and each band of the variables was assigned. Five modeling methods including partial least squares discriminant analysis（ PLS-DA）,linear discriminant analysis（ LDA）,K-nearest neighbor（ KNN）,soft independent modeling class analog（ SIMCA）and principal component analysis-discriminant analysis（ PCA-DA） were used and compared. PCA-DA was finally selected as the optimal qualitative model. The accuracy rate of training set and testing set for PCA-DA method was 98. 8% and 91. 7%,respectively. The results showed that NIR spectroscopy could be used as a rapid,non-destructive and convenient analytical method for primary screening and detecting cadmium-polluted rice.

关 键 词：近红外光谱 化学计量学 稻米 镉 污染 定性鉴别 

分 类 号：O657.33[理学—分析化学] TS210.7[理学—化学]