利用近红外反射光谱检测豆粕中掺假棉籽粕的研究  

Detection of cottonseed meal as an adulterant in soybean meal by near-infrared reflectance spectroscopy

作　　者：邓凯东[1] 潘文静 陆牡龙 边高瑞 唐倩[1] 陈玉华[1] 何晓芳 DENG Kaidong;PAN Wenjing;LU Mulong;BIAN Gaorui;TANG Qian;CHEN Yuhua;HE Xiaofang(College of Animal Science and Food Engineering,Jinling Institute of Technology,Nanjing,Jiangsu Province 210038,China)

机构地区：[1]金陵科技学院动物科学与食品工程学院,江苏南京210038

出　　处：《中国饲料》2022年第16期78-82,共5页China Feed

基　　金：国家自然科学基金面上项目(41475126);江苏省园艺学重点学科(苏教研函(2022)2号);金陵科技学院畜牧学重点学科;金陵科技学院科技创新团队。

摘　　要：本实验研究了通过近红外反射(NIR)光谱对豆粕中掺假棉籽粕的定量分析方法。在豆粕中分别掺入0%、5%、10%、15%、20%、25%、30%(m/m)棉籽粕,每个处理包含建模集样品20个、验证集样品10个,使用傅立叶变换近红外光谱仪获取掺假样本的NIR光谱,采用OPUS化学计量学软件拟合建模集样品棉籽粕掺假率的NIR光谱预测模型,再用验证集样本对预测模型进行验证,评价预测模型的相对误差。结果表明,豆粕中棉籽粕掺假率的NIR光谱预测模型的决定系数(R~2)为0.975、交叉验证均方根误差(RMSECV)为1.58、斜率为0.974,棉籽粕掺假比例为5%、10%、15%、20%、25%、30%的NIR光谱预测值的相对误差分别为15.0%、0.50%、1.87%、1.45%、1.28%、0.87%。因此,利用NIR光谱可以准确定量测定豆粕中棉籽粕的掺假比例。This experiment explored a quantitative method for measuring the adulteration of cottonseed meal in soybean meal by near-infrared reflectance(NIR)spectroscopy. Cottonseed meal was added into soya bean meal at the rate of 0%,5%,10%,15%,20%,25% and 30%(w/w)with 20 replicates of calibration and 10 of validation for each treatment. Adulterated soya meal samples were scanned for NIR spectra using a Fourier transformed NIR spectrometer,and a quantitative predictive model was developed using a chemometric software OPUS. The developed model was eventually validated for its relative errors of predictions.The results showed that the coefficient of determination(R~2)of the NIR prediction model was 0.975,the root mean square error of cross-validation(RMSECV)was 1.58,and the slope was 0.974. The relative errors of the predicted values were 15.0%,0.50%,1.87%,1.45%,1.28% and 0.87% for the adulteration rates of cottonseed meal of 5%,10%,15%,20%,25% and 30%,respectively. Therefore,adulteration of cottonseed meal in soybean meal was accurately quantified by the NIR technique.

关 键 词：豆粕 棉籽粕 掺假 近红外反射光谱 预测模型 

分 类 号：S816.17[农业科学—饲料科学]