基于近红外反射光谱快速检测豆粕中掺假菜籽粕的研究  

Study on rapid detection of adulterated rapeseed meal in soybean meal based on near infrared reflectance spectroscopy

作　　者：邓凯东[1] 陆牡龙 潘文静 边高瑞 唐倩[1] 陈玉华[1] 何晓芳 DENG Kai-dong;LU Mu-long;PAN Wen-jing;BIAN Gao-rui;TANG Qian;CHEN Yu-hua;HE Xiao-fang

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：《饲料研究》2023年第11期129-132,共4页Feed Research

基　　金：国家自然科学基金面上项目(项目编号:41475126);江苏省重点学科(项目编号:苏教研函(2022) 2号);金陵科技学院畜牧学重点学科;金陵科技学院生态养殖信息化科技创新团队。

摘　　要：试验旨在研究豆粕中掺假菜籽粕的近红外反射(NIR)光谱定量分析方法。在豆粕中掺入0、5%、10%、15%、20%、25%、30%菜籽粕,每个掺杂水平30份重复样品(建模集样本20个,验证集样本10个)。采用傅里叶变换近红外光谱仪获取样品NIR光谱,使用化学计量学软件拟合建模集样品掺假比例的NIR光谱预测模型,验证集样本评价预测模型的准确度。结果表明,豆粕中菜籽粕掺假比例的NIR光谱预测模型的决定系数R2为0.983,交叉验证均方根误差(RMSECV)为1.30。菜籽粕掺假比例为5%、10%、15%、20%、25%、30%的NIR光谱预测值的相对误差分别为8.93%、12.20%、2.21%、1.17%、1.72%、1.69%。研究表明,使用NIR光谱可以对豆粕中菜籽粕的掺假量实现准确定量测定。The experiment was to explore a quantitative method for measuring the adulteration of rapeseed meal in soybean meal by NIR spectroscopy.Soybean meal was mixed with 0,5%,10%,15%,20%,25%,30%rapeseed meal with 30 duplicate samples for each doping level(20 samples for modeling set and 10 samples for verification set).Fourier transform near-infrared spectrometer was used to obtain NIR spectra of samples,and stoichiometric software was used to fit NIR spectral prediction model of sample adulteration ratio in modeling set,and the accuracy of sample evaluation prediction model was verified.The results showed that the NIR prediction model had a coefficient of determination R^(2)was 0.983 and RMSECV was 1.30.The relative errors of the predicted values were 8.93%,12.20%,2.21%,1.17%,1.72% and 1.69%for the adulteration rates of rapeseed meal of 5%,10%,15%,20%,25% and 30%,respectively.The study indicates that adulteration of rapeseed meal in soybean meal is accurately quantified by the NIR technique.

关 键 词：豆粕 菜籽粕 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]