近红外与傅里叶变换红外光谱技术对麦麸固体发酵饲料中不同成分的监测研究  

Study on Analysis of Different Components in Solid State Fermentation of Wheat Bran Based on NIR and FT-IR Spectroscopy

作　　者：刘海涛 张宝伟 吴勃 王云龙 王昌禄[1] 刘欢欢 郭庆彬 LIU Haitao;ZHANG Baowei;WU Bo;WANG Yunlong;WANG Changlu;LIU Huanhuan;GUO Qingbin(College of Food Science and Engineering,Tianjin University of Science&Technology,Tianjin 300457,China;Tianjin Jincheng Intelligent Technology Co.,Ltd.,Tianjin 300202,China;Hangzhou Baoankang Biotechnology Co.,Ltd.,Zhejiang Hangzhou 311500,China)

机构地区：[1]天津科技大学食品科学与工程学院,天津300457 [2]天津锦城智能科技有限公司,天津300202 [3]杭州保安康生物技术有限公司,浙江杭州311500

出　　处：《饲料工业》2024年第16期115-122,共8页Feed Industry

基　　金：国家自然科学基金[32272270、32072173];省部共建食品营养与安全国家重点实验室开放课题[SKLFNS-KF-202103、SKLFNS-KF-202004];天津科学技术局计划项目[21ZYJDJC00110]。

摘　　要：为实现光谱技术对麦麸固体发酵过程中不同成分变化的在线监测,通过国家标准方法测定61份麦麸固体发酵饲料样本的蛋白质、水分、总酚和粗纤维含量,采集样本近红外光谱(NIR)和傅里叶变换红外光谱(FT-IR),经过标准正态变换(standard normal variate transformation,SNV)、多元散射校正(multiplicative scatter correction,MSC)、平滑(smoothing)等9种预处理方法对原始光谱进行校正,结合偏最小二乘法(partial least squares,PLS)建立4种成分的NIR和FT-IR定量分析模型并进行比较分析。结果表明:所建立的4种成分NIR和FT-IR模型的训练集决定系数(Rc^(2))和验证集决定系数(Rp^(2))均大于0.8,交叉验证均方根误差(root mean square error of cross validation,RMSECV)小于2.0,训练集均方根误差(root mean square error of calibration,RMSEC)和验证集均方根误差(root mean square error of prediction,RMSEP)小于1.0。因此,所建立的NIR和FT-IR定量分析模型具有较好的准确性和稳定性,能够对麦麸固体发酵过程中不同成分变化实行快速监测。To on-line monitor the process variables of different components in solid state fermentation of wheat bran by spectral technology,the contents of protein,water,total phenol and crude fiber in 61 wheat bran solid fermented feed were measured based on Chinese national standard metheds.The near infrared spectrum and Fourier transform infrared spectrometer of the samples was collected and analyzed.The spectra was performed with 9 different preprocessing methods,such as standard normal variate transformation(SNV),multiplicative scatter correction(MSC)and smoothing(SG).NIR and FT-IR quantitative models of feed nutrients were established and compared by partial least squares(PLS).The results showed that the Rc^(2) and Rp^(2) were both great than 0.8,root mean square error of cross validation(RMSECV)was less than 2.0,root mean square error of calibration(RMSEC)and root mean square error of prediction(RMSEP)were less than 1.0,respectively.NIR and FT-IR quantitative analysis models have good stability.The process parameters of solid-state fermentation of wheat bran were rapidly monitored by NIR and FT-IR is feasible.

关 键 词：饲料 固体发酵 近红外光谱 傅里叶变换红外光谱 偏最小二乘法 定量分析模型 

分 类 号：S816[农业科学—饲料科学]