PLS-BP法近红外光谱技术同时测定鲜乳中四种主成分  

Determination of four components in raw milk using near infrared spectroscopy by PLS-BP model

作　　者：刘波平[1,2] 荣菡[1] 罗香[2] 邓泽元[1] 曹树稳[1] 

机构地区：[1]南昌大学食品科学与技术国家重点实验室,南昌330047 [2]江西省分析测试中心,南昌330029

出　　处：《分析试验室》2009年第6期66-69,共4页Chinese Journal of Analysis Laboratory

基　　金：教育部南昌大学食品科学重点实验室开放基金(NCU200404);江西省星火计划项目(2005年)资助

摘　　要：基于近红外光谱技术,将偏最小二乘法(Partial Least Squares,PLS)和单隐层的反向传播网络(Back-Propagation Network,BP)联用并测定了鲜乳中4种主成分含量。用PLS法将原始数据压缩为主成分,取前3个主成分的14个数据输入网络,以Kolmogorov定理为依据,经过实验确定中间层的神经元个数为29,初始训练迭代次数为1000,建立了脂肪、蛋白质、乳糖、牛乳总固体4种主成分含量的预测校正模型。PLS-BP模型对样品4个组分含量的预测决定系数(R2)分别为:0.961、0.974、0.951、0.997;本研究为近红外光谱技术在鲜乳多组分快速检测提供了新思路。Partial least squares （PLS） and back-propagation network （BP） prediction model for the determination of fat, protein, lactose, total solids in raw milk had been established with good veracity based on near infrared spectroscopic technique. 14 peak value data from 3 principal components straight ahead compressed from original data by PLS were taken as inputs of while 4 predictive targets as outputs. According to kolmogorov theorem and experiment, 29 nerve cells were chosen as hidden nodes for its prediction with the lowest error. Its training iteration times were supposed as 1000. Predictive correlation coefficients by the model are 0.961, 0.974, 0.951, 0.997. The result shows a new idea for multi-component quantitative analysis of raw milk.

关 键 词：近红外光谱 鲜乳 偏最小二乘 BP网络 

分 类 号：O657.3[理学—分析化学]