基于偏最小二乘与广义回归神经网络的近红外光谱测定土豆中3种营养成分的研究  

Study on the determination of three components in potatoes using near infrared spectroscopy based on partial least squares and generalized regression neural network model

作　　者：刘波平[1] 秦华俊[2] 罗香[3] 曹树稳[2] 王俊德[1] 

机构地区：[1]南京理工大学化工学院,南京210094 [2]南昌大学食品科学教育部重点实验室,南昌330047 [3]江西省分析测试中心,南昌330029

出　　处：《分析试验室》2007年第9期38-41,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(2005年)项目资助

摘　　要：偏最小二乘(partial least squares,PLS)与广义回归神经网络(generalizedregression neural networks,GRNN)联用对土豆样品建立起粗纤维、淀粉、蛋白质含量的预测校正模型,用PLS法将原始数据压缩为主成份,取前3个主成份的12个特征吸收峰输入GRNN网络,网络光滑因子iσ为0.1。PLS-GRNN模型对样品3个组分含量的预测决定系数(R2)分别为:0.945、0.992、0.938。结果表明,近红外光谱技术可以快速、准确地同时测定土豆中的粗纤维、淀粉、蛋白质,该方法可应用于果蔬产业的品质管理与控制。Partial least squares （PLS） and generalized regression neural network （GRNN） prediction model for the determination of fibre, starch and protein in potatoes had been estabhshed with good veracity. 12 peak value data from 3 principal components straight compressed from original data by PLS were taken as inputs of GRNN while 3 predictive targets as outputs. 0.1 was chosen as smoothing factor for its good approximation and prediction with the lowest error compared with 0.2, 0.3, 0.4, 0.5. Predictive correlation coefficient of three components by the model are 0. 945, 0. 992, 0.938. The results show that PLS-GRNN used in NIRS is a rapid, effective means for measuring fibre, starch and protein in potatoes. The results are important for quality control and evaluation in fruit and vegetable industry, and can also be used for quantitative analysis of other samples.

关 键 词：近红外光谱 土豆 偏最小二乘 GRNN网络 多组分检测 

分 类 号：O657.33[理学—分析化学]