摘要
采用近红外光谱法测定清汁中还原糖含量,获得良好效果,所建模型标准误差小而决定系数高。数据采用修正的最小二乘法(MPLS)进行计算,所得多元回归线性方程的定标决定系数R2、交互定标标准误差(SECV)、交互定标决定系数(1-VR)分别为0.952,0.067,0.813,具有较好的相关性。用36个随机清汁样品检验该模型,近红外法预测结果与传统滴定法测定结果的检验工作标准误差(SEP)为0.070,检验决定系数为0.876,证明测定所建近红外法定标模型具有较好的稳定性。
The suitability of near infrared（NIR） spectroscopy for the rapid analysis of the content of reducing sugar in clarified juice was tested.Data were analyzed using Modify Partial Least Squares（MPLS）.the Multivariate Linear Regression（MLR） equation shows a high determination coefficient （ R^2 values in calihration）, low standard error of cross validation （SECT）and cross validation （1- VR）, （R^2 = 0.952;SECT =0.067;1- VR =0.813）.36 different data points were used to validate the model developed employing a MPLS regression.The result was good;the coefficient of multiple determination was high（0.876）with a standard error of 0.070. This correlation between NIR and the content of reducing sugar allows us to predict the content of reducing sugar with NIR technology.
出处
《中国甜菜糖业》
2005年第3期1-3,共3页
China Beet & Sugar
基金
广西科学基金"近红外光谱分析在制糖工业的应用机理与数学模型的建立"项目(桂科基0009010)资助
关键词
近红外光谱
清汁
还原糖
甘蔗糖厂
Near infrared spectroscopy,clarified juice,Reducing sugar, Sugarcane factories
