近红外光谱法在线测定白砂糖二氧化硫的研究  

Study on determination of white granulated sugar SO_2 by on-line near infrared spectroscopy

作　　者：黎庆涛[1] 王远辉[1] 潘路路[1] 姜毅[1] 黄康宁[1] 李震宇[1] 

机构地区：[1]广西大学轻工与食品工程学院,南宁530004

出　　处：《食品科技》2010年第9期324-327,共4页Food Science and Technology

摘　　要：针对糖厂的需求,提出了用近红外光谱技术在线快速检测白砂糖中二氧化硫的新方法。通过比较不同的光谱预处理方法,得到最佳方法,即Savitzky-Golay一阶求导和正交信号校正(OSC)联合应用。通过比较主成分回归(PCR)、偏最小二乘法(PLS)和人工神经网络(ANN),得到最优模型,即主成分数选4的PLS模型,校正集的相关系数Rc为0.966,校正标准偏差(SEC)为1.086,预测标准偏差(SEP)为2.519。结果表明,近红外光谱法测定白砂糖二氧化硫是可行的,为在线快速检测白砂糖中的二氧化硫提供了新的途径。A new method of on-line fast determination of white granulated sugar SO2 by near infrared spectrum（NIRS） was developed for sugar-refinery.By comparison of different pretreating methods,Savitzky-Golay first derivative and orthogonal signal correction（OSC） was the best pretreating method.By comparison among principal component regression（PCR）,partial least square（PLS） and artificial neural networks（ANN）,PLS modern with main component 4 was the best modern,calibration volume＇s Rc was 0.966,standard error of calibration variable（SEC） was 1.086,standard error of prediction（SEP） was 2.519.The ultimate experimental results indicated that NIRS was feasible in determination of white granulated sugar SO2.This paper could offer a new approach of on-line fast determination of white granulated sugar SO2.

关 键 词：近红外光谱 白砂糖 二氧化硫 偏最小二乘法 人工神经网络 

分 类 号：O657.39[理学—分析化学]