摘要
利用乙醇、果糖和葡萄糖在波数为１４０００～７５００ｃｍ－１范围内的吸收值，在近红外谱区用偏最小二乘法（ＰＬＳ）对体系进行建模，并通过内部和外部样品校验确认最佳数据预处理方法。结果表明，乙醇在０．０５～０．２５ Ｌ／Ｌ，果糖在０．０１～０．０５ ｇ／ｍｌ／Ｌ及葡萄糖在０．００５～０．００９ｇ／ｍＬ的范围内外部校验模型较好的分别是在ＰＬＳ中所用的一阶偏导（平滑点数是５）、二阶偏导（平滑点数是５）和直线消除法的数据预处理方法所建立的模型，其外部校验预测结果的相对误差在２％左右。此方法具有可同时测定不同样品，简便快速及成本低等优点。
Partial least square was used to establish the prediction model for simultaneous determination of ethanol, fructose and glucose by using the absorbance data in the range of 14000 similar to 7500 cm(-1). Eight data preprocessing methods were examined by both the internal validation of the calibration set and the external validation of the test samples. It was observed that the best result was obtained by using different pre-treating methods for each component, the first derivative for ethanol, second derivative for fructose, and straight-line subtraction for glucose respectively. At the concentration level of 0.05 similar to 0.25 L/L for ethanol, 0.005 similar to 0.009 g/mL for fructose and 0.01 similar to 0.05 g/mL for glucose, the relative errors of prediction were all about 2% for the three components. The results show that near infrared spectroscopy combining with chemometrics method may be a simple, effective and low cost method for the simultaneous determination of the multi-component system.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
2001年第10期1213-1215,共3页
Chinese Journal of Analytical Chemistry
基金
浙江省分析测试基金资助项目
