一种葡萄糖发酵液中乙醇含量的拉曼光谱分析方法

A Detection Method for Alcohol Content in Glucose Fermentation Based on Raman Spectrum

拉曼光谱法可实现样品的快速定量分析,但发酵液等复杂体系中样品的拉曼强度与被测组分含量之间的良好线性关系难以直接得到。本实验中将待测样品与纯被测组分的拉曼光谱均视为向量,并按照向量点积公式计算向量之间的夹角。在被测组分的拉曼光谱特征峰波段,对多组样品在不同光谱区间的向量夹角矩阵和被测组分含量进行偏最小二乘回归,建立定量分析数学模型,从而以光谱向量之间的夹角作为定量指标实现葡萄糖发酵液中乙醇含量的快速测定。选取的主成分数为6时,所建模型校正集和预测集中预测值与实测值的线性相关系数分别为0.9993和0.9933;校正集均方根误差为0.2171。其建模所用样本数量少、对样品无损坏,方法适用于发酵在线监测和过程分析。

The Raman spectral has been used in quantitative measure in recent years,but good linear relationship between intensity of Raman peak and concentration of measured compo- nent in complex systems was very difficult to obtain. The Raman spectral of measured com- ponent and the glucose fermentation sample are considered as two vectors in n-dimensional space,and their angular as a quantitative criterion in this paper. The Partial Least Squares （PLS） model of alcohol content in glucose fermentation was established by multivariate line- ar regression analysis between the angular and the measured values of alcohol content at Ra- man characteristic peaks ranges. When the principal components number is 6, the correlation coefficient between predicted and measured values in calibration set and prediction set are 0. 9993 and 0. 9933,and the Root Mean Square Error of Calibration （RMSEC） is 0. 2171. The rapid method need small samples number, non-invasive and suitable for on-line monitoring and process analysis.