多元组分含量的在线定量检测对于大多数过程分析具有重要意义, 但是, 在多元统计方法中, 由于组分光谱相似、 叠加会导致多元统计误差增大, 在高相似组分体系中尤为突出。 基于子空间角度转换的化学计量学方法, 求取体系拉曼光谱响应后将光谱降噪平滑, 消除差异后求取待测组分的纯光谱的角度值, 将待测物样品含量转为百分比, 计算角度值与组分含量的线性关系, 建立角度方差值与建模样本组分含量值的关系模型, 可实现对组分含量的实时跟踪。 以蔗糖水解过程为分析对象, 以蔗糖含量为分析目标, 采集蔗糖、 果糖、 葡萄糖不同浓度配比下的混合溶液拉曼光谱信号, 先进行二阶求导消除加性误差, 再经平滑降噪后, 将其转化为空间向量角度以消除批次间光谱差异, 建立移动窗口, 求取系列夹角值的方差Dxi, 由此利用角度转换算法, 建立了蔗糖含量和光谱夹角值方差的相关模型, 相关系数r可达0.997, 验证可得模型相对误差在-3.390%～7.333%, 模型预测效果良好。 通过拉曼光谱对不同条件下的蔗糖水解过程进行监测, 分别计算出反应过程中光谱的系列角度值方差Db, 带入模型得到组分含量, 进而求得不同条件下的反应速率r, 发现反应速率随相应条件变化而变化, 验证了蔗糖水解过程是一级反应, 表明采用拉曼光谱结合角度转换法可快速跟踪蔗糖水解过程中组分含量。 利用该方法采集的数据计算了26.5 ℃下蔗糖水解反应速率常数K1为0.031, 同温度下旋光法测定反应速率常数为0.031 5, 二者相接近; 利用该方法计算40 ℃下反应速率常数K2为0.197 8, 带入阿伦尼乌斯方程得到活化能Ea=107.1 kJ·mol-1, 与文献值相符。 表明光谱分析与角度值转换的化学计量学方法相结合可用于蔗糖水解过程动力学研究, 为多组分定量分析提供一种高效快捷的方法。

Online quantitative detection of the content of multivariate components is of great significance for most process analysis. However, in multivariate statistical methods, the spectral similarity and superposition of components will lead to an increase of multivariate statistical error, which is particularly prominent in systems with high similarity components. Chemometrics method based on subspace Angle conversion, calculate the system after the Raman spectral response spectrum noise smoothing, eliminate the differences after calculating the Angle of the pure spectra of the component under test values, a sample content object under test to the percentage, calculated Angle value and component content, the linear relationship between the Angle values of variance and modeling sample components content value relation model, which can realize real-time tracking of component content. Based on sucrose hydrolysis as analysis object, with the aim to the analysis of sucrose content, acquisition of sucrose, fructose, glucose mixture under different concentration ratio of Raman spectrum signal, the first to the second order derivative, eliminate the additive error after smoothing noise is transformed into space vector Angle to eliminate spectral differences between batches, move the window, set up to calculate the Dxi series Angle value of variance, this algorithm of Angle transformation, established the sucrose content and relevant model of spectral Angle value variance, correlation coefficient r of 0.997, model validation available relative error in 3.390%~7.333%, The prediction result of the model is good. Through the Raman spectra of different conditions of sucrose hydrolysis process monitoring, respectively to calculate the response spectra in the process of value variance Db series Angle, to the model component content, then the reaction rate of different conditions is r, found that the reaction rate changes over the corresponding conditions, both close to the same proportion decreases, and verified the sucrose hydrolysis process is first order reaction, shows that the Raman spectroscopy combined with Angle conversion method can fast track component concentration of sucrose hydrolysis process. Using the data collected by this method, the reaction rate constant K1 of sucrose hydrolysis at 26.5 ℃ was calculated as 0.031, and the reaction rate constant 0.031 5 was determined by the optical rotation method at the same temperature. The reaction rate constant K2 at 40 ℃ was calculated to be 0.197 8, and the activation energy Ea=107.1 kJ·mol-1 was obtained by substituting the Arrhenius equation, which was consistent with the literature value. The results show that the Raman spectrum combined with Angle conversion algorithm can be used to study the dynamics of sucrose hydrolysis process and provide an efficient and fast method for multi-component quantitative analysis.