基于光学性质的多组分液体混合物定量分析

Quantitative Analysis of Multicomponent Liquid Mixtures Based on Optical Properties

为了对多组分液体混合物进行定量分析,建立了窗片-待测液-窗片三层光学腔双光程光学常数反演模型。以乙醇和蒸馏水的混合物为研究对象,利用双光程模型对各混合物样本的光学常数进行反演;引入牛顿混合规则,以0.01的混合梯度对乙醇-蒸馏水混合物样本的折射率进行预测;联合混合物折射双光程反演值和混合规则预测值组成定量分析模型折射率数据集,利用Savitzky-Golay法对折射率数据集进行预处理;基于折射率预处理光谱利用区间偏最小二乘法建立液体混合物定量分析模型。结果表明:最佳区间划分方式为30等分,最佳子区间为第21区间,最佳建模波段为0.428~0.448μm,此时预测集样本预测值的相对误差分别为7.15%和4.93%,表明定量分析模型具有较高的预测精度,能准确预测液体混合物各组分的含量。

In order to quantitatively analyze multicomponent liquid mixtures,a double optical path inversion model of three-layer optical cavity with window-liquid-window was established.Taking the mixture of ethanol and distilled water as the research object,the optical constants of different mixture samples were inverted by using the double optical path inversion model;The refractive index of ethanol-distilled water mixture was predicted by introducing Newton mixing rule with the using mixing gradient of 0.01;The refractive index data sets of the quantitative analysis model was composed of the double optical path inversion value of mixture and the predicted value of mixing rule;The Savitzky-Golay method was used to preprocess the refractive index data sets,and the interval partial least squares method was used to establish the quantitative analysis model of liquid mixtures based on the refractive index pretreatment spectra.The results showed that the optimum interval dividing method was 30 equal parts,the optimum sub-interval was interval 21,and the optimum modeling band was 0.428~0.448μm.The relative errors of the predicted values of the predicted samples were 7.15%and 4.93%,respectively,which indicated that the quantitative analysis model has high prediction accuracy and can accurately predict the content of each component of multicomponent liquid mixtures.