近红外光谱混合模型定量分析不同物理状态样品的研究

发布时间：2018-07-08 09:02

本文选题：校正模型 + 均质样本　；参考：《分析化学》2017年07期

【摘要】：近红外光谱(NIRS)以漫反射模式对非均质样本进行测量时,由于其光谱散射和吸收系数差异较大,建立的校正模型准确性和稳健性较低,因此,本研究提出了一种基于均质样本和模型转移方法建立混合模型的策略,解决非均质样本近红外光谱检测的问题。以烟叶样本为研究对象,分别建立了基于Shenk专利算法(Shenk's)、分段直接标准化(PDS)和基于典型相关分析的模型转移算法(CTCCA)的烟粉+烟丝、烟粉+烟片混合模型,用于烟丝和烟片样本中烟碱含量的预测。结果表明,混合模型对烟丝和烟片样本的预测均方误差(RMSEP)较直接建模分别降低了1.39%和2.73%,预测结果有一定的改善,稳健性提高,3种方法中CTCCA表现最优。因此,采用近红外光谱均质模型和模型转移方法建立的混合模型对非均质样本的测定具有可行性,有利于在线近红外光谱分析技术的发展,可为近红外光谱模型的共享提供参考。
[Abstract]:Near infrared spectroscopy (NIRS) is used to measure heterogeneous samples in diffuse reflectance mode. Due to the difference in spectral scattering and absorption coefficients, the accuracy and robustness of the calibration model are low. In this study, a hybrid model based on homogeneous sample and model transfer method is proposed to solve the problem of near-infrared spectrum detection of heterogeneous samples. Based on Shenk patent algorithm (Shenks), piecewise direct standardization (PDS) and model transfer algorithm based on canonical correlation analysis (CTCCA), a mixture model of tobacco powder and tobacco powder was established. It is used to predict nicotine content in tobacco and tobacco samples. The results showed that the mean square error (RMSEP) of the mixed model for tobacco samples and tobacco samples was reduced by 1.39% and 2.73%, respectively, and the prediction results were improved to some extent. The CTCCA performance was the best among the three methods. Therefore, the hybrid model based on NIR homogeneous model and model transfer method is feasible for the determination of heterogeneous samples, which is beneficial to the development of on-line NIR analysis technology. It can be used as a reference for the sharing of near infrared spectral model.
【作者单位】：湖南农业大学生物科学技术学院;中南大学化学化工学院;
【基金】：国家自然科学基金项目(No.21275164) 湖南省研究生科研创新项目(No.CX2015B237)资助~~
【分类号】：O657.33

【相似文献】