三维演进特征投影图及其量化方法识别联用色谱数据中选择性区域

Identification of Selective Regions in Overlapping Hyphenated Chromatographic Data Using Three-dimensional Latent Projective Graph(3D-LPG) and Chemometric Technique

提出三维演进特征投影图 ( 3 D-LPG)和定量识别三维 LPG上直线段的方法 ,即 T-图法 ,并采用上述方法对模拟体系和实验体系进行了解析 .结果表明 ,三维 LPG可提供比二维 LPG更丰富的信息 ,当组分的光谱非常相似时 ,只有三维 LPG才能正确地显示出重叠色谱峰中选择性区域的范围 ;同时 ,T-图法可准确地指示出三维 LPG上直线段区域的所在范围 ,从而可正确识别出体系的选择性区域 。

Selective regions of the overlapping hyphenated chromatographic data are crucial for the resolution via chemometric methods. The latent projective graph(LPG) is the common technique for identifying the selective regions, in which the selective regions are revealed by the appearance of sequences of consecutive points mapping as segments of straight lines. However, according to our results, the straight lines in the LPG of the system comprising some closely eluting components with fairly similar spectra may not always necessarily correspond to the real selective regions. In this work, three-dimensional LPG(3D-LPG) was developed, and a new chemometric technique T-plot was proposed to identify the linear segments of 3D-LPG corresponding to the selective regions. The performance and efficiency of the proposed methods were illustrated on both the simulated and experimental data, and the results show that the 3D-LPG can provide much more information and deeper insights into the structure of the data, outperforming the present 2D version in revealing the selective regions when the spectra of some coeluting components are quite similar. Moreover, T-plot is proved to be effective in identifying the linear segments in 3D-LPG.