正交投影回归法和HLAE用于解析灰色体系

提出一种基于正交投影的用于判断未知干扰组分的新方法--正交投影回归法(OrthogonalProjection-basedRregression,OPR),可指示出复杂体系中未知干扰组分存在与否及存在区间;在此基础上,应用最近提出的扩展的HLA算法(HLAE),对一维生素混合体系进行了分析,并与HLA和PLS的结果进行了比较.结果表明,OPR是HLAE得以成功应用的前提;不需待测组分的精确光谱,HLAE可得到优于PLS和至少相当于HLA的结果,在纯光谱不够精确的情况下,完全可以用HLAE代替HLA.

Application of Orthogonal Projection-based Algorithms and HLAE in the Analysis of Grey System

A new wavelength selection method, orthogonal projection-based regression(OPR), was presented, which can indicate whether the unknown interferences exist in the determined system and show the wavelength region where the unknown interferences do not have absorbance. A multi-vitamine system was analyzed by using the newly developed multivariate calibration method-extension of hybrid linear analysis(HLAE) after the wavelength selection of OPR, and the results were compared to those of HLA and PLS. The results showed that only after the wavelength selection of OPR, can satisfactory results be got by using HLAE; and without pure spectra of analyzed components, HLAE can produce the results better than PLS and at least equivalent to HLA, and can be used as a replacement of HLA when it is difficult to get the precise pure spectra of the determined components.