目标识别因子分析解析完全未知混合体系的研究

本文以目标识别因子分析为基础, 将因子分析与聚类分析相结合, 给出了从完全未知混合体系中提取纯物种光谱的新方法。所得纯物种光谱经光谱检索作定性判别, 然后利用外标法对混合物中的各组份进行标定。该方法用于计算机模拟体系及三组份实际混合体系的红外光谱解析, 结果令人满意。

Studies on target recognition facto r analysis resolving unknown multicomponent mixture

In this paper, target recognition factor analysis (TRFA) is applied to estimating the component spectras from unknown multicomponent mixture system. With the use of pattern recognition techniques, TRFA derived target vector from the original data. A set of procedures which inlude extracting pure component spectra, spectra searching and standardizing are developed. This method made it possible to complete qualitative and quantitative analysis of an unknown mixture system in the meantime. The computer simulation technique is first used to verify the algorithm, and then the practical data collected from FTIR are also analyzed by this method. Satisfactory results are obtained.