Multicomponent Analysis with Chemometrics Methods

近年来，随着生物学、生命科学、药物学、环境科学等学科的迅猛发展，使得人们对于分析化学的要求也越来越高。由于这些学科一般都涉及对复杂多组分体系的定性、定量及结构分析，传统的单组分分析手段已经无法满足需要，这就迫切需要新的分析理论和技术来解决日益复杂的分析问题。而化学计量学的出现和迅速发展，为广大化学工作者提供了强有力的理论工具和分析手段，使得他们能够根据具体问题的不同而采用适当的分析技术来解决传统方法无法解决的难题。论文工作围绕因子分析类方法和小波分析方法在多组分分析中的应用进行了较系统的研究，主要内容归纳如下：1）应用渐进因子分析法和固定尺寸移动窗口渐进因子分析法对模拟的重叠HPLC-DAD数据进行解析，并对两种方法的解析结果进行比较。2）应用子窗口因子分析法解析模拟的重叠HPLC-DAD数据。3）应用小波分析方法进行数据压缩、平滑和滤噪、背景扣除和基线校正、重叠信号的解析，得到了满意的结果。4）将二维小波变换与子窗口因子分析方法相结合，用于高噪声的模拟HPLC-DAD数据的解析。5）将启发渐进式特征投影法和局部正交投影法分别应用于丹参和川芍提取物的HPLC-DAD重叠峰解析，所得结果可以作为进一步定性定量分析的依据。With the rapid development of biology, life sciences, pharmacology, environment sciences in recent years,, the requirements for analytical chemistry have been increasing, and the traditional single-component analytical methods can not be suitable for multi-component qualitative, quantitative and structure analysis which are basically needed for these disciplines. As a consequence, new analytical theory and technology are in urgent need to be developed for solving these increasingly complicated problems. Fortunately, the newly emerged and rapidly developing chemometrics methods provide powerful theoretical tools and analytical means, which come up with tailored solution for the problems. This dissertation is concentrated on the applications of factor analysis and wavelet transform for multicomponent analysis. The main contents are as follows: 1)Application of evolving factor analysis and fixed size move window evolving factor analysis in the resolution of simulated overlapping HPLC-DAD data, and the comparison of results obtained by the two methods. 2)Application of subwindow factor analysis in resolution of simulated overlapping HPLC-DAD data. 3)Application of wavelet analysis in data compression, smoothing and denoising, baseline and background correction, resolution of multicomponent overlapping signals. 4)Combination application of two-way wavelet transform and subwindow factor analysis in resolution of high noise simulated HPLC-DAD data. 5) Heuristic evolving latent projection and local orthogonal projection were applied to resolve overlapping HPLC-DAD peaks of salvia miltiorrhiza bunge and liguisticum chuanxiong Hort. respectively. The result can be as the basis of further qualitative and quantitative analysis