The Applications Of Chemometric Multivariate Resolution Methods In The Analysis Of Complex Multi-component Systems

There are five parts in this article. The applications of chemometric multivariate resolution and its development are reviewed in recent years. These methods are mainly some two-way data and three-way data resolution methods. Moreover, these methods are applied in the analysis of some complex multi-component systems which can not be analyzed by some routine means qualitatively and quantitatively because physical separation of these systems can not completely accomplished and overlapped peaks of multi-component are formed.In the first chapter, the development and applications of multivariate resolution are summarized and the studies of some two-way data and three-way data resolution methods are introduced.In the second chapter, the principles of some algorithms based on matrix and array resolution in multivariate resolution are introduced in detail.In the third chapter, the performance of combination of three two-way data resolution methods (heuristic evolving latent projections [HELP], sub-window factor analysis [SFA] and orthogonal projection resolution [OPR]) with two rank map estimation methods (evolving factor analysis [EFA] and fixed size moving window-evolving factor analysis [FSMW-EFA]) is investigated about the identification and quantification of overlapping peaks in capillary electrophoresis-diode array detector (CE-DAD). All resolution methods are applied to the simulated and experimental data respectively. The resolved results are compared qualitatively and quantitatively. In addition, the further comparison are carried out with the results obtained by multivariate curve resolution- alternate least square (MCR-ALS) using the initial estimates provided by EFA.In the forth chapter, three three-way data resolution methods (Tucker 3, Parallel factor analysis [PARAFAC] and alternative trilinear decomposition [ATLD]) are applied in the analysis of overlapped peaks of three kinds of amino acid in fluorescence. The results indicate that three methods can be used to the qualitative and quantitative analysis of multi-component overlapped peaks of three-way fluorescence spectra andhave comprehensive theoretic and practical significance.In the fifth chapter, partial least squares (PLS) and artificial neural network (ANN) are applied to the quantitative analysis of multi-component overlapped peaks in fluorescence. The results indicate that back-propagation of error (BP) ANN which uses genetic algorithm (GA) to choose input variables can predict the concentration of each component of overlapped peaks more exactly than PLS and it provides a good routine method for the analysis of this kind of sample.