New Chemometric Algorithms And Their Application Studies For Mutiway Spectroscopy Data

With the rapid development of modern analytical technology, many sophisticated instruments such as EEM, HPLC-DAD, GC-IR have come out, and the automatization of data collection and transmission have been realized. Analysts can now obtain two-dimensional, three-dimensional and even four-dimensional data arrays, which are composed of hundreds and thousands data points, easily and quickly. These data arrays not only contain a mass of chemical information, but also involve some interferers including irrelevance chemical copmponents, noise and background. However, it is difficult, if not impossible, to extract meaningful chemical information from the multi-dimensional data arrays just in virtue of conventional analytical techniques. Analysts have to resort to chemometrics, the art of extracting interested information from chemical data consisting of vast data points by the combination of mathematics, statistics and computer science.Nowadays,in the field of chemometrics methodologies, studying multi-way data analysis in analytical chemistry is one of the most active areas with practical significance. It can offer promising tools for the analysis of complex chemical systems, which are hard to be analyzed by conventional analytical means. The research work in this thesis focuses on the methodologies and their applications for three-way data analysis and two-way data analysis of mutiway spectroscopy.Asymmetric alternating least squares decomposition (AALSD) algorithm with partial bilinearization is developed for three-way data arrays analysis. The new proposed algorithm is based on the least squares principle and an asymmetric way with partial bilinearization to calculate the three underlying profile matrices in the resolution of the trilinear model. The results obtained by simulated data and real excitation–emission spectral data sets have shown that AALSD method retains the second-order advantage of quantification for analyte(s) of interest even in the presence of potentially unknown interferents. Another salient virtue of AALSD is insensitive to the estimated component number, which can make the need for exact determination of the actual component number omitted in the resolution of three-way data arrays. Comparing with parallel factor analysis-alternating least squares (PARAFAC-ALS), the developed AALSD method can supply acceptable results in most cases and has a relative higher convergence rate. Three-way calibration methodologies coupled with the excitation–emission fluorescence (EEM) are applied to direct and quantitative determination of reserpine in human urine and doxazosin in human plasma. With the property of"mathematical separation"to displace"chemical separation", the alternating trilinear decomposition (ATLD), the self-weighted alternating trilinear decomposition (SWATLD) and the alternating fitting residue method (AFR) are employed to resolve the data of reserpine and doxazosin, respectively. These chemometric methodologies have the second-order advantage, which is the ability to get accurate concentration estimates of interested analyte(s) even in the presence of uncalibrated interfering components. The results have indicated that the combination of EEM detection coupled with second-order calibration algorithms is a powerful tool to quantify the analytes of interest from overlapped fluorescence spectra for complex analysis of drugs in urine or plasma.An improved method based on an ensemble of Monte Carlo uninformative variable elimination (EMCUVE) has been presented for wavelength selection in multivariate calibration of spectral data analysis. The proposed algorithm introduces Monte Carlo (MC) strategy to uninformative variable elimination-partial least square (UVE-PLS) instead of leave-one-out strategy for estimating the contributions of each wavelength variable in the PLS model. In EMCUVE, wavelength variables are evaluated by different Monte Carlo uninformative variable elimination (MCUVE) models. Moreover, a fusion of MCUVE and the vote rule can obtain an improvement over the original uninformative variable elimination method. Results obtained from simulated data and real data sets demonstrate that EMCUVE can properly carry out wavelength selection in the course of data analysis and improve predictive ability for multivariate calibration model.Monte Carlo strategy is used to implement a wavelength selection procedure for the multivariate calibration model based on partial least squares regression, named Monte Carlo partial least squares (MCPLS). Firstly, the new method allows wavelength variables to be separated into small sets, then and constructs different PLS sub-models by themselves to be optimized along with the selection of the wavelengths. The proposed procedures are investigated by one simulate data and two near-infrared data sets. The results have revealed that the proposed method is promising for multivariable spectroscopy analysis and gives much better prediction than the full-spectrum PLS model.A homemade apparatus has been designed to monitor the process of gas-solid adsorption non-invasively with near-infrared spectroscopy (NIR). With the apparatus, it is convenient to study adsorption equilibrium in-line in the laboratory. A series of local PLS alibration models are suggested to replace a normal global PLS model because the concentration of the adsorbate in the adsorbent varies greatly during the period of an adsorption process. Moreover, as an example, three isotherms of gas-solid adsorption have been obtained with the apparatus and multivariate calibration model.A novel experimental methodology is proposed to study on the adsorption kinetics of orthoxylene on silica gel with near-infrared spectroscopy. There is a differential adsorption bed (DAB) where the solid adsorbent always contacts with the same bulk concentration of the adsorbate vapor, and the DAB is monitored with near-infrared diffuse reflectance spectroscopy (NIRDRS) continuously. In the same way, we have obtained instantaneous adsorption rates of several orthoxylene/silica gel adsorption processes under different conditions like partial pressure of orthoxylene vapor and velocity of gas, and discovered that the adsorption process is physical adsorption, and mainly controlled by external diffusion.