Xueguang Shao

A new regression method based on independent component analysis

Based on independent component analysis (ICA), a new regression method, independent component regression (ICR), was developed to build the model of NIR spectra and the routine components of plant samples. It is found that ICR and principal component regression (PCR) are completely equivalent when they are applied in quantitative prediction. However, independent components (ICs) can give more chemical explanation than principal components (PCs) because independence is a high-order statistic that is a much stronger condition than orthogonality. Three ICs are obtained by ICA from the NIR spectra of plant samples; it is found that they are strongly correlated to the NIR spectra of water, hydrocarbons and organonitrogen compounds, respectively. Therefore, ICA may be a promising tool to retrieve both quantitative and qualitative information from complex chemical data sets.

Wavelet transform and its applications in high performance liquid chromatography (HPLC) analysis

Abstract Wavelet transform has been proven to be a high performance signal processing technique. In this paper, a novel algorithm which is more suitable to process analytical signals is proposed, and application of the algorithm in denoising, baseline correction, determination of component number in overlapping chromatograms, and preprocessing the data matrix for window factor analysis (WFA) in resolving multi-component overlapping chromatograms is reported. The main characteristic of the wavelet transform is that it decomposes a signal into localized contributions, and each of the contributions represents the information of different frequency contained in the original signal. Therefore, the noise can be filtered by separating the high frequency contributions from the chromatogram, the baseline correction can be obtained by removing the low frequency contributions from the chromatogram, and the resolved chromatogram can be retrieved from the overlapping signal by choosing a discrete detail at certain scales from the decomposed contributions. The determination of component number in overlapping chromatograms can be obtained by simply counting the number of peaks in the resolved chromatogram.

Optimization of Lennard-Jones atomic clusters

A fast annealing evolutionary algorithm was applied to determine the minimum configurations of Lennard-Jones (LJ) clusters. Other techniques such as limited memory quasi-Newton algorithm (L-BFGS), seeding, similarity checking, and moving, were also used in this method. It has been shown that these techniques dramatically speed up the evolutionary procedure. The lowest known energies were located for the LJ clusters containing up to 74 atoms. It has been proven that the algorithm is a fast and high efficient optimizing tool, which can be used in energy minimization problems.

Determination of the component number in overlapping multicomponent chromatogram using wavelet transform

Abstract Component number in overlapping multicomponent chromatogram was determined by a novel method—wavelet transform. Because of the characteristic of the double localization in time and frequency domain, the wavelet transform can decompose an overlapping chromatogram into contributions of different frequency. Among these contributions, there will be contributions which will represent the resolved chromatographic signals because their frequency is higher than the overlapping signal and lower than the high frequency noise. Therefore, the component number of an overlapping chromatogram can be determined by the number of peaks in the resolved chromatogram. Simulated data sets and a seriously overlapping 5-component chromatogram were investigated by the method. It was proved that the wavelet transform is a very easy and convenient method for detecting the component number in overlapping multicomponent chromatograms.

A new stochastic resonance algorithm to improve the detection limits for trace analysis

Based on the stochastic resonance theory, a new stochastic resonance algorithm (SRA) to improve analytical detection limits for trace analysis is presented. In the new algorithm, stochastic resonance takes place in a bistable system driven only by the inherent noise of an analytical signal. The effect of the system parameters on the proposed algorithm is discussed and the optimization of parameters is studied. By using experimental chromatographic and spectroscopic data sets, it is proven that the signal-to-noise ratio (SNR) of the analytical signal can be greatly enhanced by the method, and an excellent quantitative relationship between different concentrations and their responses can be obtained. Stochastic resonance may be a promising tool to extend instrumental linear range and to improve the accuracy of micro- or trace analysis.

A method based on stochastic resonance for the detection of weak analytical signal.

An effective method for detection of weak analytical signals with strong noise background is proposed based on the theory of stochastic resonance (SR). Compared with the conventional SR-based algorithms, the proposed algorithm is simplified by changing only one parameter to realize the weak signal detection. Simulation studies revealed that the method performs well in detection of analytical signals in very high level of noise background and is suitable for detecting signals with the different noise level by changing the parameter. Applications of the method to experimental weak signals of X-ray diffraction and Raman spectrum are also investigated. It is found that reliable results can be obtained.

A Partial Least Squares‐Based Consensus Regression Method for the Analysis of Near‐Infrared Complex Spectral Data of Plant Samples

Abstract A consensus regression approach based on partial least square (PLS) regression, named as cPLS, for calibrating the NIR data was investigated. In this approach, multiple independent PLS models were developed and integrated into a single consensus model. The utility and merits of the cPLS method were demonstrated by comparing its results with those from a regular PLS method in predicting moisture, oil, protein, and starch contents of corn samples using the NIR spectral data. It was found that cPLS was superior to regular PLS with respect to prediction accuracy and robustness.

An efficient approach for theoretical study on the low-energy isomers of large fullerenes C90-C140.

An approach that consists of a molecular mechanics method based on the second generation reactive empirical bond order (REBO) potential and the more accurate semiempirical method PM3 (Parametric Method No. 3) was proposed to predict the energetically favored isomers of the fullerenes from C90 to C140 at the semiempirical level. All the 578,701 isolated-pentagon-rule isomers of fullerenes from C90 to C140 were enumerated from topological structures and systematically searched using an energy minimization method to select the best 100 low-energy isomers based on the REBO potential for each fullerene. Then these candidate isomers were further optimized by PM3 and ranked again to determine the top low-energy isomers. This approach was applied to calculate the energetically favored isomers of C90-C140. The results of C90-C120 are in good agreement with the published results by quantum-chemical methods. Furthermore, the top five low-energy isomers of C90-C120, as well as C122-C140 which have scarcely been systematically studied before, are also predicted with the approach. The analysis of the structures showed that the hexagon-neighbor rule is an important factor to the stability of C90-C140. The time cost for the systematical search based on the REBO potential was also discussed. It indicates that the approach proposed is efficient for predicting the energetically favored isomers of large fullerenes at the semiempirical level.

Molecular interactions of α-cyclodextrin inclusion complexes using a genetic algorithm

Abstract Molecular interactions of inclusion complexes of mono- or 1,4-disubstituted benzenes and α-cyclodextrin have been studied in this paper. Two types of energy terms were considered in the total interaction energies, non-bonded term and desolvation term, and minimized by a genetic algorithm (GA). Using the consistent force field (CFF91), the non-bonded energies between all pairs of atoms in different molecules were determined by a Coulomb potential term for electrostatic interactions and a Lennard-Jones potential for van der Waals interactions. The desolvation energy term was modeled by a simple constant term corresponding to a penalty when polar atoms are placed in the hydrophobic cavity. The total interaction energies for 15 inclusion complexes with experimental association constants (lnxa0 K ) were optimized by the GA method. Linear regression analysis of the observed association constants against the total energies was performed. It was found that the interaction energies of these complexes obtained by the simple interaction energy model could be correlated with their experimental association constants, and also the desolvation term should be included.

AN APPLICATION OF THE CONTINUOUS WAVELET TRANSFORM TO RESOLUTION OF MULTICOMPONENT OVERLAPPING ANALYTICAL SIGNALS

The continuous wavelet transform (CWT) is applied successfully to the resolution of overlapping chromatograms. CWT tends to reinforce the traits of a signal and makes all information more visible due to its redundancy. This is especially true of very subtle information. It is proven by the results that the method is superior to the discrete wavelet transform (DWT) and wavelet packet transform (WPT). CWT has much stronger ability to extract subtle information from seriously overlapping chromatograms, and can process more conveniently than WPT. Furthermore, the linearity of chromatographic signal remains in the resolved results, which ensure the results can be used for quantitative determination. The resolving ability of different wavelet basis and for signal with different noise levels are also discussed.