Application Of Tchebichef Moment Method In The Analysis Of Complex Systems And Research Of Related Extentions

The rapid development of modern analytical instruments and the realization of automatic acquisition for the instrument data enable us to obtain massive data in a relatively short time.Therefore,how to acquire as much useful information as possible from a large amount of data has become an urgent problem that need to be solved by researchers.Chemometrics is a useful method that were based on the mathematical,statistical and computer techniques.It can extract as much useful informations as possible from chemical data by analyzing and processing complex measurement data.There are two main research contents of chemometrics,one is the development of algorithm for processing chemical data and the other is the application of new theories and chemometrics methods in practical problems.The works in this dissertation are focused on these two issues mentioned above.One part is the application of Tchebichef image moment method in complex system?chapters 2 to 4?,and the other one is the expansion and application of new moment methods?chapters 5 and 6?.The main contents of each chapter are shown as follows:Chapter 1 IntroductionThis part mainly summarizes the development of chemometrics and its application in complex system analysis,the development of moment methods and their applications in conventional chemical spectra analysis.Then the research works of this dissertation were introduced briefly.Chapter 2 An effective approach to quantitative analysis of ternary amino acids in foxtail millet substrate based on Terahertz Spectroscopy with Tchebichef image moment methodAs a novel detection technology,Terahertz time-domain spectroscopy?THz-TDS?has been applied to many fields.However,it still encounters drawbacks in multicomponent mixtures analysis due to serious spectral overlapping.In this chapter,an effective approach to quantitative analysis was proposed,and applied it on the determination of the ternary amino acids?glutamic acid,glutamine and tyrosine?in foxtail millet substrate to simulate the real food environment.Utilizing three parameters derived from the THz-TDS,the data of three properties are constructed into two-dimensional matrix and the Tchebichef image moments were used to extract the information of target components.Then the quantitative models were obtained by stepwise regression.The correlation coefficients of leave-one-out cross-validation(R²_loo-cv)were more than 0.9595 for the three obtained models.As for external test set,the predictive correlation coefficients?R_p²?were more than 0.8026 and the root mean square error of prediction?RMSE_p?were less than 1.2601.Compared with the traditional methods PLS?partial least squares?and N-PLS?N-Way partial least squares?methods,our approach is more accurate,robust and reliable,and can be a potential excellent approach to quantify multicomponent with THz-TDS spectroscopy.Chapter 3 An effective approach to the rapid quantitative analysis of three pesticides in two food substrates based on HPLC-DAD with Tchebichef image momentsDiverse interferences exist in different substrates which often affect the determination of pesticide residues in various kinds of food,and complex pretreatments to eliminate interference information before analysis are usually necessary.We proposed an approach for rapidly quantitative analysis of multi-target components in different substrates,and applied it to the determination of ternary pesticides?thiabendazole,thiophanate-methyl and 2-aminobenzimidazole?in cherry tomatoes and red grape samples.By means of HPLC-DAD,the three-dimensional chromatographic data were obtained under isocratic elution.Utilizing the Tchebichef image moments,unified quantitative models were established for the three target components of two sample substrates,respectively.The correlation coefficients of leave-one-out cross-validation(R²_loo-cv)of the obtained models were more than 0.9975.The predictive correlation coefficients?R_p²?were more than 0.9831.Compared with the N-PLS and MCR-ALS?Multivariate curve resolution alternating least square?methods,the proposed approach is not only more effective in modeling,more accurate and reliable in compositional analysis,but also can eliminate the interference of different sample substrates and extract the most essential information of the target components.And this work shows that the Tchebichef image moment method is expected to be a potential tool to rapidly quantify multi-component in different sample substrates without complex pretreatments.Chapter 4 Rapid quantitative analysis of the eight active components in one Chinese patent medicine based on HPLC-DAD with Tchebichef image momentsDifferent analytical techniques,analytical conditions and complicated pretreatments are often necessary to quantify the alkaloids,statins,and glycosides in Chinese patent medicine?CPM?,which are not only time consuming and test cost increasing,but also easy to cause environmental pollution with chemical reagents.In this chapter,the Tchebichef image moment is used for the rapid and effective determination of four alkaloids,one statin and three glycosides in the CPM samples simultaneously.Based on the 3D spectra of samples obtained from HPLC-DAD,the Tchebichef image moments?TMs?were calculated and employed to establish the quantitative linear models of the eight target active components in the CPM samples.The correlation coefficients of leave-one-out cross-validation(R²_loo-cv)of the obtained models were more than 0.9545,and the predictive correlation coefficients?R_p²?were more than 0.9159;the recoveries were 82.8%to 116.4%.The obtained results indicated that the proposed approach is more accurate and reliable compared with the two popular chemometrics methods of N-PLS and MCR-ALS.We also expected our approach to be a potential tool to quantify multi-components rapidly in other complex compound preparations.Chapter 5 The extensions based on the method of second order Tchebichef image momentAnalytical chemistry often requires the use of chemical instruments to convert chemical signals into image information,and analyze.Various analytical instruments can obtain not only two-dimensional matrix data of a single sample?HPLC,EEM,LC-MS,etc.?,but also one-dimensional matrix data of a single sample data?HNMR,NIR,CD,UV-Vis,etc.?and three-dimensional matrix data?EEM-dynamics,EEM-pH,multiple instruments combined,etc.?.Therefore,it is necessary to develop the first-order and third-order data processing methods.In this chapter,the formula of the Tchebichef image moment is mathematically derived and extended to the Tchebichef first-order curve moment and Tchebichef third-order cubic moment.And then the global feature extraction ability of moment method and the invariance of moment are verified by image reconstruction.Chapter 6 The initial application attempt of the Tchebichef third-order cubic momentWith the rapid development of modern analytical instruments and the complexity of research objects,more and more four-dimensional data can be obtained.In this chapter,the excitation emission-time data and the data of multiple instruments were selected,and these two groups of data which represents different ways to acquire the four dimensional data,were applied to study the feature information extraction ability of Tchebichef cubic moment respectively.The correlation coefficient(_cR²,R²_{ad j}andR²_cv)of linear model established by two components of excited emission fluorescence data were better than 0.9937 and the root-mean-square error(RMSE_c and RMSE_cv)were less than 0.0250.The predictive correlation coefficients of the multiple instruments were all higher than 0.9846 and the prediction of root mean square error were less than 0.2267.The modeling results were also compared with those in the literature,which showed that our method was more convenient,faster and more accurate.This study effectively expanded the application of Tchebichef moment method in complex systems.