【作者】 莫翠云；

【导师】 吴海龙；

【作者基本信息】 湖南大学 ， 分析化学， 2002， 硕士

【摘要】 本论文的绪论部分概述了化学计量学多元分辨与校正这一分支的发展情况以及研究现状，重点描述了三阶数据分析即立方阵数据分析的原理和方法，在此基础上，将立方阵数据分析方法应用于多样品复杂多组分荧光数据的解析，成功地解决了稀土、药物等样品分析中混合物光谱重叠严重所引起的分离和分析极为困难的问题。此外，还提出了一个改进的立方阵数据分解的算法，用以解决传统平行因子分析法(PARAFAC)的收敛缓慢和对组分数估计敏感等问题。第二章利用立方阵数据分析中交替三线性分解算法(ATLD)与三维荧光分析法相结合，对三种广泛用于治疗心血管疾病的药物--阿米洛利、心得安和潘生丁进行同时测定。测定结果表明尽管这三种药物荧光光谱严重重叠，但通过运用立方阵数据分析方法，以“数学分离”代替“化学分离”，可在不增加经济和时间耗费的大前提下，可在背景干扰的共存下有效地对其进行同时快速定量测定。氨苯喋啶、呋塞米和阿米洛利是三种常用的利尿药，由于其用于比赛可以提高运动员成绩，同时又能影响某些其他药品的检测，故被国际奥委会定为利尿类违禁药品。因此，对其进行同时测定具有极其重要的现实意义。本文第三章利用ATLD算法与三维荧光分析法相结合，对该体系进行了研究，结果表明，化学计量学方法与荧光分析法相结合，可用于氨苯喋啶、呋塞米和阿米洛利的同时定量测定，是不经分离且快速有效解决荧光分析体系中背景干扰共存下多组分同时快速定量分析问题的较佳途径。第四章利用化学计量学中的立体阵数据分解方法--平行因子分析法(PARAFAC)与荧光分析法相结合，对多组分稀土元素进行了荧光定量同时测定。利用其二阶校正优势，以“数学分离”代替传统的化学分离，无需对复杂混合体系中的干扰物或其他待测组分进行化学分离或采用其它掩蔽技术，在钐共存条件下，对混合物中感兴趣的组分铽和镝进行了同时荧光测定，结果令人满意。第五章利用ATLD算法与荧光激发发射矩阵相结合对吡多醛、吡多胺和吡多酸混合物体系进行了研究。吡多醛、吡多胺和吡多酸与维生素B6的生理功能及体内的多种代谢有着密切的关系，它们可共存于人体内，其荧光光谱严重重叠，用常规方法难以同时测定。本章将基于立方阵分解的交替三线性分解算 湖南大学硕土学位论文法（ATLD）与荧光激发发射矩阵分析法相结合，对水溶液中共存的毗多醛（PL）。毗多胺（PM ）和4－毗多酸（4PA）进行了同时分辨和测定。研究结果表明了ATLD算法亦能有效地解决这一混合物体系在干扰物共存下的同时分辨和定量测定问题。 第六章提出了一个用于立方阵数据分辨的同时最小化协变矩阵误差算法 （SMCME人 该算法不仅能克服传统 PARAFAC算法对组分数估计的敏感性问题，而且可有效地解决药物分析中的因光谱重叠严重而引起的误差放大或无法得到正确解的问题。更多还原

【Abstract】 In the introduction section of this thesis, the development of multivariateresolution and calibration in chemometrics, especially three-way data analysis, hasbeen reviewed. Applications in the analysis of rare earth and drugs by fluorometrycombining with the three-way data resolution and calibration method have beenresearched. The problem of serious spectra overlapping has been successfully solved.Furthermore, an improved algorithm which can overcome the problems of slowconvergence and being sensitive to the estimation of component number of thetraditional PARAFAC algorithm has been proposed.In chapter 2, the system of Amiloride(AMI), propranolol(PRO) anddipyridamole(DIP) are widely used in the treatment of angiocardiopathy. It wouldbe difficult to be determined simultaneously by routine method since theirfluorescence spectra overlapped closely. AMI, PRO and DIP in aqueous solution areresolved and determined simultaneously using alternating trilinear decompositionalgorithm and three-dimensional fluorescence spectrometry. The satisfactory resultsshowed that the ATLD algorithm can solve the problem of closely spectraoverlapping at ease without resorting to expensive techniques and time-consumingprocedure.Triamterence (TAT), furosemide (FUR) and amiloride (AMI) are three staplediuretic drugs, widely used to treat several diseases. These components are classifiedin doping substances and have been included in the list of forbidden substances bythe International Olympic Committee. Simultaneous determination of the three drugsis of great importance. In chapter 3, the system composed of the three drugs has beenresolved using ATLD algorithm. The results showed that the ATLD algorithm cansolve the problem of serious spectral overlapping rapidly and effectively.In chapter 4, simultaneous determination of terbium (Tb) and dysprosium (Dy) inthe presence of interference of samarium (Sm) was successfully performed using theclassical three-way data array decomposition method -- the PARAFAC algorithm.Due to "second-order calibriation advantage", this algorithm can successfullyimplement direct calibration of the components of interest without resorting to time-consuining and expensive separation procedure.In chapter 5, the system of pyridoxal (PL), pyridoxamine (PM) and 4-pyridoxicacid (4-PA) are three drugs which have closely relationship with physiologicalfunction of vitamine B6 and several metabolities in human bodies. They aresimultaneously determined using the ATLD algorithm. The satisfactory resultsfurther demonstrated the ATLD algorithm can solve the closely spectra overlappingproblem at ease and be insensitive to the estimation of component number.In chapter 6, an improved algorithm, simultaneously minimizing covariant matrixerror algorithm fOr three-way data analysis, was proposed. This algorithm canovercome the problems of slow convergence and being sensitive to the estimation ofcomponent number, which are inevitable while using the traditional PARAFACalgorithm.更多还原