INATION TO USERS分析化学双语论文.docx

INFORMATION TO USERS 分析化学论文This manuscript has been reproduced from the microfilm master. UMIfilms the text directly from the original or copy submitted. Thus, somethesis and dissertation copies are in加ewriter face, while others maybe from any加e of computer printer.The quality of this reproduction is dependent upon the quality of thecopy submitted. Broken or indistinct print, colored or poor qualityillustrations and photographs, print bleedthrough, substandardand improper alignment can adversely affect reproduction.In the unlikely event that the author did not send UMI a completemanuscript and there are missing pages, these will be noted. Also, ifunauthorized copyright material had to be removed, a note will indicatethe deletion.Oversize materials (e.g., maps, drawings, charts) are reproduced bysectioning the original, beginning at the upper left-hand corner andcontinuing from left to right in equal sections with small overlaps. Eachoriginal is also photographed in one exposure and is included inreduced form at the back of the book.Photographs included in the original manuscript have been reproducedxerographically in this copy. Higher quality 6 x 9 black and whitephotographic prints are available for any photographs or illustrationsappearing in this copy for an additional charge. Contact UMI directlyto order. University Microfilms International A Bell&Howell Information Company300 North Zeeb Road, Ann Arbor, MI 48106-1346 USA 3131761-4700 800/521-0600ABSTRACT Analytical techniques often require measuring transient chemical species.Here, transient chemical species are considered to arise from changes in concentra-tion over time. Chemical kinetics, chromatography, and flow injection analysis, allmake use of measurements of transient chemical species,妙measuring chemicalmixtures of rising or falling chemical concentration, as chemical reactions pro-ceed or as the chemical species pass through a flow-through detector. The studiesreported here show how chemical estimation is performed with a Kalman filteron multicomponent transient chemical species. A new approach to Kalman filterarchitecture, referred to as hierarchical Kalman filtering, makes use of the statespace formulation, as it is ap皿ed to problems in analytical chemistry. Two kindsof Kalman filter architecture, described as master-slave and multiple-peer filters,are used to do multicomponent analysis of chromatographic, flow-injection andchemical kinetic data. The data analysis performed in the studies was applied tohigh information content diode-array spectrophotometric data. Information-based application of the Kalman filter algorithm was a centraltheme of the work. In the first study, the information matrix determinant wasevaluated to determine information content. By using these results, the extendedKalman filter was applied only to high information content data, resulting inreduced computational burden. In another study, new information arising fromthe evolving chemical estimates, while analyzing chromatographic data, was usedin a dynamic filter model. This resulted in improved filter estimates by eliminatinginduced model errors. Another application to chromatographic data addressed aXiiicoeluting chromatographic pair and used a Kalman filter model that could beadjusted for subtle variations from baseline drift. Finally, flow injection datawere analyzed by using a calculation of matrix condition number. In this way,system observability was evaluated for determining the choice of wavelengths ofthe Kalman filter models. The calculation burden was reduced by eliminating lowinformation wavelengths from the filter model. In each study, filtering efficiencywas addressed by considering information location and content or by incorporatingnew information that becomes available after the data analysis was initiated.xivChapter 1INTRODUCTION Scientists and engineers have long studied the spatial and temporal charac-teristics of chemical species in various media. The time-dependent functionality ofchemical species is of common interest to these groups. Time dependency occurs asthe result of transit through conduits or as the result of changes in concentrationsfrom chemical reactions. Workers in these groups have developed tools designedto accomplish the objectives of the separate disciplines. This study merges toolsfrom engineering with tools of analytical chemistry. The result is a unique way toto perform chemical an由ses. Information theory establishes a metric for information by describing infor-mation content in a quantitative way. Chemical applications of information theoryseek to maximize the efficiency of the practice of analytical chemistry by directingattention to methods which provide high information content, or场directing at-tention to particular regions of data having higher information. A technique thatproduces greater quantities of information expedites decision-making processes(governing recalibration or the number of repeat an吻ses), potentially improvinganalysis efficiency. The efficiency of analytical techniques may also improve byconsidering the location of data-information. Portions of data which contain rmation and redundant data can be eliminated from the scrutiny of the dataanalysis algorithm, thus speeding the analysis of the data. In this study, both the amount and location of information in chemicaldata are studied. By using multisensor detection, hi沙information data can be12collected. After data acquisition, a novel hierarchical, multivariate digital filter isapplied to the data to do chemical estimation. The more common analytical chemistry techniques often rely on zeroth-order sensing measurement devices, each measurement providing a single numer-ical value that is related to a quantity such as concentration. An example ofthis is a chromatographic analysis with single-wavelength detection. This experi-ment results in a series of single point measurements, which are taken over time.However, with the advent of powerful computers in the laboratory and the use ofrapid-scanning instruments and diode-array based instrumentation, it is possibleto reconsider how measurements can be made of such events. In the examplementioned, chromatographic an吻sis can now be performed using a higher or-der analytical device such as a diode array instrument. As a result, the quantityof information available per measurement is increased greatly. The informationavailable from such instrumentation is also great enough to allow for performingmulticomponent analysis of complex mixtures衍using mathematical data analy-sis algorithms. In this study, diode一二ay, ultraviolet-visible spectroscopy is usedto an由ze chemical data. The relatively higher information content available fromthese measurements is then used to analyze chemical mixtures. When using multivariate data to perform multicomponent analysis, the in-formation content of each measurement vector (for example, an ultraviolet-visibleabsorption spectrum) must be considered. In many cases, parts of a measurementvector may be eliminated since they do not contribute to decision making processesor do not enhance the ability to perform quantitative analysis. In the followingchapters, a hierarchical architecture is implemented that couples a digital Kalmanfilter to calculations that evaluate information content. This allows for flexible ap-plication of the chemical estimation algorithm by adapting to information contentin the data. This approach addresses the issue mentioned earlier, that the data3analysis algorithm be applied to regions in the data consisting of high informationdensity, to ensure that chemical estimation is accomplished efficiently and quickly. Until recent玩an由tical techniques have relied on adequate prior separa-tion of mixtures. Analyses which rely on a physical or chemical process to separatemixture responses are hindered when mixed detector responses occur. Methodswhich incorporate mathematical multicomponent resolution are more robust to-ward the problem, allowing for a mathematical solution when prior separationfails. The methods used in the past that incorporate zeroth order sensing devicescan often miss possible sources of information. Using multivariate mathematicalmodels, other dimensions of chemical information can be incorporated in the ana-lytical technique. Each dimension contributes to the selectivity and specificity ofthe technique. One experiment which has held great promise for development ofnovel analytical techniques is flow injection analysis. It produces transient chem-ical species in a flowing carrier medium. Flow injection data is information-rich,incorporating information from chemical kinetic, diffusion and dispersion-basedphenomena. To investigate modeling of these transient chemical species, the prob-lem was broken into its component parts for study. These studies comprise thesubsequent chapters presented here.Chapter 2KALMAN FILTER MODELING FOR STUDY OFTRANSIENT CHEMICAL SPECIES IN FLOWINGMEDIA In analytical chemistry, the success of chemical and physical separations,applied prior to performing the measurement, affects the ability to obtain highquality quantitative or qualitative chemical estimates. Recently, because of theappearance of powerful computers in the an吻tical chemistry laboratory, a newway to achieve these separations has emerged: the separation of chemical responsesthrough mathematical multicomponent analysis. In contrast to performing chem-ical separations prior to the measurement, mathematical resolution of mixtureresponses is done after the measurement. The power of thisapproach canbe measured by the economic impact of applying a technique that requires lessexpenditure of reagents and time, which come as a result of increased analysisthroughput. Another way to use the power of mathematical-an吻sis separationsis by considering that in situ analyses are now more feasible. This is becausethe analyte can be measured directly within the original sample environment. Byeliminating any physical or chemical separation before making a measurement,variability in the chemical estimates is subdued. The integrity of the analytespecies can be better preserved by avoiding steps which might lead to analytedegradation or losses due to sample handling.4汉语：信息给用户这个手稿已复制缩微胶片主。 UMI电影直接提交的原件或复制的文本。因此，一些论文和毕业论文的副本加ewriter面对，而另一些可能从任何计算机打印机加。这种复制的质量取决于质量副本提交。破损或模糊不清的打印，彩色或质量差插图和照片，打印bleedthrough，不合和对齐方式不当，可以产生不利影响繁殖。万一，作者没有派一个完整的UMI手稿和有缺页，这些会注意到。此外，如果未经授权的版权材料已被删除，将显示的说明删除。特大型的材料（如地图，图纸，图表）转载切片原来，在上层的左上角开始，在与小重叠平等的部分继续由左到右。每个原来也拍到在一次曝光，包括在减少在书背的形式。原稿中的照片已转载xerographically在这个副本。更高质量的6“x 9”的黑色和白色可用于任何照片或插图摄影照片在这个副本中出现的额外收费。直接联系UMI秩序。国际大学缩微胶卷贝尔和霍威尔信息公司北Zeeb 300路，安阿伯，MI 48106-1346美国3131761-4700 800/521-0600摘要分析技术，往往需要测量瞬态化学物种。在这里，被认为是暂时性的化学物质的变化浓度出现TION以上时间。化学动力学，色谱仪，流动注射分析，所有利用测量瞬态化学物种，妙测量化学化学物的浓度上升或下降的混合物，作为化学反应的亲CEED或化学物种通过流过探测器通过。该研究此报告显示化学的估计是如何与卡尔曼滤波器进行多元瞬态化学物种。卡尔曼滤波的一种新方法架构，使得使用的状态，称为分层卡尔曼滤波空间的制定，因为它是鸭皿在分析化学中的问题。两种卡尔曼滤波器架构，主从和多个同行过滤器，是用来做多组分色谱，流动注射分析和化学动力学数据。在研究中进行数据分析应用高信息含量二极管阵列分光光度数据。信息化应用卡尔曼滤波算法是中央工作的主题。在第一项研究，信息矩阵的行列式评估，以确定信息内容。通过使用这些结果，延长卡尔曼滤波器只适用于高信息含量的数据，导致降低了计算负担。在另一项研究中，所产生的新信息不断变化的化学估算，分析色谱数据的同时，被用来在一个动态的过滤模型。这导致提高滤波器的估计通过消除引起的模型误差。另一种应用色谱数据处理了第十三共流出色谱对使用卡尔曼滤波器模型可调整为从基线漂移的微妙变化。最后，流动注射数据通过使用矩阵的条件数的计算分析。通过这种方式，系统可观测性进行了评估确定波长的选择卡尔曼滤波模型。计算负担，减少了消除低从筛选模型的信息波长。在每项研究中，过滤效率考虑信息的位置和内容，或纳入解决新的信息，成为提供的数据进行分析后启动。第十四第1章引言科学家和工程师长期研究的空间和时间的字符在各种媒体上的化学物质的teristics。时间依赖的功能化学物质是这些群体的共同利益。发生的时间依赖性如下：通过导管或浓度变化的结果导致过境从化学反应。这些团体的工作人员已经开发工具设计完成的独立学科的目标。本研究合并工具从工程设计与分析化学的工具。其结果是一个独特的方式来进行化学由SES。信息理论建立描述信息数据信息mation内容定量的方式。信息理论的化学应用力求最大限度地提高指挥效率分析化学的实践注意方法，提供的信息内容，或场指挥在tention特定区域具有较高的信息数据。一种技术，产生更大量的信息，加快决策进程（校准或重复一个吻SES），有可能改善分析的效率。分析技术的效率也可能提高考虑到数据信息的位置。部分含有少量的数据。从审议的数据信息和冗余数据可以消除分析算法，