人工神经网络方法预测气相色谱保留值

本文运用一典型的人工神经网络模型－“反向传播“模型的改进形式，研究了诱导效应指数Ｉ，摩尔折射度Ｒｏ，疏水亲脂参数ＩｇＰ，以及分子联通性指数与气象色谱保留行为的关系，实现了对色谱保留植的预测。神经网络预测模型的最大相对误差不超过８．７％。结果表明，该方法性能良好，可望成为色谱保留值预测的有效手段。     

色谱保留值方程的统一形式

本文采用点阵模型,运用统计热力学方法推导出色谱保留值方程的统一形式。根据不同的条件从此统一形式可获得描述气-固色谱、气-液色谱、液-固色谱、液-液色谱以及超临界流体色谱保留值规律的方程。上述方程均采用实验数据予以验证。     

色谱灰色数模的建立及应用研究(Ⅳ)-CGM(1,1,N)模型的建立及烷基苯异构体保留行为的预测

应用灰色系统理论,推导出灰色N参数模型CGM(1,1,N),突破了现有单参数灰色模型的局限性.并用该模型讨论了色谱保留值与分子结构参数间的关系,对烷基苯的同分异构体色谱保留值及疏水性参数进行了预测.经实例验证,其预测精度较单参数灰色模型高,这为溶质保留行为的预测提供了一个实用方法,同时也拓宽了灰色模型的应用范围.     

氨基苯磺酸和氨基萘磺酸在反相离子对色谱和离子色谱中分离行为的比较

本文考察了氨基苯磺酸和氨基萘磺酸染料中间体在反相离子对色谱和离子色谱中的分离行为，结果表明一些异构体在反相色谱中的保留顺序与离子色谱中的顺序发生颠倒。静电作用力在离子色谱中对保留值的贡献比在反相离子对色谱中对保留值的贡献要大，而疏水作用力在反相离子对色谱中对保留值的贡献却较在离子色谱中对保留值的贡献要大。离子对色谱更有利于有机离子的分离分析。     

氨基苯磺酸和氨基萘磺酸在反相离子对色谱和离子色谱中分离行为的比较

 本文考察了氨基苯磺酸和氨基萘磺酸染料中间体在反相离子对色谱和离子色谱中的分离行为，结果表明一些异构体在反相色谱中的保留顺序与离子色谱中的顺序发生颠倒。静电作用力在离子色谱中对保留值的贡献比在反相离子对色谱中对保留值的贡献要大，而疏水作用力在反相离子对色谱中对保留值的贡献却较在离子色谱中对保留值的贡献要大。离子对色谱更有利于有机离子的分离分析。     

色谱灰色数模的建立及应用研究（Ⅳ）：CGM（1,1,N）模型的 …

应用灰色系统理论，推导出灰色Ｎ参数模型ＣＧＭ（１，１，Ｎ），突破了现有单参数灰色模型的局限性，并用该模型讨论了色谱保留值与分子结构参数间绎烷基苯的同分异构体色谱保留值及疏水性参数进行了预测，经实例验证，其预测精度较单参数灰色模型高，这为溶质保留行为的预测提供了一个实用方法，同时也拓宽了灰色模型的应用范围。     

反相液相色谱中保留值变化规律的研究——保留值收敛性的理论探讨与实验验证

本文根据反相液相色谱中顶替吸附多种相互作用模型,系统地研究了不同色谱参数(如流动相组成、柱温、溶质、溶剂及键合相配体分子中的碳数)对保留值收敛性的影响,并用热力学方法导出了4个新的保留值收敛方程,已经在各种同系物实验中得到了验证。另外,本文还建立了一种能够采用计算机直接从实验数据计算保留值收敛点坐标的方法,预测值与做图法得到的结果完全一致。     

从分子微观参数预测反相色谱保留值方程系数

在考虑氢键作用能随流动相组成改变的情况下,重新推导了液相色谱保留值方程,进而得到液相色谱保留值方程系数与分子微观参数之间的关系。在分子母体结构相同条件下,提出采用五个系数预测反相色谱保留值的方法,并用文献数据给予验证。     

有机磷农药分子空间坐标指数与色谱保留值的QSPR研究

根据分子空间坐标数据,定义了原子距离指数Ys、分子空间距离指数yF和空间坐标特征值指数Yλ来表征有机磷农药（OPS）的分子结构特征;运用误差反向传输（BP）人工神经网络方法建立21种OPS空间坐标指数与气相色谱保留值定量结构性质分析（QSPR）模型,预测了气相色谱保留值.结果得出实验室值与预测值的相关系数R=0.950 2,所建模型稳定性和预测能力均良好.     

人工神经网络用于有机磷酸酯类化合物的定量结构色谱保留相关研究

采用误差反传前向人工神经网络(ANN),研究了35种有机磷酸酯类化合物在3种不同极性固定相上的结构与其色谱保留(QSRR)之间的定量关系。以其分子电性距离矢量(或分子拓扑指数)作为输入、色谱保留值作为输出,采用内外双重验证的办法分析和检验所得模型的稳定性和外推能力。结果表明,ANN模型获得了比多元线性回归(MLR)模型更好的拟合效果。使用MLR模型时QSRR模型相关性受色谱固定相极性的影响,而采用ANN模型无此现象。同时,ANN模型解决了QSRR中预测维数为1时耗时较长的问题。通过ANN建模可以同时预测3种不同极性固定相上的色谱保留值,可大大缩短建模和预测所需的时间。     

Evaluation of the Linear-Elution-Strength Model for the Prediction of Retention and Selectivity in Isothermal Gas Chromatography

Linear-elution strength theory and temperature-programmed gas chromatography is evaluated as a rapid method for predicting isothermal retention factors and column selectivity. Retention times for a wide range of compounds are determined at the program rates of 3 and 12 °C/min for the temperature range 60 to 160 °C on three open-tubular columns (DB-1701, DB-210 and EC-Wax) and used to predict isothermal retention factors for each column over the temperature range 60 to 140 °C. The temperature-program predicted isothermal retention factors are compared with experimental values using linear regression and the solvation parameter model. It is shown that isothermal retention factors predicted by the linear-elution-strength model only approximately represents the experimental data. The model fails to predict the slight curvature that exists in most plots of the experimental retention factor (log k) as a function of temperature. In addition, regression of the temperature-program predicted isothermal retention factors against the experimental values indicates that the slopes and intercepts deviate significantly from their target values of one and zero, respectively, in a manner which is temperature dependent. The temperature-program predicted isothermal retention factors result in system constants for the solvation parameter model that are different to those obtained from the experimental retention factors. These results are interpreted as indicating that linear-elution-strength theory predicts retention factors that fail to accurately model stationary phase interactions over a wide temperature range. It is concluded that temperature-program methods using linear-elution-strength theory are unsuitable for constructing system maps for isothermal separations.     

反相液相色谱中同系物体系实验点与预测方程参数间的匹配研究

以波相色谱中溶质计量置换统一保留模型（SDM-R）的二参数、三参数和四参数方程为基础,研究了在预测同系物保留值时实验点与预测方程参数间的匹配关系。用两个实验点,分别以二、三和四参数方程对反相色谱中同系物保留值进行预测,发现预测值与实验值基本相符,但以二参数方程预测结果为最佳。又与其它保留方程的预测结果进行了比较,证实了上述结论。当实验点数增加到4～5个时,多参数方程预测同系物的准确度增加,且明显高于二参数方程。同时还讨论了方程中各参数与同系物碳数间的线性关系,发现二参数方程中的参数遵循同系物变化规律。     

Predicting retention data by target factor analysis and multiple regression analysis

Target factor analysis is used to predict gas-chromatographic retention indices from a training-set data matrix for 13 solutes and 15 stationary phases. In the target-combination approach, sets of data vectors are target-tested in combination and the resulting coefficients for the best model are used for prediction. Retention indices for 42 solutes and 24 stationary phases are predicted to better than 1% even with a three-factor model. In the target free-float approach, values for missing retention indices on target test vectors are predicted. Predictions from sets of target-test data selected by chemical intuition are compared to those obtained from sets of target-test data selected by using models from the combination step. The target-combination approach and multiple-regression approach are overall of similar utility for predicting new data.     

Prediction of protein retention times in gradient hydrophobic interaction chromatographic systems

A two-step methodology has been developed for the prediction of protein retention time in linear-gradient HIC systems. Isocratic retention parameters were determined from ln(k')-salt concentration plots for a number of commercially available proteins with a range of properties. Quantitative structure property relationship (QSPR) models based on a support vector machine (SVM) approach were generated for predicting isocratic retention parameters for proteins not included in the model generation. The predicted parameters were then used to calculate protein gradient retention times and the results indicate that this approach is well suited for predicting experimental gradient retention data. The approach presented in this paper may have implications for HIC methods development at both the bench and process scales.     

Retention models for ionizable compounds in reversed-phase liquid chromatography: Effect of variation of mobile phase composition and temperature

General models in reversed-phase liquid chromatography that have been extended to relate retention of ionizable compounds to mobile phase composition, pH and/or temperature are reviewed. In particular, the fundamentals and applications of the solvation parameter model, the polarity parameter model and several classical models based on empirical equations are presented and compared. A main parameter in all these models is the degree of ionization of the acid–base compound, which depends on both the pH of the mobile phase and the acid–base constant of the compound. Thus, on one hand, the different procedures for pH measurement in the mobile phase and their influence on the performance of the models are outlined. On the other hand, equations that relate the variation of the pH of the buffer and the pK_a of the compound with the mobile phase composition and/or temperature are reviewed and their applicability to the retention models critically discussed.     

Reduced combustion time model for methane in gas turbine flow fields

Computational fluid dynamics (CFD) modeling of the complex processes that occur within the burner of a gas turbine engine has become a critical step in the design process. However, due to computer limitations, it is very difficult to completely couple the fluid mechanics solver with the full combustion chemistry. Therefore, simplified chemistry models are required, and the topic of this research was to provide reduced chemistry models for CH4/O2 gas turbine flow fields to be integrated into CFD codes for the simulation of flow fields of natural gas-fueled burners. The reduction procedure for the CH4/O2 model utilized a response modeling technique wherein the full mechanism was solved over a range of temperatures, pressures, and mixture ratios to establish the response of a particular variable, namely the chemical reaction time. The conditions covered were between 1000 and 2500 K for temperature, 0.1 and 2 for equivalence ratio in air, and 0.1 and 50 atm for pressure. The kinetic time models in the form of ignition time correlations are given in Arrhenius-type formulas as functions of equivaience ratio, temperature, and pressure; or fuel-to-air ratio, temperature, and pressure. A single ignition time model was obtained for the entire range of conditions, and separate models for the low-temperature and high-temperature regions as well as for fuel-lean and rich cases were also derived. Predictions using the reduced model were verified using results from the full mechanism and empirical correlations from experiments. The models are intended for (but not limited to) use in CFD codes for flow field simulations of gas turbine combustors in which initial conditions and degree of mixedness of the fuel and air are key factors in achieving stable and robust combustion processes and acceptable emission levels. The chemical time model was utilized successfully in CFD simulations of a generic gas turbine combustor with four different cases with various levels of fuel-air premixing.     

Retention in coupled capillary column supercritical fluid chromatography with lipids as model compounds

Summary We have investigated to which extent retention data, acquired on single capillary columns, can be used for predicting retention factors in a coupled column system. For this purpose we utilized a model mixture of 18 lipid components with widely different vapor pressures and polarities. The sample was chromatographed on two columns, SB-biphenyl-30 (70% methyl-30% biphenylpolysiloxane) and SB-cyanopropyl-50 (50% methyl-50% cyanopropylsiloxane). Experimental retention factors, acquired in coupled column systems with two columns connected in different order, were thus compared with values calculated from runs on each single column. The agreement between calculated and experimental values generally was better than 5% without any pressure drop correction.To study the possibility of predicting retention behavior in a wide pressure range from a limited number of experiments, we also investigated the relation between solute retention and mobile phase density. We found that all data could be fitted to second order equations, which gives the possibility to optimize the resolution with respect to pressure from a limited number of runs at different pressures.