有机醇的结构与色谱保留指数的相关性拓扑指数法研究

通过计算40种有机醇化合物分子结构的拓扑指数，用多元线性回归将它与气相色谱保留指数进行关联分析，结果获得了良好的相关性，因此，对有机醇各组分在气相色谱上保留指数的理论预测，可作为在色谱分离过程中识别各组分的一种参考方法。     

Soft sensor for continuous product quality estimation (in crude distillation unit)

Due to the strict norm requirements of keeping products in crude refining units within specifications, laboratory testing and quality control of the products are necessary. Given this reason, virtual soft sensor for continuous quality estimation of light naphtha as the crude distillation unit (CDU) product was developed. Experimental data included available continuous measurements of CDU process streams (temperatures, pressures and flowrate) and laboratory analyses undertaken twice a day. The results are soft sensor models for light naphtha vapor pressure (RVP) estimation.Soft sensor models have been developed conducting multiple linear regression analysis and using neural network-based models such as LNN, MLP and RBF. Considering statistical and sensitivity analysis, the best results for both oils were obtained with MLP and RBF neural networks. The results show possible application of the soft sensor models for estimating light naphtha RVP as an alternative for laboratory testing.     

硫代氨基甲酸酯类衍生物抗HIV-1逆转录酶抑制活性的QSAR研究

采用以MLR为基学习器的Boosting算法模型,对79种硫代氨基甲酸酯类衍生物做抗HIV-1逆转录酶抑制活性的QSAR研究。以E-Dragon软件计算的7组描述符分别为自变量,以化合物的半数效应浓度EC_(50)值为因变量构成7个原始数据集,用PSO算法筛选变量并建立MLR模型。各描述符建立的MLR模型中仅有RDF描述符模型同时通过外部预测和内部验证,故确定以其建立关于硫代氨基甲酸酯类衍生物抗HIV-1逆转录酶抑制活性的Boosting-MLR预测模型。Boosting-MLR模型与MLR模型相比,训练结果的决定系数R~2分别为0.728和0.741,预测结果R~2则分别为0.718和0.667,表明其泛化能力明显增强。对Boosting-MLR模型进一步进行稳定性验证,证明其预测稳定性较高。     

原油闪点预测模型的研究

针对原油这类结构组成复杂、差异性大、可燃的复杂混合体系,选取各地区共计101种原油的恩氏蒸馏温度、20℃密度、20℃粘度作为输入变量,建立原油闪点预测模型。采用主成分分析法对输入变量进行降维,除去恩氏蒸馏系列数据中的信息冗余,分别采用多元线性回归(MLR)、BP神经网络、RBF神经网络三种方法建模,并对模型的预测结果进行对比,RBF神经网络模型的预测准确度与稳定性均为最优,绝对误差期望为2.94℃,相对误差期望为3.45%,BP神经网络模型的准确性优于多元线性回归模型,稳定性不如MLR模型。     

Evaluation of apple texture with contact acoustic emission detector: A study on performance of calibration models

Performance of calibration models for evaluation of apples sensory texture with contact acoustic emission detector (CAED) was studied. For model evaluation and testing, 2500 apples of 19 cultivars were harvested over two seasons. Apples were stored at normal atmosphere (NA), controlled atmosphere (CA) for different periods or were treated with 1-methylcyclopropene (1-MCP) in order to obtain a high variability of texture and fruit maturity. Apples were tested simultaneously in two distinct laboratories. The models were created and validated on averaged values from 10 fruits using simple linear regression, multiple linear regression (MLR) and principal component regression (PCR). Performance statistics of the models were expressed in terms of determination coefficient (R²), root mean square errors of cross validation (RMSECV) or prediction (RMSEP) and ratio of prediction to deviation (RPD). Firmness and total acoustic emission counts were predictors of sensory texture in the models. MLR and PCR models show better performance for prediction of sensory data than simple linear regression models however PCR models show the best results among models tested in this study. Common PCR models for several cultivars allow for successful prediction of hardness (RPD > 2.0), crispness and overall texture (1.5 < RPD < 2.0). The single-cultivar PCR models, constructed on data sets containing 26-39 averaged values, reveal significantly better performance (RPD > 2.0 for most of the cases) than the common PCR models for many varieties.     

Poisson parameters of antimicrobial activity: a quantitative structure-activity approach

A contingency of observed antimicrobial activities measured for several compounds vs. a series of bacteria was analyzed. A factor analysis revealed the existence of a certain probability distribution function of the antimicrobial activity. A quantitative structure-activity relationship analysis for the overall antimicrobial ability was conducted using the population statistics associated with identified probability distribution function. The antimicrobial activity proved to follow the Poisson distribution if just one factor varies (such as chemical compound or bacteria). The Poisson parameter estimating antimicrobial effect, giving both mean and variance of the antimicrobial activity, was used to develop structure-activity models describing the effect of compounds on bacteria and fungi species. Two approaches were employed to obtain the models, and for every approach, a model was selected, further investigated and found to be statistically significant. The best predictive model for antimicrobial effect on bacteria and fungi species was identified using graphical representation of observed vs. calculated values as well as several predictive power parameters.     

Probabilistic simulation of cable performance and water based protection in cable tunnel fires

Nuclear power plants contain a significant amount of fire load in form of electrical cables. The performance of the cables is interesting both from the fire development and system failure viewpoints. In this work, cable tunnel fires are studied using numerical simulations, focusing on the fire spreading along power cables and the efficiency of the water suppression in preventing the cable failures. Probabilistic simulations are performed using Monte Carlo technique and the Fire Dynamics Simulator (FDS) as the deterministic fire model. The primary fire load, i.e. the power cables are modelled using the one-dimensional pyrolysis model, for which the material parameters are estimated from the experimental data. Two different water suppression systems are studied. The simulation results indicate that using either suppression system decreased the heat release rate in the tunnel to less than 10% of the non-suppressed case. Without water suppression, the cables of the second sub-system were damaged in almost all fires, but when either of the studied water suppression systems was used, the probability of the cable failures was decreased to less than 1%. This result indicates that in current scenario, the probability of losing both sub-systems is determined directly by the suppression system unavailability.     

脂肪族醇的酸碱离解常数(pKa)的预测研究

酸碱离解常数(pKa)在理解受体与药物反应和药物如何选择辅料上起着重要的作用。因此,在药物前期的开发中,能有效地预测pKa值可很好地降低开发成本,降低资源消耗。本文通过运用定量构效关系的方法(QSAR)对脂肪醇的数据库展开研究,以进行pKa值的预测。一共选用了253种各类描述符,包含原子键的性质的描述符48个,分子整体描述符205个,其中重点描述符为原子键的性质类描述符。采取多元线性回归的方法对数据进行研究。该预测模型能取得较好的效果,可以应用于药物的前期开发中。