论文部分内容阅读
采用量子化学密度泛函B3LYP/6-311+G~*,在高斯09软件上计算了32个多氯联苯类化合物的电子结构参数:筛选出影响化合物色谱保留时间显著的5个变量,并建立其结构与保留时间之间的定量关系(MLR模型);同时,利用人工神经网络(artificial neural network,ANN)法建立相应的QSRR模型(ANN模型)与之对比。所建MLR模型的相关系数R=0.904,标准误差Se=0.542;ANN模型的相关系数R=0.981,标准误差Se=0.213。表明所建立的QSRR模型的稳定性和预测能力良好。结果表明,多氯联苯化合物的色谱保留时间与前沿轨道能级差ΔE和分子最高占有轨道能E_H成正比例关系。所建模型为预测多氯联苯化合物的色谱保留时间提供理论指导。
The electron structure parameters of 32 polychlorinated biphenyls (PCBs) were calculated on Gaussian 09 software by using the quantum chemical density functional theory B3LYP / 6-311 + G ~ *: Five variables that affected the chromatographic retention time of the compounds were screened out. The quantitative relationship between structure and retention time (MLR model) was established. Meanwhile, the artificial neural network (ANN) method was used to establish the corresponding QSRR model (ANN model). The correlation coefficient R = 0.904, the standard error Se = 0.542, the correlation coefficient R = 0.981 and the standard error Se = 0.213 of the MLR model. It shows that the established QSRR model has good stability and prediction ability. The results show that the chromatographic retention time of PCBs is directly proportional to the difference in frontier orbital energy levels ΔE and the highest occupied molecular orbital energy E_H. The model provided theoretical guidance for predicting the chromatographic retention time of PCBs.