QSRR结合人工神经网络预测多氯联苯色谱保留时间  

Prediction for the retention time of PCBs compounds with QSRR-ANN

作　　者：陈欣[1] 金俏[2] 金飙[1] 

机构地区：[1]沈阳化工大学环境与安全工程学院,辽宁沈阳110142 [2]北京石油化工学院化学工程学院,北京102600

出　　处：《计算机与应用化学》2013年第5期515-518,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金(20773014)

摘　　要：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。For 32 PCBs, quantum chemistry calculation of electronic properties were carried out at density functional theory （DFT） B3LYP/ 6-311 ＋G＊ level by Ganssion09. 5 important parameters were selected and the quantitative structure retention relationship （QSRR） model was set up by multiple linear regressions （MLR） method. Furthermore, using artificial neural network （ANN）, the QSRR model was obtained in order to make contrast. For the artificial neural network method, the correlation coefficient R=0.981 and the standard error Se=0.213, while for the multiple linear regression analysis R=0.904 and Se=0.542. These shows that the QSRR models have both favorable estimation stability and good prediction capability. This indicates that the retention time of PCBs and both AE and Enare in direct proportion. Successful QSRRs were developed to predict the retention time of PCBs.

关 键 词：多氯联苯 人工神经网络 定量结构保留关系 密度泛函 

分 类 号：TQ015.91[化学工程] O6-39[理学—化学]