The molecular electrongativity-distance vector(MEDV) was used to describe the chemical structure of 209 polybrominated diphenyl ethers(PBDEs). The quantitative structure-retention relationship (QSRR) model of 6 MEDV variables was built by multiple linear regression (MLR) with the correlation coefficients (R) being 0. 995. A new model of 5 MEDV descriptors could be obtained by stepwise multiple regression(SMR)with the correlation coefficients (ft) being 0. 995 too. The predicted and observed chromatograph-ic relative retention times(RRT)were proved to be in accord with cross-validation (CV) .performed by leave-one-out (LOO) procedure with good results of correlation coefficients(R2cv)being 0. 987 and 0. 988 for all 6 MEDV variables and 5 MEDV variables respectively. It was concluded that the models can provide satisfactory estimation stability and favorable predictive ability.%应用分子电性距离矢量(MEDV)对多溴联苯醚(PBDEs)的209种同系物进行结构表征.通过多元线性回归的方法,建立了PBDEs定量结构-色谱保留(QSRR)关系的6个变量和5个变量的两种模型.两种模型的建模计算值复相关系数R均为0.995;用留一法(LOO)进行了交互检验,其复相关系数(R2cv)分别为0.987和0.988;结果表明:此定量结构活性相关模型具有较高的结构选择性和性质相关性,良好的估计能力和稳定性以及计算简便等优点.并用这两个模型对未知的多溴联苯醚( PBDEs)的色谱保留时间(RRT)进行了预测.
展开▼