【摘要】 本文将分子结构表征方法(MEDV)进行改进,得到按非氢原子间距离分类的分子电性距离矢量,将该矢量用于酚类化合物结构表征,并与其土壤吸收系数(lgKoc)建立定量结构与性质关系(QSPR)模型。利用逐步回归(SMR)得到的4变量模型复相关系数(R2)为0.970、标准偏差(SD)为0.198,留一法(LOO)交互校验(CV)预测值的复相关系数(R2cv)为0.916、标准偏差(SDcv)为0.337。结果表明该矢量具有较强的分子结构表达能力,模型具有良好的预测能力与稳定性。更多还原

【Abstract】 A molecular structure characterization method called distance-association classified molecular electronegativity-distancevector(D-MEDV)was developed and used to build quantitative structureproperty relationship(QSPR)model of soil sorption coefficient(lgKoc)of phenols.Through the stepwise multiple regression(SMR)method,4vectors were selected to build the model with the correlation coefficient(R2)of 0.970and the standard deviation(SD)of 0.198,respectively.The model was evaluated by performing the cross validation with the Leave-One-Out(LOO)procedure and the results with correlation coefficient(R 2c v)of 0.916and standard deviation(SDcv)of 0.337could be obtained,respectively.The results show that the vector has a strong ability to express the molecular structures and the model has good predictive ability and stability.更多还原