【摘要】 依照经济合作与发展组织(OECD)关于定量结构-活性关系(QSAR)模型的构建和使用导则,采用偏最小二乘(PLS)法进行变量筛选和回归分析,建立了不同温度下酞酸酯类化合物(PAEs)蒸气压的定量预测模型。模型仅包含温度、相对分子质量、氢原子最高净正电荷3个参数。模型的应用域(AD)可由williams图较好地表征。结果表明,温度和相对分子质量是影响PAEs蒸气压的主要因素,且温度越高、相对分子质量越小时,蒸气压越大。内部验证与外部验证结果表明该模型具有良好的稳健性和预测能力。更多还原

【Abstract】 The paper is inclined to introduce a temperature-dependent quantitative predictive model for vapor pressures of environmental chemistry of phthalic acid esters （PAEs）.As we know,since vapor pressures （P）is one of the key physicochemical properties to deter- mine the environmental features of PAEs but it is hard to define them experimentally,it is of great significance to develop necessary tem- perature-dependent predictive models for testing such pressures.Ac- cording to the OECD （Organization for Economic Co-operation and Development）guideline for the validation of quantitative structure-ac- tivity relationship （QSAR）models,we have developed such an im- proved predictive model,in which 3 temperature related variables are introduced to illustrate the environmental temperature data with 15 molecular structural descriptive indexes listed by the PM3 method in MOPAC 2000.In addition,partial least squares （PLS）regression process has been adopted to describe the model by eliminating redun- dant descriptive indexes.Actually,the optimal model involves merely 3 variables,i.e.temperature,relative molecular mass,and the most positive net atomic charge on the hydrogen atoms.The other correlat- ed data are:R_γ²=0.972,RMSE=0.396,Q_CUM²=0.962,which account for the goodness-of-fit and robustness.Besides,Q_EXT² proves to be 0.880 with RMSE being 0.391,which helps to heighten the ef- ficient predictability of the model.Trial application shows that the proposed model is both simple to use and highly efficient.From the results of the primary application,it can be concluded that tempera- ture and relative molecular mass play a dominant role in governing the values of lgP,which proves to be increasing with the rise of the tem- perature and decreasing with the rise of relative molecular mass.更多还原