论文部分内容阅读
根据592种有机化合物实测Koc数据,研究了用片段常数法估算有机化合物土壤吸着系数的可能性。将592种化合物随机划分为建模组和验证组。前者用于模型建立,后者用于模型检验。计算结果证明,包括74种结构片段和24个结构校正因子的回归模型能够很好地预测有机化合物的吸着系数。对所有592种化合物而言,模型的可次系数高达0.9696。估算结果的平均绝对误差仅为0.37个对数单位。
Based on the measured Koc data of 592 organic compounds, the possibility of estimating the soil sorption coefficient of organic compounds by fragment constant method was studied. 592 compounds were randomly divided into modeling and validation groups. The former is used for modeling and the latter for model checking. The calculation results show that the regression model of 74 structural fragments and 24 structural correction factors can predict the adsorption coefficient of organic compounds well. For all 592 compounds, the model has a secondary coefficient of up to 0.9696. The average absolute error of the estimates is only 0.37 log units.