摘要:
酚类化合物(BP)是重要的工业原料或中间体,但工业废水含有的酚类化合物会对环境造成污染。为建立酚类化合物臭氧氧化速率的QSPR(quantitative structure-property relationship)预测模型,分析了23种酚的分子结构与臭氧氧化速率之间的相关关系,计算了这些酚的分子连接性指数和分子形状指数,优化筛选了连接性指数的1χ和2χ、分子形状指数的K1和K2共4种参数,将其作为BP神经网络的输入层变量,臭氧氧化速率作为输出层变量,采用4:2:1的网络结构,获得了令人满意的QSPR神经网络预测模型,模型总相关系数r为0.976,计算得到的臭氧氧化速率的预测值与实验值较为吻合,平均残差仅为0.05;为检验结构参数建立模型的普适性,同样方法建立对酚类化合物的辛醇-水分配系数的预测模型,模型总相关系数r达到0.993,辛醇-水分配系数的预测值与实验值吻合度较为理想,结果表明,本法建构的神经网络模型具有良好的稳健性和预测能力。
Abstract:
Phenolic compounds were important industrial raw materials or intermediates, but industrial wastewater containing phenolic compounds was polluted to the environment. In order to establish QSPR (quantitative structure-property relationship) model of ozonation rate of phenolic compounds, the relationship between molecular structure and the ozonation rate of 23 kinds of phenolic compounds was analyzed. Moreover, the molecular connectivity indices and molecule shape indices of these compounds were calculated. 1χ and 2χ of the molecular connectivity indices, K1 and K2 of the molecule shape indices were optimized. The four parameters were used as input variables of neural network and the ozonation rate was used as output variable, and the 4:2:1 network structure was adopted and BP neural network method was used to establish a satisfying QSPR prediction model. The total correlation coefficient r was 0.976. The predicted values and experimental values were very close, and the mean error was 0.05. In order to test the generality of our method, a QSPR model of octanol-water partition coefficient lgp of phenolic compounds was established using the same method. The total correlation coefficient r was 0.993. The predicted values of lgp agree with the experimental values. The results showed that the neural network model had good stability and predictive ability.
点击查看大图
下载:
