饮用水中挥发性有机物色谱保留时间的神经网络研究

研究饮用水中挥发性有机物的色谱保留时间与分子结构之间的定量结构-保留相关关系，基于分子结构和邻接矩阵，计算了56?个挥发性有机物的分子连接性指数、形状指数、电性拓扑状态指数和电性距离矢量，建立挥发性有机物的保留时间与0X、1X、2X、3X、K1、E43和M91指数的定量结构-保留相关性（quantitative structure-retention relationship，QSRR）模型。将这7 种结构参数作为BP（back propagation）人工神经网络法的输入变量，采用7∶4∶1的神经网络结构，建立了令人满意的QSRR预测模型，模型的总相关系数r总为0.999 1，利用本模型计算得到色谱保留时间的预测值与相关实验值相对平均误差2.17%，吻合度较为理想。结果表明，饮用水中挥发性有机物的色谱保留时间与7 种结构参数之间具有良好的非线性关系，本研究对快速评价水质对生态环境的影响具有参考价值。

Predicting Retention Times of Volatile Organic Compounds in Drinking Water by Neural Network

In order to study the quantitative structure-retention relationship (QSRR) between the chromatographic retention times and molecular structures of volatile organic compounds in drinking water, the molecular connectivity index, shape index, electrotopological state index and electrical distance vector of 56 volatile organic compounds were calculated based on their molecular structures and conjugation matrix. Further, the QSRRs between the retention times (tR) and seven structural parameters (0X, 1X, 2X, 3X, K1, E43 and M91) of these volatile organic compounds were developed. Using the structural parameters as the input variables of artificial neural network, satisfying QSRR models whose network structure was 7:4:1 were constructed by the back-propagation neural network (BNN) method. The total correlation coefficient rT was 0.999 1. The average relative error between the experimental and the predicted values (tR) was 2.17%, indicating good agreement. These results showed that there was a good non-linear relationship between the retention times and the seven structural parameters. This research would be helpful to quickly test the impact of water quality on the environment.