将自相关拓扑指数作为分子描述符 ,研究了多氯联苯类化合物的水溶解性与分子结构的定量关系。将遗传算法 (GA)引用于本研究中的特征提取 ,并将神经网络法用于数学建模 ,与相应体系的水溶解性进行相关分析时发现 ,自相关拓扑指数能较好地反映化合物的结构特征 ,基于GA算法提取的参量用于神经网络法研究得到较优结果。 Using the self-correlated topological index as a molecular descriptor, the quantitative relationship between water solubility and molecular structure of polychlorinated biphenyls (PCBs) was studied. The genetic algorithm (GA) is applied to the feature extraction in this study, and the neural network method is used in mathematical modeling. Correlation analysis with the water solubility of the corresponding system shows that the autocorrelation topological index can better reflect the compound Structural characteristics, GA-based parameters extracted by the neural network method to get better results.