稠环烃有机半导体禁带宽度的人工神经网络预报  

The ANN Prediction of the Band Gap ofFused Cyclic Hydrocarbons

作　　者：张兆春[1] 彭瑞伍[1] 陈念贻[1] 詹千宝[1] 

机构地区：[1]中国科学院上海冶金研究所

出　　处：《Chinese Journal of Chemical Physics》1998年第5期427-431,共5页化学物理学报（英文）

摘　　要：稠环烃有机半导体的禁带宽度与其分子构型以及π电子数有关，将可表征分子几何结构特征的拓扑指数一顶点复杂度信息指数和π电子数作为人工神经网络的输入特征量，利用经已知样本集训练的人工神经网络对稠环烃有机半导体的禁带宽度进行预报，预报结果与实测结果符合较好，表明人工神经网络是进行定量结构—性质或定量结构—活性相关性研究的一种有效方法。The band gap of some fused cyclic hydrocarbons,organic semiconductors,dependsupon both the numbers of electron and the molecular constitution of them.An informationtheoretic topological index,namely degree complexity,and the number of electron are choosenas the input features of an artificial neural network,which is trained by available experimentaldata and has been used to predict the values of the band gap of some fused cyclichydrocarbons.The good agreement between the predicted results and the experimental ones indi-cates that artificial neural network is an efficacous approch for quantitative structure-propertyand structure-activity relationship studies.

关 键 词：禁带宽度 人工神经网络 稠环芳烃 有机半导体 

分 类 号：TN304.51[电子电信—物理电子学] O625.15[理学—有机化学]