摘要
本文从有机化合物分子三维空间结构出发,用其原子间的空间距离以替代二维距离矩阵,传统表示方法中的拓扑距离(键的数目),首次提出并建立表示分子空间结构的三维距离矩阵计算模型,并基于三维距离矩阵提出三维结构拓扑指数3- DM(W),将分子所含原子个数和支化度组合作为BP人工神经网络(BPANN)的输入参数,利用BPANN预测烷烃化合物的临界温度(T_c),结果精度符合化工计算的要求。可以预料三维距离矩阵计算模型极可能在定量结构-性能关系研究中起重要作用。
Base on the introduction of the space distance between atoms of the compound moleculars instead of the traditional topographic distance in two-dimensional topographic distance matrix, a model of three-Dimensional distance matrix which reflected the space structures of the compound moleculars was developed. A novel three-dimensional topological index 3-DM （W） was proposed on the basis of this three-dimensional distance matrix and the same calculation formular of Wiener index （W）. A back-propagation artificial neural networks （BPANN） model was used to predict the critical temperatures of alkanes with 3-DM （ W）, number of atom and degree of branching as the input parameters. The predicted results showed a better agreement with the experimental values. It can be seen that the three-dimensional distance matrix may play an important role in quantitative structure-property relationship （QSPR） studies.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2008年第1期123-125,共3页
Computers and Applied Chemistry
关键词
三维距离矩阵
拓扑指数
BP人工神经网络
烷烃
临界温度
three-dimensional distance matrix, topological index, BPANN, alkanes, critical temperature
