摘要
采用分子子图编码法将烷烃的分子子图码作为人工神经网络的输入参数 ,对烷烃的吉布斯自由能进行预测 .将预测值与文献值进行线性拟和 ,其回归系数达到 0 9812 .
On the basis of molecular subgraph coding method, a back propagation neural network was trained on the molecular subgraph code to predict Gibbs energy of alkanes. A satisfied result is given. The correlation coefficients (R) of linear regression equation is 0 9812.
出处
《深圳大学学报（理工版）》
CAS
2000年第2期41-45,共5页
Journal of Shenzhen University(Science and Engineering)
基金
"九五"国防预研课题资助项目! (1 2N1 961 3 - 2 )
