计算物理 ›› 1997, Vol. 14 ›› Issue (6): 770-776.

• 论文 • 上一篇    下一篇

人工神经元网络模拟卤代烷烃的电离势

贺黎明, 包于诗, 陆慧, 金乾元   

  1. 华东理工大学物理系, 上海 200237
  • 收稿日期:1996-04-04 修回日期:1997-04-12 出版日期:1997-11-25 发布日期:1997-11-25

MODELLING THE IONIZATION POTENTIAL OF HALO ALKANES BY NEURAL NETWORK APPROACH

He Liming, Bao Yushi, Lu Hui, Jin Qianyuan   

  1. Department of Physics, East China University of Science and Technology, Shanghai 200237
  • Received:1996-04-04 Revised:1997-04-12 Online:1997-11-25 Published:1997-11-25

摘要: 利用人工神经网络反向传播(BP)模型计算了卤代烷烃第一电离势。通过优化神经网络结构,在已知样本范围内由leave-one-out方法得到的标准预报误差(SDEP)为0.34eV,好于文献报导的用PLS方法的计算和预报结果。

关键词: 神经网络, 优化, 预报, 电离势

Abstract: The Back propagation(BP) model of neural network is used to calculate the first ionization potention(IP) of haloalkanes.By optimizing the neural network,the standard deviation of errors of prediction(SDEP) using leave one out method within the known samples is 0.34eV,which is better than the reported predicting and calculating results by PLS.

Key words: neural network, optimization, prediction, ionization potential

中图分类号: