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ｌｅａｖｅｏｎｅｏｕｔ方法得到的标准预报误差（ＳＤＥＰ）为０３４ｅＶ，好于文献报导的用ＰＬＳ方法的计算和预报结果。
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.
出处
《计算物理》
CSCD
北大核心
1997年第6期770-776,共7页
Chinese Journal of Computational Physics
关键词
神经网络
优化
预报
电离热
卤代烷烃
neural network
optimization
prediction
ionization potential.
