摘要
对45种烷烃的0阶到4阶分子连接性指数进行主成分分析,将得到的4个主成分变量作为贝叶斯正则化神经网络的输入特征参数,对烷烃的密度进行预测.预测结果与实验值的相关系数为0.996,预测标准偏差为0.002 8.因此,人工神经网络可以作为预测烷烃密度的有效手段.
The densities of 45 kinds of alkanes are predicted successfully by Bayesian-Regularization neural network in this paper. 4 principal component variables are used as input parameters of the BRNN. Correlation coefficient between the predictions and experiments is 0. 996 and standard deviation is 0. 0028. The ANN is proved to be an effective method of predicting organic properties of compounds.
出处
《内蒙古师范大学学报（自然科学汉文版）》
CAS
2006年第1期80-82,共3页
Journal of Inner Mongolia Normal University(Natural Science Edition)
关键词
主成分
分子连接性指数
贝叶斯正则化神经网络
密度
principal component
molecular connective index
bayesian-regularization neural network
density
