摘要
本文用人工神经网络方法，以ＣｒａｍｅｒⅢ的ＢＣＤＥＦ参数作为输入参数，预测有机物物性．对于１５３个不同类型有机物的亨利常数（Ｈ）的预测结果为：相关系数ｒ＝０．９８１３，标准差ＳＤ＝０．３３６９ｌｏｇＨ单位，对于９１种有机物在水中溶解度（Ｓ）的预测结果为：相关系数ｒ＝０．９８５５，标准差ＳＤ＝０．２８２７ｌｏｇＳ单位。
A back-propagation neural network was trained on the Cramer's param-eters (BCDEF)to predict the physical properties of organic compounds. The predictionresults of Henry's Law constant of 153 organic compounds are well correlated with theexperiment values(r= 0. 9813 and standard deviation is 0. 3369 log unit).The estima-tion of the aqueous solubilities of 91 organic compounds using neural networks is alsowell correlated with the observed values(r= 0. 9855 and standard deviation is 0. 2827log unit).
出处
《辽宁大学学报（自然科学版）》
CAS
1995年第2期38-41,78,共5页
Journal of Liaoning University：Natural Sciences Edition
