镧系元素定量构效关系的神经网络研究  

Study on the neural network of lanthanide's quantitative structure-property relationship

作　　者：王琴[1] 

机构地区：[1]青海民族学院化学系,青海西宁810007

出　　处：《青海大学学报（自然科学版）》2008年第1期56-59,共4页Journal of Qinghai University(Natural Science)

摘　　要：研究了影响镧系元素物理化学性质的各种因素。以镧系元素的基态L值、鲍林电负性、4f电子排布周期因子q以及离子半径等为参数,使用以L-M优化算法为训练方法的BP神经网络,对9种镧系元素物理化学性质进行了关联。得到的结果:该人工神经网络方法比普通的线性回归方法要好,适用于复杂的定量构效关系研究。The various influemce factors to the physicochemical properties of lanthanide are studied in this paper and the physicochemical properties of nine kinds of lanthanide are associated too using lanthanide foundation state value L, electronegativity 4f, electronic arrangement periodic factors q and ionic radius as the parameters and using the BP neural network based on L- M optimization algorithm. The obtained results are satisfactory and this method is better than the ordinary linear regress method and suitable to the study on completed quantitative structure - property relationship.

关 键 词：镧系元素 定量构效关系 L—M优化算法 BP神经网络 

分 类 号：O642[理学—物理化学]