摘要
以拓扑指数为结构描述符 ,用基于Levenberg_Marquardt优化的BP神经网络建立了醇类化合物的结构与色谱保留值的相关性模型 ,用于未知醇类化合物在SE_30和OV_3两根色谱柱上保留指数的同时预测 ,其学习速率优于文献中普通BP神经网络法,预测准确度与普通BP神经网络法接近,但优于多元线性回归法 。
A model of back _ propagation artificial neural network based on Levenberg _ Marquardt algorithm for determining the relations between the structure of alcohols and their chromatographic retention indices has been set by means of topological indices. The model was used for the simultaneous prediction of retention indices of alcohols on SE _ 30 and OV _ 3 columns. The results showed that the method had higher training rate and equal accuracy in comparison with common BP artificial neural network,and better accuracy than that of multiple linear regression. Therefore, the model is a more satisfactory method for prediction of chromatographic retention indices of organic compounds.
出处
《分析测试学报》
CAS
CSCD
北大核心
2003年第1期21-23,共3页
Journal of Instrumental Analysis
基金
安徽省自然科学基金资助项目(00046509)