神经网络模型预测有机酸的毛细管电泳淌度  

Neural Network Model Prediction of the Capillary Electrophoretic Mobilities of Organic Acdis

作　　者：谢善梅[1] 

机构地区：[1]湖南人文科技学院化学系,湖南娄底417001

出　　处：《湖南人文科技学院学报》2008年第4期22-25,共4页Journal of Hunan University of Humanities,Science and Technology

摘　　要：通过分析化合物结构对毛细管电泳淌度的影响,提取三个分子结构参数——有机酸的元数(Q)、组成化合物的原子总数(N)及相对分子质量(M),对训练集中92个有机酸的毛细管电泳淌度建立径向基神经网络模型。对训练集的交叉验证及对测试集化合物的预测结果表明,该模型具有良好的稳健性和预测能力。与文献结果比较,该模型所用参数少,计算简便,物理意义明确。According to the influence of chemical structure on the capillary eleetrophoretic mobility,three molecular structur- al descriptors——the number of acidic groups(Q),the sum of atoms in compound(N)and the relative molecular mass(M)were employed as the input variables of radical basis function neural network to model the capillary electrophoretic mobilities of 92 or- ganic acids in the training set.The cress validation against the training set and the prediction result of the test set indicated that the neural network model built in this paper was of good stability and predictive ability.Compared with the reported results,this model employed less molecular descriptors,which were calculated easily and had clear physical meaning.

关 键 词：毛细管电泳淌度 神经网络 有机酸 分子结构参数 

分 类 号：O657.8[理学—分析化学]