基于计算机化学方法的中药膜过程研究  

Study on membrane separation process of Chinese medicine's water extraction based on the method of computer chemistry

作　　者：郭立玮[1] 李玲娟[1,2] 董洁[1] 

机构地区：[1]南京中医药大学中药复方分离工程重点实验室,南京210029 [2]南京中医药大学信息技术学院,南京210046

出　　处：《膜科学与技术》2011年第3期196-204,共9页Membrane Science and Technology

基　　金：国家自然科学基金资助项目(30171161;30572374;30873449)

摘　　要：中药膜分离过程是一个非线性系统,其工艺数据具有多变量、非线性、强噪声、自变量相关、非正态分布、非均匀分布等全部或部分特征.为了探索中药复杂体系膜过程优化原理与方法,提出使用神经网络、支持向量机等计算机化学方法建立预测模型,并开发实现集成化的综合分析和预测系统.结果表明,该集成化的综合分析系统较单一分析建模预测精度更高.同时介绍了定量结构-活性/性质相关性(QSAR/QSPR)研究的主要步骤及采用该方法开展的"超滤膜对生物碱类等物质的透过/截留及其定量结构关系的研究".The membrane separation process of Chinese medicinels water extraction is a non-linear system and its process data have the features of multi-variable, non-linear, strong noise, independent varieties relat ed to each other,abnormal and nonuniform distribution etc. In order to research the principles and methods of membrane separation process in the complex system of Chinese medicine, computer chemistry methods of artificial neutral network, support vector machine are proposed to build up a forecasting model in this paper, and integrated analysis and forecasting system was developed. The result shown that this integrated analysis system is more accurate than single analysis modeling in forecasting. At the same time, the main steps of quantitative structure-activity/property relationships（QSAR/QSPR） and the study on transmission rate and QSPR of alkaloids with UF membranes were introduced.

关 键 词：中药 膜分离过程 人工神经网络 支持向量机 分子结构参数 

分 类 号：TQ461[化学工程—制药化工]