基于特征提取的中药水提液膜分离预测系统  

Forecasting system of Chinese medicine’s water extract in membrane separation process based on feature selection

作　　者：李玲娟[1] 郭立玮[2] 

机构地区：[1]南京中医药大学信息技术学院,江苏南京210046 [2]南京中医药大学中医药研究院,江苏南京210046

出　　处：《计算机工程与设计》2010年第9期2023-2026,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(30572374);江苏省高校自然科学研究指导性计划基金项目(06KJD520144)

摘　　要：为了找出中药水提液膜过程中影响膜污染的主要因素和预测膜污染的程度以防止膜污染,提出了应用遗传神经网络提取影响中药水提液膜分离过程的特征因素的方法,并以特征因素为输入向量,使用神经网络、支持向量机等建立预测模型,开发并实现了集成化的综合分析和预测系统。介绍了中药水提液膜分离预测系统的体系结构、主要功能、运行情况及开发的关键技术。实验结果表明,该集成化的综合分析系统较单一分析建模预测精度更高。With the purpose of finding the critical factors which exert great influences on membrane pollution in the filtration of Chinese medicine’s water extract and preventing membrane pollution by forecasting the pollution degree, an account of how to select features in membrane separation process by genetic algorithm and artificial neutral network, and how to build up a forecasting model by artificial neutral network and support vector machine, taking selected features as input vector. With characterization data of membrane separation of Chinese medicine’s water extract, by means of module development and system integration. Aimed at developing a forecasting system whose systematic structure, main function, operation situation and key developing technology of are also discussed, which can predict and analyze membrane fouling in membrane separation process of Chinese medicine’s water extract. Besides, the finding indicated that this integrated analysis system is superior to the mono-analysis in forecasting accuracy of modeling building.

关 键 词：特征提取 遗传算法 神经网络 支持向量机 模型预测系统 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]