用支持向量机预测中药水提液膜分离过程

为了找出中药水提液膜过程中影响膜污染的主要原因和预测膜污染的程度以防止膜污染,研究用支持向量机分类、遗传神经网络于中药水提液膜中属性筛选。以筛选出的主要属性用支持向量机回归建模预测,讨论确定模型参数、模型优化等关键问题,并与神经网络运行结果对比分析。分析结果表明支持向量机回归算法对膜污染度的拟合效果和预测能力均好于对该问题分析的其他方法。

Forecasting membrane separation process of Chinese medicine's water extract with support vector machine

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,we applied support vector classification(SVC)and genetic neural network to select main features,then we applied support vector regression(SVR)with main features to build model and forecast the pollution degree in the process.In addition,parameter selection and model optimization as key issues were fully discussed in this study.Meantime,model built from SVR was compared with artificial neutral network(ANN),thus it can be concluded that SVR is superior to other algorithms in forecasting and fitting membrane pollution degree.