基于正交实验和最小二乘支持向量机的泡沫复合体系配方预测研究  

Research on forecast method of foam composite system based on quasi-orthogonal experiment and support vector machines

作　　者：刘鸿博[1,2] 罗陶涛[1] 

机构地区：[1]成都理工大学能源学院,四川成都610059 [2]四川水利职业技术学院,四川都江堰611830

出　　处：《应用化工》2010年第5期713-715,719,共4页Applied Chemical Industry

摘　　要：将正交实验与最小二乘支持向量机结合,形成复合预测新方法,以泡沫复合体系配方为研究对象,将聚合物和表面活性剂等影响因素进行综合考虑,对影响因素的定量化评价、实验步骤和预测方法进行详细说明。应用实例表明,该方法能较为全面的反映泡沫复合体系配方的影响因素,避免历史数据中的不确定因素所带来的干扰,有效的解决预测模型的选择、非线性、维数灾难和计算时间长等问题。A new method of forecasting was formed by combined with orthogonal test and least squares support vector machine,considerating the factors of polymer,surfactants and alkali for foam formula.The affecting factors were evaluated,testing procedures and the forecasting methods were introduced.Application showed that the method can be more fully reflected foam composites formula factors.The method can avoid the uncertainly historical data caused by interference and effective solute the choice of forecasting models,non-linear,the problem of dimensionality curse and computing long time.

关 键 词：泡沫复合体系 正交实验 最小二乘支持向量机 方差分析 

分 类 号：O212.6[理学—概率论与数理统计] TH12[理学—数学]