基于核主成分-最小二乘支持向量机的区域物流需求预测  

Forecast of Regional Logistic Demand based on KPCA-LSSVM

作　　者：梁毅刚[1] 耿立艳[1] 张占福 

机构地区：[1]石家庄铁道大学经济管理学院,河北石家庄050043 [2]石家庄铁道大学四方学院,河北石家庄051132

出　　处：《铁道运输与经济》2012年第11期63-67,共5页Railway Transport and Economy

基　　金：河北省社会科学基金项目(HB12YJ035);教育部人文社会科学研究青年基金项目(11YJC790048);国家软科学研究计划项目(2010GXQ5D320)

摘　　要：概述区域物流需求预测方法,分别阐明核主成分分析(KPCA)和最小二乘支持向量机(LSSVM)模型的原理,提出将核主成分分析(KPCA)与最小二乘支持向量机(LSSVM)相结合,建立核主成分-最小二乘支持向量机(KPCA-LSSVM)预测模型。先利用KPCA对数据进行预处理,消除变量之间的相关性,提取非线性主成分,再通过LSSVM对提取的非线性主成分进行训练,建立预测模型。最后,通过实例验证比较LSSVM与KPCA-LSSVM两种模型的预测性能。结果表明,KPCA-LSSVM的预测精度较LSSVM明显提高,是一种有效的中短期区域物流需求预测方法。This paper summarizes regional logistic demand,expounds the principle of kernel principal component analysis（KPCA） and least squares support vector machine（LSSVN）,puts forward that the KPCA could be combined with LSSVM,and the forecast model of KPCALSSVM could be established.Firstly,making data pretreatment by KPCA,eliminating the relativity between variable and distilling nonlinearity principal component,then training the distilled principal component by LSSVM and establishing forecast model.In the end,the forecast function of LSSVM and KPCA-LSSVM models is validated and compared by examples.The result shows the forecast precision of KPCA-LSSVM is obviously higher than LSSVM,and the KPCA-LSSVM model is a effective forecast method of regional logistic demand in middle-term and short-term.

关 键 词：区域物流 需求预测 最小二乘支持向量机 核主成分分析 

分 类 号：F252.21[经济管理—国民经济]