供应链需求预测的非线性方法研究  

Nonlinear Method Research on Demand Forefoundry for Supply Chain

作　　者：封云[1] 马军海[1] 

机构地区：[1]天津大学管理学院,天津300072

出　　处：《北京理工大学学报（社会科学版）》2008年第5期82-86,共5页Journal of Beijing Institute of Technology：Social Sciences Edition

摘　　要：文章总结了国内外关于供应链需求预测的研究工作,大致分为线性预测方法和非线性预测方法。非线性预测方法由于可以准确预测实际需求随机波动,已经成为供应链需求预测问题研究的热点。其中以局域法加权一阶预测、最大Lyapunov指数预测和全域法支持向量机预测最为常用。通过实证比较研究,基于相空间重构理论的支持向量机预测方法可以准确预测实际需求的波动趋势,其预测精度和准确度很高。The research on demand forefoundry for supply chain at home and abroad can be roughly divided into linear and nonlinear method. As nonlinear method can accurately forefoundry the random fluctuation of practical demand, it has become a hotspot in research on the demand forefoundry for supply chain. The Local Area Single-Order Weighting Forefoundry, Maximum Lyapunov Index Forefoundry and Macrocosm Support Vector Machine Forefoundry are used most frequently. Through the empirical comparative study, it proves that, the method of Support Vector Machine （SVM）based on the space reconstructive theory can forefoundry the fluctuating trend of practical demand with high degree of accuracy.

关 键 词：供应链 需求预测 非线性 支持向量机 

分 类 号：F252.21[经济管理—国民经济]