基于多变量支持向量机的供应链需求预测分析  

Supply Chain Demand Forecasting Based on Multivariate Support Vector Machine

作　　者：荆浩 刘垭 唐金环 JING Hao;LIU Ya;TANG Jing-huan(School of Economics and Management,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学经济与管理学院,辽宁沈阳110136

出　　处：《系统工程》2018年第11期121-126,共6页Systems Engineering

基　　金：国家自然科学基金资助项目(71702112);教育部人文社科基金资助项目(18YJC630219);辽宁省自然科学基金资助项目(20170540698)

摘　　要：在供应链中,市场需求往往受到多个因素的影响,单一变量的预测方法已经无法达到预测期望。为了提高预测精度,削弱牛鞭效应,根据Agent理论构建供应链模型,提出一种基于多因素的支持向量机需求预测方法;通过选取主要影响因素,建立多变量SVM预测模型,并利用遗传算法对SVM模型的参数进行优化选择,最后利用算例进行仿真分析。实验结果表明,与单变量支持向量机相比,多变量支持向量机充分考虑影响市场需求的相关因素,最大程度挖掘了数据中的有效信息,提高了供应链需求预测的准确性,从而缓解了供应链中的牛鞭效应。In the supply chain,the market demand is often affected by many factors,the single variable forecasting method has been unable to meet the forecast expectations.In order to improve prediction accuracy and weaken the bullwhip effect,this paper builds a model of supply chain based on Agent theory and proposes a demand forecasting method based on multiple factors.By selecting the main influencing factors,a multivariable SVM prediction model is established,and genetic algorithm is used to optimize the parameters of the SVM model.Finally,an example is used to analyze the simulation results.Experimental results show that,compared with single variable support vector machine,multivariable support vector machine fully consider the impact of market demand related factors,maximize the effective mining of the information,and improve the accuracy of demand forecasting,thus alleviating the bullwhip effect in the supply chain.

关 键 词：供应链需求预测 AGENT系统 多变量支持向量机 

分 类 号：F252[经济管理—国民经济]