基于关联价格网的产品价格预测方法  

Product price forecasting based on correlative price net

作　　者：严洪森[1,2] 施文武[1,2] 汪峥[1,2] 孟宪刚[1,2] 

机构地区：[1]东南大学自动化学院,南京210096 [2]东南大学复杂工程系统测量与控制教育部重点实验室,南京210096

出　　处：《东南大学学报（自然科学版）》2008年第A02期1-8,共8页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(50875046;60574062);国家高技术研究发展计划(863计划)资助项目(2007AA04Z112)

摘　　要：为实现产品价格影响因素较大波动情况下的价格快速预测,提出了一种关联价格网产品价格预测模型.对引起产品价格变化的主要因素进行了分析,分别以产品、原材料、竞争产品的价格、供需量的变化引起产品价格的变化为基础构建产品的关联价格网.然后,对关联价格网性质进行理论推导,得出可以用产品的关联价格子网替代整个关联价格网对产品价格进行预测的结论,从而简化模型.进而采用神经网络构建一个反映关联价格子网中价格波动因素的预测网络(PFN),并利用神经网络学习和计算方法对产品的价格进行预测.理论分析和仿真实验显示该价格预测方法具有很好的应用价值.A correlative price net forecasting model is presented to implement quick forecasting when the influence factors of product price fluctuate changing abruptly. This paper analyzes the main factors which result in product price change, and establishes correlative price net based on price and supply/demand change of product, materials and competing products which influence price change of product. And then, correlative price sub net is presented by theoretical derivation from correlative price net. Model of correlative price sub net is simple and can forecast in place of correlative price net. Then neural networks are adopted to build up a new network, known as price forecasting net （ PFN）, which detects the changes of factors in the correlative price sub net, and price can be forecasted with the data processing methods of neural networks. The theoretic analysis and simulation experiment show the practical value of the method.

关 键 词：价格预测 自学习 关联价格网 神经网络 

分 类 号：TH201[机械工程—机械制造及自动化]