波动性产品需求预测的神经网络模型  

Neural Network Forecasting Model for the Fluctuating Product Demand

作　　者：赵宏霞[1] 杨皎平[1] 

机构地区：[1]辽宁工程技术大学工商管理学院,葫芦岛125105

出　　处：《价值工程》2008年第8期97-100,共4页Value Engineering

基　　金：辽宁省社科联基金(2007lslktglx-12);辽宁工程技术大学校基金(07A202&)。

摘　　要：随着市场竞争的日益激烈,消费者的心里越来越复杂,这样导致了产品的需求的波动性大大增加。这种强波动性的产品需求序列中除了随机性外还存在混沌性,根据混沌理论可知,混沌的短期预测是可行的。为了有效的对这些混沌性进行预测,选择了神经网络作为预测模型,因为神经网络对非线性具有较好逼近能力。在网络结构选择中考虑了混沌序列的嵌入维数,并在隐层中加入了径向基以更好的拟合数据。在针对目前很多企业具有数据库和数据仓库的背景,给出了基于数据挖掘的具体预测方法,并通过实例演示了预测的有效性。Along with the intense competition in market, the consumer＇s mentality become more and more complicated, it result that the fluctuation of product demand more and more acute. The time series of the fluctuate product demand include the randomness and chaos, according to the chaos theory, the chaos can be forecast short-tenn. How to forecast the chaos action, we can use the neural network because the NN process the excellent fitness ability. In the selection of the structure of NN, the embedding dimensions is considered, in order to fit the data accurately the RBF network is recommend. Base on the fact that many enterprise possess data base and data warehouse, this paper give the forecasting method under the data mining thinking, and demonstrate the effectiveness through example.

关 键 词：混沌 预测 神经网络 数据挖掘 

分 类 号：G303[文化科学] TP183[自动化与计算机技术—控制理论与控制工程]