混沌时间序列的组合网络预测研究  

Chaotic Time Series Forecasting Based on Conbination Network

作　　者：文娟[1] 谭阳红[1] 龙伯华[1] 周野[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082

出　　处：《计算机仿真》2009年第11期166-169,共4页Computer Simulation

基　　金：国家自然科学基金(50277010);湖南省自然科学基金(07JJ6132;04FJ2003;05GK2005);教育部博士基金(20020532016);高等学校新世纪优秀人才支持计划(NCEF-04-0767)

摘　　要：针对混沌时间序列的解析方程的不确定性,提出了自适应模糊递归网络组合预测网络。网络不依赖混沌时间序列的确定性方程和初始条件,根据给定的训练数据,先采用Anfis网络对训练数据进行处理,然后与Elman网络组合预测,从而提高了预测能力和精度。通过对Mackey2Glass混沌时间序列仿真实验,表明了Anfis-Elman网络增强了局部泛化能力,其预测精度明显高于BP网络和Elamn等网络。In view of the uncertainty of chaotic time series analysis equation a combination forecast network of Antis - Elman is proposed. The qualitative equation and initial condition of chaotic time series are not related to the network. For improving the ability and accuracy of forecast, the training data are first processed by the network of Antis, then forecasted by Elman network. The result from simulated experiment of chaos time series of the Mackey2Glass indicates that the Antis - Elman network has strengthened the part general melt ability, and its prediction precision is higher than that of the BP network and the Elman network obviously.

关 键 词：混沌时间序列 自适应模糊递归网络 组合网络 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]