• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2010, Vol. 32 ›› Issue (8): 134-137.doi: 10.3969/j.issn.1007130X.2010.

• 论文 • 上一篇    下一篇

混沌时间序列的Anfis-量子网络预测

蒋文科,谭阳红   

  1. (湖南大学电气与信息工程学院,湖南 长沙 410082)
  • 收稿日期:2009-01-17 修回日期:2009-04-13 出版日期:2010-07-25 发布日期:2010-07-28
  • 作者简介:蒋文科(1970),男,湖南长沙人,工程师,研究方向为智能控制和图像识别;谭阳红,教授,研究方向为智能信号处理、电路测试与诊断。
  • 基金资助:

    国家自然科学基金资助项目(50277010);湖南省自然科学基金资助项目(07JJ6132,04FJ2003,05GK2005);教育部博士基金资助项目(20020532016);高等学校新世纪优秀人才支持计划资助项目(NCET040767)

Chaotic Time Series Forecasting Research Based on the Combination Network

JIANG Wenke,TAN Yanghong   

  1. (School of Electrical and  Information Engineering,Hunan University,Changsha 410082,China)
  • Received:2009-01-17 Revised:2009-04-13 Online:2010-07-25 Published:2010-07-28

摘要:

针对混沌时间序列的解析方程的不确定性,本文提出了Anfis量子组合预测网络。此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,再与量子神经网络组合预测,从而提高预测能力和精度。最后以MackeyGlass混沌时间序列进行实验,结果表明Anfis量子网络具有良好的局部泛化能力,其预测精度明显高于BP神经网络和量子神经网络。

关键词: 混沌时间序列, 量子网络, 组合网络, 预测

Abstract:

A combination forecast network of the Anfisquantum which are suggested in view of the uncertainty of the chaotic time series analysis equation. The qualitative equation and initial condition of the chaotic time series are not related to the network. For improving the ability and accuracy of forecast,the training data is beforehand handled by the network of Anfis, and is then forecasted by the quantum network. The results of the experiment of the chaos time series of the MackeyGlass indicate the Anfisquantum network is characterized by its superior generation ability, and its predictability precision is obviously higher than that of the BP network and that of the quantum network.

Key words: chaotic time series;quantum network;combination network;forecast