中国农业气象

• 论文 • 上一篇    下一篇

和田绿洲空气相对湿度的混沌神经网络预测模型

张高锋;黄领梅;沈冰;张晓伟;秦胜英;   

  1. 西安理工大学西北水资源与环境生态教育部重点实验室;和田河管理局;
  • 出版日期:2008-06-10 发布日期:2008-06-10
  • 基金资助:
    国家自然科学基金(50779052)

Chaotic Neural Network Model for Predicting Relative Humidity in Hotan of Xinjiang Autonomous Region

ZHANG Gao-feng1,2,HUANG Ling-mei1,SHEN Bing1,ZHANG Xiao-wei1,QIN Sheng-ying3(1.Key Lab of Northwest Water Resources and Environmental Ecology,Ministry of Education,Xi'an University of Technology,Xi'an 710048,China; 2.Shaanxi Modern Architecture Design and Research Institute;3.Administration Bureau of Hotan River)   

  • Online:2008-06-10 Published:2008-06-10

摘要: 针对用单一指标判断时间序列的混沌特性的不足,本文应用Hurst指数、Lyapunov指数以及饱和关联维数从不同的角度对和田绿洲空气相对湿度的混沌特性进行了识别。在此基础上将混沌理论与神经网络相结合,建立了混沌神经网络预测模型,利用此模型分别对1954-2002年和2003-2004年和田河流域月平均相对湿度进行模拟和预测,其平均相对误差分别为2.96%和0.85%,表明模型具有较高的精度。

关键词: 相对湿度, Hurst指数, Lyapunov指数, 饱和关联维数, 混沌神经网络

Abstract: According to the insufficiency of a single index in identifying the chaotic character,the Hurst exponent,lyapunov exponent and saturated correlation dimensions were used to identify the chaotic characteristics of the relative humidity in Hotan.The chaotic neural network model was established.The monthly average relative humidity for the period of 1954-2002 was simulated and predicted for the period of 2003-2004.The average relative error was 2.96% for the simulation and 0.85% for the prediction respectively.The validation results indicated that the model had a relative high accuracy.

Key words: Relative humidity, Relative humidity, Hurst exponent, Lyapunov exponent, Saturated correlation dimensions, Chaotic neural network