混沌时间序列的2阶预测模型  

2-order Prediction Model of Chaotic Time Series

作　　者：张春涛[1] 关文忠[2] 刘学飞[1] 

机构地区：[1]重庆三峡学院数学与统计学院,重庆万州404100 [2]重庆三峡学院经济与管理学院,重庆万州404100

出　　处：《数学的实践与认识》2016年第5期186-193,共8页Mathematics in Practice and Theory

基　　金：重庆市教委科技项目(KJ111106)

摘　　要：为了克服传统预测方法对混沌时间序列预测精度不高的缺点,提出一种新的基于1阶预测模型(1-OP)和信息融合理论的混沌时间序列2阶预测模型(2-OP).首先根据相空间重构理论建立2个1阶预测模型,然后根据融合估计原理建立2阶预测模型.最后利用Lorenz和Mackey-Glass时间序列对该模型进行验证,结果表明,2阶预测模型对多变量和单变量混沌系统都是有效的.It introduces a novel 2-order prediction model（2-OP）for chaotic time series based on 1-order prediction model（1-OP） and information fusion theory,in order to overcome the prediction accuracy disavantage of traditional forecasting methods.First,two 1-order prediction models are established based on the Takens theorem,then,2-order prediction model is established according to fusion estimation principle.Finally,the Lorenz equations and the Mackey-Glass equation are applied to test this model,and the results show that the 2-order prediction model is effective for univariate and multivariate chaotic systems.

关 键 词：混沌时间序列 2阶预测模型 神经网络 

分 类 号：O415.5[理学—理论物理]