中国物理B ›› 2004, Vol. 13 ›› Issue (5): 633-636.doi: 10.1088/1009-1963/13/5/012

• GENERAL • 上一篇    下一篇

Determining the minimum embedding dimension of nonlinear time series based on prediction method

卞春华, 宁新宝   

  1. State Key Laboratory of Modern Acoustics, Institute for Biomedical Electronic Engineering, Nanjing University, Nanjing 210093, China
  • 收稿日期:2003-08-12 修回日期:2003-09-24 出版日期:2005-07-06 发布日期:2005-07-06

Determining the minimum embedding dimension of nonlinear time series based on prediction method

Bian Chun-Hua (卞春华), Ning Xin-Bao (宁新宝)   

  1. State Key Laboratory of Modern Acoustics, Institute for Biomedical Electronic Engineering, Nanjing University, Nanjing 210093, China
  • Received:2003-08-12 Revised:2003-09-24 Online:2005-07-06 Published:2005-07-06

PDF (PC)

578

赞

摘要： Determining the embedding dimension of nonlinear time series plays an important role in the reconstruction of nonlinear dynamics. The paper first summarizes the current methods for determining the embedding dimension. Then, inspired by the fact that the optimum modelling dimension of nonlinear autoregressive (NAR) prediction model can characterize the embedding feature of the dynamics, the paper presents a new idea that the optimum modelling dimension of the NAR model can be taken as the minimum embedding dimension. Some validation examples and results are given and the present method shows its advantage for short data series.

Abstract: Determining the embedding dimension of nonlinear time series plays an important role in the reconstruction of nonlinear dynamics. The paper first summarizes the current methods for determining the embedding dimension. Then, inspired by the fact that the optimum modelling dimension of nonlinear autoregressive (NAR) prediction model can characterize the embedding feature of the dynamics, the paper presents a new idea that the optimum modelling dimension of the NAR model can be taken as the minimum embedding dimension. Some validation examples and results are given and the present method shows its advantage for short data series.

Key words: nonlinear time series, embedding dimension, NAR model, prediction

中图分类号:  (Time series analysis)

  • 05.45.Tp
05.45.Pq (Numerical simulations of chaotic systems)