摘要
针对目前常用的特征量有关联维数和近似熵这两个指标在应用中存在不足,提出了一种新的替代数据法对时间序列中的非线性特性进行检测。替代数据法由零假设和检验特征量两部分组成。笔者提出将模糊熵作为特征量引入到替代数据法中检测时间序列的非线性特征,并在Logistic方程产生的非线性时间序列,以及线性AR模型产生的线性时间序列上进行了验证。研究结果表明,对于不同长度的时间序列,基于模糊熵的替代数据法是一种稳定、有效的非线性检测方法。
In this study, a new method of surrogate data for testing nonlinearity is described. Surrogate data is composed of null hypothesis and inspection characteristics. The common characteristics quantity includes correlation dimension and approximate entropy, both of which have some insufficiency in application. Fuzzy entropy is an newly emerged complex measure index and an improved algorithm of approximate entropy. This study introduces fuzzy Entropy as the characterlstlc quan validation on t duced by linea entropy is a st tity into surrogate data, testing the non-linear feature of the time series and making he nonlinear time series produced by Logistic equation and linear time series pror AR model. The result of this study shows that surrogate data based on the fuzzy able and efficient nonlinearity test for the time series of different length.
出处
《太原理工大学学报》
CAS
北大核心
2014年第3期369-371,407,共4页
Journal of Taiyuan University of Technology
基金
国家自然科学基金资助项目(61170136
6137101)
关键词
替代数据
模糊熵
时间序列
非线性检测
surrogate data
fuzzy entropy
time series
nonlinearity test
