改进的时间序列关联维数快速算法  

Improved fast algorithm to calculate correlation dimension of time series

作　　者：王基[1] 杨爱波[1] 刘树勇[1] 郑建华[1] 

机构地区：[1]海军工程大学船舶与动力学院,湖北武汉430033

出　　处：《华中科技大学学报（自然科学版）》2012年第10期101-104,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(51179197);上海交通大学海洋工程国家重点实验室开放课题资助项目(1009)

摘　　要：针对传统关联维数的计算方法耗时量过大的问题,通过改进点对距离的度量方法,采用空间分块策略技术对重构相空间进行分块并将每个网格进行统一编号,加快了点对的搜索速度,实现了关联积分的快速计算,从而较大程度地提高了关联维数的计算速度.仿真结果表明:提出的算法可以快速有效地计算时间序列的关联维数,为工程实际应用奠定了基础.As traditional algorithms for calculating correlation dimension consume the time greatly, the measured method of distance between points was improved. The space block strategy of dividing the reconstructed phase space into small spaces was used, and every subspace was marked with a u- nique identifier, all of which lead to the searching speed and the calculating speed of correlation inte- gral improved greatly and thus the calculating speed of correlation dimension improved. Simulation re- sults show that the correlation dimension is calculated effectively, which settles the basis of the corre- lation dimension being applied in engineering.

关 键 词：关联维 混沌 相空间 时间序列 空间分块 网格 

分 类 号：TN710[电子电信—电路与系统] O415.5[理学—理论物理]