求解关联维数的快速算法研究  

Research of the Optimized Algorithm for the Calculation of Correlation Integrals

作　　者：周越[1] 杨杰[1] 

机构地区：[1]上海交通大学图像处理与模式识别研究所,上海200030

出　　处：《电子学报》2002年第10期1526-1529,共4页Acta Electronica Sinica

基　　金：国家自然科学基金 (No 30 1 70 2 74;50 1 740 38)

摘　　要：针对关联维数计算速度慢、耗时多和运算效率低的缺陷 ,本文从算法和结构两个方面研究了关联维数的快速求解方法 ,提出了区间阈值累加法和区间维值累加法 .这两种算法通过对 [1×mmax]·[1×rmax]表格的批处理填写以达到减少重复运算和提高计算效率的要求 .文中详细地介绍了算法步骤 ,分析了算法的运算量 ,通过与标准算法和一般改进算法的比对 ,显示出本文提出算法的优越性 .In order to solve the problem of slow speed, based on the canonical algorithm of calculation of the correlation integrals, this paper not only provides several improvements, but also puts forward two optimized algorithms, which are based on the region of embedded dimension and the region of distance threshold. The performance of the optimized algorithms are compared with that of canonical algorithm and improvement method. The experimental results show that the optimized algorithms need least time to calculate the correlation dimension for same length time series with same number of embedded dimension and distance threshold, since they have little redundant calculation and enhance the calculation efficiency obviously.

关 键 词：关联维数 维值区间累加法 阈值区间累加法 混沌分形信号处理 

分 类 号：TN911.7[电子电信—通信与信息系统]