摘要
针对关联维数计算耗时量大的问题，通过改进点对距离的度量方法，以及采用K-NN技术进行点对的搜索实现了关联和的快速计算，较大程度地提高了关联维数的计算速度。验证表明：对于长度为20480的时间序列，采用快速算法计算关联维数，其耗时量是G-P算法的1／60。
Aiming at the matter of time-consuming in calculating the correlation dimension, by improving the method of measuring the point-point distance, and by using the K-Nearest Neighbor Search technology to calculate the correlation sum, we increase the calculating speed greatly. The actual computation indicates that when the time series length is 20 480, the time consumption of calculating the correlation dimension using the improved algorithm is 1/60 of that of using the G-P algorithm.
出处
《装甲兵工程学院学报》
2007年第6期58-61,共4页
Journal of Academy of Armored Force Engineering
基金
军队科研计划项目
关键词
时间序列
关联维数
G-P算法
K最近邻搜索
time series
correlation dimension
G-P algorithm
K-Nearest Neighbor Search
