摘要
本文提出了一种无限长时间序列的分段线性拟合(Infinite Time Series-Piecewice Linear Fitting,简称ITS-PLF)算法,该算法根据关键点保持时间段的统计特性,确定选择关键点的区间范围;若极值点的保持时间段不在区间范围,则根据包含极值点的连续三个时间数据之间的夹角与筛选角度之间的关系,判断该极值点成为关键点的可能性.实验表明,ITS-PLF算法的执行不依赖于时间序列长度及领域知识,可以有效识别关键点,并可根据数据压缩率的变化实现自适应拟合.
In order to resolving the problem of depending on the length of time series and domain knowledge of traditional PLF algorithm, we proposed a Piecewise Linear Fitting algorithm for Infinite Time Series （ ITS_ PLF）. To determine the interval of Key Points selecting, the statistical attributes of maintaining time of these Key Points was considered. If the maintaining time of a Extreme Point beyond the selection interval, the relation between the threshold angle and the angle of three consecutive data points containing the Extreme Point was selected to determine whether the Extreme Point was a Key Point or not. The experimental results show that ITS _ PLF algorithm does not depend on the length of time series and domain knowledge, can effectively identify the Key Point and adaptively fit the time series according to the changing of the data compression ratio.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2010年第2期443-448,共6页
Acta Electronica Sinica
基金
国家自然科学基金(No.50674086)
中国矿业大学青年科研基金(No.2008A041)
关键词
时间序列
分段线性拟合
压缩率
time series
piecewise linear fitting
compression ratio
