基于关键点的小波提取时间序列近似表示方法  

A Wavelet Extraction Time Series Approximation Methods Based on Key Point

作　　者：韩波[1] 章荣丽[1] 鱼先锋[1] 

机构地区：[1]商洛学院数学与计算机应用学院,陕西商洛726000

出　　处：《商洛学院学报》2015年第4期29-32,39,共5页Journal of Shangluo University

基　　金：陕西省教育厅专项科研计划项目(12JK0950)

摘　　要：为了克服常见的时间序列近似表示方法参数精确控制不准确而带来的源序列数据表示减损问题,提出一种基于关键点技术的小波变换分解近似表示方法。来自于不同时间序列的仿真实验表明:相比已有的近似表示方法,它保留了原有表示方法的尺度近似系数,摆脱了用户对参数的精确操作控制,在保留住了时序数据的主要特征的同时还极大地实现了时序数据维度的有效约简。In order to overcome the problem that the source series date control was not accurate, which is usually seen in the common approximate representation methods, an approximate wavelet transform decomposition representation method based on the key points technology with the existing approximate representation, the method retains the approximate coefficient of the scale of the existing approximate representation. The method gets rid of the user precise operation on the parameters control. And the approximation wavelet coefficient of the approximate presentation scale of the key point sequence is given, which can not only keeping the main features of time-series data, but also can greatly realize the effective reduction of the time-series data dimension.

关 键 词：时序数据 数据维度 小波变换 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]