基于小波变换和反馈的时间序列相似模式搜索算法  

Algorithm for Finding Similar Patterns over Time-Series Data Based on Wavelets and Feedback

作　　者：秦吉胜[1] 王淑静[2] 宋瀚涛[1] 

机构地区：[1]北京理工大学信息科学技术学院计算机科学工程系,北京100081 [2]中国航空结算中心,北京100028

出　　处：《北京理工大学学报》2004年第12期1070-1073,1095,共5页Transactions of Beijing Institute of Technology

基　　金：国家部委预研项目(G1998030414)

摘　　要：为得到有价值的相似时间序列,分析小波变换及其在寻找相似时间序列上的优越性;指出了现有的基于小波变换的相似时间序列搜索算法的2个缺点,提出了基于小波变换和加权反馈的时间序列相似模式匹配算法和基于验证深度的验证方法;通过实验证明了算法的有效性和实用性.To obtain valuable similar time-series, the advantage of the wavelet transform and its performance in finding similar patterns over time-series data are analyzed. The disadvantages of existing algorithms based on wavelet transform are summarized. An algorithm for finding similar patterns over time-series data based on wavelets and feedback is put forward, besides the method for validating results based on the depth of verification. The validity and practicability of the algorithm are proved through experiments.

关 键 词：时间序列 相似模式 小波变换 反馈 验证深度 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]