一种多尺度的时间序列相似模式匹配算法

A Muti scale Similar Pattern Match Approach for Times Series Databases

提出一种多尺度的时间序列相似模式匹配算法 .该算法用离散小波变换对时间序列进行多分辨分析 ,在多尺度上提取序列的形状特征 ,然后在不同的尺度上找出相似的序列和子序列模式 .该算法可以匹配不同长度的序列 。

This paper proposed a multi scale similar pattern match approach for Times Series Databases. Time series is analyzed by discrete wavelet transformation at multi resolution. Shape features of time series is extracted at multi scale. Then Sequence or sub sequence Similar to pattern sequence can be found from raw time series at multi scale. It has capability of matching time series of different lengths. It can support vertical shifting and amplitude scaling transformation.