一种基于滑动窗口的时间序列异常检测算法

AN OUTLIER DETECTION ALGORITHM OF TIME SERIES BASED ON SLIDING WINDOW

时间序列的异常检测的应用越来越广泛,本文是讨论在基于分段线性的FKD时间序列模式表示基础上时间序列的异常检测。文中提出了一种基于滑动窗口的时间序列模式偏离和窗口异常度的概念,并在此基础上提出了基于滑动窗口的时间序列模式异常的检测算法。通过实验证明了该算法是合理的、有效的。

With the background of the extensive application of the outlier detection on time series,this paper discusses the outlier detection using FKD time series pattern based on piece-wise linear.This article also proposes a concept of the deviation of time series pattern and window abnormal degree base on sliding window and a detection algorithm accordingly.The related experiment showed that this detection algorithm is reasonable and effective.