Title:
METHOD AND DEVICE FOR IDENTIFYING TIME SERIES ABNORMAL PATTERN BASED ON FUZZY MATCHING
Document Type and Number:
WIPO Patent Application WO/2021/012315
Kind Code:
A1
Abstract:
A method and device for identifying a time series abnormal pattern based on fuzzy matching. The method comprises the following steps: performing missing value processing, white noise processing, and outlier processing on an original time series to obtain a time series satisfying a preset condition (S101); detecting a turning point in the time series satisfying the preset condition to divide the time series satisfying the preset condition into a plurality of segmented sub-series (S102); and searching for the obtained sequence segments according to start and end points of each segmented sub-series, and matching by means of a DTW algorithm to obtain the matching result (S103). According to the method, all abnormal patterns are not needed to be defined precisely, pattern fragments of interest can be searched only by few templates, and the purpose of sequential pattern search on the basis of the appropriate scale and the effectiveness can be achieved.
Inventors:
MA XIAOBIN (CN)
DU ZHIHUI (CN)
DU ZHIHUI (CN)
Application Number:
PCT/CN2019/099467
Publication Date:
January 28, 2021
Filing Date:
August 06, 2019
Export Citation:
Assignee:
UNIV TSINGHUA (CN)
International Classes:
G06K9/62
Foreign References:
| CN104820673A | 2015-08-05 | |||
| CN105205112A | 2015-12-30 | |||
| US20160019415A1 | 2016-01-21 | |||
| US20110225196A1 | 2011-09-15 |
Other References:
FENG, HONGWEI : "Fuzzy Set in Time Series Matching", COMPUTER SCIENCE, vol. 29, no. 4, 31 December 2002 (2002-12-31), pages 138 - 140, XP009525601, ISSN: 1002-137X
Attorney, Agent or Firm:
TSINGYIHUA INTELLECTUAL PROPERTY LLC (CN)
Download PDF: