Method for detecting abnormal subsequence of single time series

Abstract

The invention provides a method for detecting an abnormal subsequence of a single time series. In the method, abnormal subsequences are redefined and TSMBRB representation sets of all subsequences are extracted to perform clustering operation. According to clustering information obtained, if some subsequence is far away from a clustering center, the subsequence is more likely to be an abnormal subsequence, or if the number of elements in the cluster is less, then it is more likely that the cluster contains an abnormal subsequence. This method adopts a double-layer cycle structure. The outer cycle detects a candidate subsequence, and the inner cycle finds k neighbors of the candidate subsequence. The distance between the candidate subsequence p and the abnormal subsequence position q in the inner cycle is smaller than the abnormality degree of the current abnormal subsequence as early as possible, so that calculation of the candidate subsequence is terminated prematurely. In the method, by analyzing and utilizing the law of data change of adjacent times, a large number of candidate subsequences are reduced, the number of times of distance calculation is greatly reduced and the algorithm efficiency is improved.

Description

technical field [0001] The invention belongs to the technical field of data mining, and in particular relates to a method for detecting abnormal subsequences applicable to multi-field single time series. Background technique [0002] The current research on anomaly detection mainly focuses on the common detection of data points, that is, finding abnormal data points from a large number of unordered data points, without considering the order characteristics of the data. However, in many application fields, sometimes it is not meaningful to study a single data point. Many time series data belong to this category, and it is more meaningful to study several continuous data. Various fields contain a large amount of time-series data, such as ECG data of patients, EEG data, parameter data of a large number of sensors in power plants, network flow data, and so on. The abnormal subsequence (pattern) detection of time series is a very important field. Most of the time series data con...

AI Technical Summary

Problems solved by technology

The comparison-based method has many advantages, but some problems still need to be solved and improved. The original anomaly subsequence definition of the comparison-based method has the disadvantage of not being able to find similar anomalies, and most of the current anomaly subsequence detection algorithms are only suitable for static time series data.

Images