Time sequence approximate match-based big data abnormal state detection method and device

Abstract

The invention relates to a time sequence approximate match-based big data abnormal state detection method and system. The method comprises the following steps: partitioning a to-be-detected time sequence into multiple sets according to a data range of the to-be-detected time sequence and a preset partitioning coefficient, and expressing the to-be-detected time sequence through a one-dimensional to-be-detected sequence consisting of numbers of sets where data points are located; expressing a standard time sequence by a one-dimensional standard sequence in the same way; performing Harsh calculation on the one-dimensional to-be-detected sequence and the one-dimensional standard sequence; calculating Jaccard coefficients of the one-dimensional to-be-detected sequence and the one-dimensional standard sequence, and judging that the time sequence with the Jaccard coefficient less than a preset threshold value is a sequence in an abnormal state. According to the time sequence approximate match-based big data abnormal state detection method, by combination of set-based time sequence partitioning and Harsh calculation, the calculation amount of the Jaccard coefficients is reduced; furthermore, the sequence can be further partitioned from coarse to fine, so that the whole calculation speed is guaranteed, and the abnormal state detection precision is also guaranteed.

Description

technical field [0001] The present invention relates to a large data abnormal state detection technology, in particular to a large data abnormal state detection method and device based on time series approximate matching. Background technique [0002] Abnormal state detection is widely used in various fields such as electric power, remote sensing, road and bridge, machinery, etc., and plays an extremely important role in the normal operation of the system. In recent years, due to the large volume and various types of online detection data, people have introduced big data technology into anomaly detection, providing new solutions and ideas for improving the accuracy of system anomaly detection. [0003] In abnormal state detection, time series analysis is a commonly used method. In the prior art, a set-based time series analysis method is proposed. This method first divides the data distributed on the time axis into sets according to certain rules and segmentation coefficien...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to provide a large data abnormal state detection method and device based on time series approximate matching, aiming at the defect that the existing abnormal state detection method increases the amount of data calculation when the accuracy is improved. The combination of set-based time series segmentation and hash segmentation further reduces the dimensionality of data and reduces the amount of computing data, making it more suitable for anomaly detection in big data computing systems

Images