An Alarm Suppression Method Based on Distributed Clustering of Historical Alarms

Abstract

The invention discloses an alarming restraining method based on distributed clustering of historical alarming. The alarming restraining method comprises: performing clustering distributed storage on a large number of historical alarming data according to alarming points, and designing a corresponding indexing main key according to structure features of the alarming data; selecting clustering center alarming points with a pre-set quantity, and separately carrying out parallel calculation on the comprehensive similarity of each locally stored alarming point so as to determine clustering groups of the alarming points; separately calculating clustering scores of the local clustering groups and carrying out result collection on the clustering scores; taking one group with the best clustering effect as a final clustering result; and selecting alarming with the highest priority of different alarming point groups from an alarming list to be displayed. According to the alarming restraining method, the parallel calculation is realized through a distributed manner; time consumed in an alarming clustering analysis process can be remarkably shortened, and the alarming analysis efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of process industry alarm management, in particular to an alarm suppression method based on historical alarm distributed clustering. Background technique [0002] In the "alarm flood" state in the process industry, a large number of alarms generated in real time and displayed in the alarm list are likely to overwhelm the really important and urgent alarms, and the operator does not have enough time and energy to deal with each alarm perfectly. May cause production accidents. [0003] One of the important technologies to reduce the number of alarms displayed in the alarm list of the operator station is alarm suppression, and the rule generation of alarm suppression is mainly obtained through the operation experience or the machine learning process based on historical data. One of the main machine learning technologies in the field of alarm analysis is alarm cluster analysis, which uses clustering methods to ...

AI Technical Summary

Problems solved by technology

[0004] Due to the increase in the production scale of enterprises and the sharp increase in the number of historical alarms, the time-consuming process of alarm clustering analysis has increased significantly. The alarm analysis based on a single machine can no longer meet the application requirements, and corresponding technologies are urgently needed to improve the efficiency of alarm analysis.

Images