Studies On An Approach To Process Alarm Event Grouping

Process alarm event grouping is recognized as a core task ofintelligent alarm management systems.Taking advantage of relevanceinformation,process alarm events can be divided into separate groupsmutually independent,which is considered imperative and important foreffective management and utilization of process alarm events.In this paper, a so called Petri-FCM approach to process alarm eventgrouping is explicitly introduced.Therein,a Petri-net based process alarmevent evolution model is established based on both process priorknowledge and historical data.Taking advantage of feature that timeintervals associated with timed Petri time correspond to alarm triggertime, a timed Petri-net alarm event model is well established, which couldbe used to figure out the center of process alarm events as well as tosynchronize the candidate data.Subsequently, Fuzzy C-means Clusteringmethod is employed to group the process alarm events.The proposed approach is applied to TE process,leading to satisfactory results, andshowing benefits of the contribution. In addition,an industrial DMFrecovery process is employed to validate the proposed approaches, givingrise to expected results. Oriented to application,a developed industrialmonitoring alarm system is introduced in details together with keyfunctional modules of the system.It is worth noting that the methods presented in this paper could bequite attractive to dealing with relevant problems associated withindustrial processes.