Alarm Rationalization Based on Process Mining Techniques

Industries in general need a reliable system for fault identification and alarm management. The high incidence of alarms can overload the operator, exposing him to conditions that may exceed their ability to perform effective actions, impairing his performance during his workday. In this context, the present paper proposes an approach for alarms rationalization based on process mining techniques. The alarm rationalization is in accordance to the alarm lifecycle management model suggested by ANSI / ISA-18 standard. The aim of this paper is to improve the alarm system through its rationalization, allowing an adequate organization and data layout of the whole set of alarms. Thus, it is expected a better interpretation and understanding of the industrial process. Performance metrics recommended by ANSI / ISA-18 standard are used. Such metrics are used for analyzing a database of alarms coming from an industrial plant. Preliminaries results have demonstrated the feasibility of the present approach. The results show that the use of the process mining technique can provide support on rationalization alarms, standing for a promising method in alarm management domain.