Identification Method of Multiple Sequential Alarms that Occurred Simultaneously in Plant-operation Data

Abstract

Abstract Advances in distributed control systems in the chemical industry has made it possible to inexpensively and easily install numerous alarms in them. A poorly designed alarm system might cause nuisance alarms. One type of nuisance alarm is a sequential alarm, which reduces the capability of operators to cope with plant abnormalities because critical alarms are hidden in them. We propose an identification method of sequential alarms that occurred at the same time. With this method, similarities of all combinations of an alarm subsequence are compared using the Smith-Waterman algorithm (Smith et al ., 1981). We introduced a new objective function considering the time differences between alarms into this algorithm. We applied the proposed method to the simulation data of an extractive distillation column, and the simulation results indicate that the method can extract sequential alarms that occurred simultaneously in plant-operation data.