Optimal alarm threshold under time-varying operating conditions

Abstract

Alarm is an important method to detect the abnormal situations in modern industrial plants. Reducing the number of false alarms and missed alarms is significant for improving the performance of the alarm systems. In this paper, the optimal alarm threshold under time-varying operating condition has been studied. An alarm system model is introduced first, then the computational method of optimal alarm threshold without any data processing techniques is discussed. Moving average filter is then used to reduce the impact of measurement noise and its effect on the threshold design is further explained. An alarm design procedure based on these analysis is presented for the fast computation of alarm threshold. An example is provided to illustrate the effectiveness of the proposed method.