Contents 1 Introduction
Modern large-scale industrial facilities, such as the oil extraction, refinery, metallurgy, and power generation plants, are commonly deployed with alarm systems, which perceive abnormalities and notify operators by audible or visible means. However, due to the process complexity and poor alarm rationalization, most real alarm systems suffer from false alarms, chattering alarms, and alarm floods, which overload plant operators and distract them from critical alarms. As reported in [1], the alarm rates in many real industries were much higher than the benchmarks in the industrial standards [2], [3]. As a result, the efficiency of alarm systems was significantly lowered, which compromised process safety and might lead to serious consequences. In the past decade, the research area of industrial alarm monitoring and management has been attracting increasing attentions, and a variety of advanced alarm management techniques have been proposed to improve the performance of alarm systems and handle complex alarm problems.
