Alarm deadband design based on the reliable distribution estimation of alarm deviations

Deadbands are widely used in industrial alarm systems to decrease nuisance alarms as much as possible. This paper proposed a way to design deadband widths based on alarm deviations. A key technical issue is to determine an appropriate sample size of alarm deviations, in order to yield a reliable estimation of the probability distribution of alarm deviations. This issue is resolved via Bayesian estimation theory, and an optimal deadband width is designed by exploiting historical data samples in normal operating conditions. Industrial examples are provided to validate the proposed method.