Design of multivariate alarm systems based on online calculation of variational directions

Alarm systems are critically important for safety and efficiency of industrial plants, but are severely suffering from alarm overloading. This paper proposes a method to design a multivariate alarm system based on variational directions of involved process variables, in order to alleviate the severity of alarm overloading. By using adaptive time scales, time gradients of signals are extracted to calculate variational directions in an online manner. Adaptive time scales are determined from a function between time scales and volatilities of process variables. Alarms arise at the moment that a nominal relationship of variational directions among process variables is invalidated. Numerical and industrial examples illustrate the effectiveness of the proposed method.