FAULT INFERENCE USINGENHANCEDBAYESIAN NETWORKSFOR ABNORMAL SITUATION MANAGEMENT

Nowadays, chemical plants are becoming complex due to high dependency among operational variables. Control loops are interdependent to optimize production. Therefore, the triggered floods of alarms complicate tracking the root fault among different process systems. Nevertheless, the alarm systems could have diverse failures leading to uncertainty in decision-making of Abnormal Situation Management (ASM). For these flooding and reliability issues in alarm systems, Bayesian Networks(BNs)are increasingly employed to model the relationships among the operational variables. However, fault inference using BN has structuring and learning issues for complex systems and little fault history respectively.