IDENTIFICATION OF FAULTS IN A LEVEL CONTROL DYNAMIC

Identification of faults is an important task in system safety analysis. The large number of system scenarios considered in a realistic safety assessment need to be post-processed to identify the system critical states. The task can be quite complex when the dynamic aspects of the system behavior and its control play a relevant role in the analysis. For example, a fault event may lead to different scenario evolutions depending on the time of failure occurrence and the corresponding state of the controlled process variables. In this work, a Fuzzy C-Means clustering algorithm is applied for the classification of the fault scenarios of a level control system. The classification is based on information from both the stochastic events sequence and the patterns of the process variables evolution