Model-Based Fault Diagnosis Using a Hybrid Dynamic Simulator: Application to a Chemical Process

Abstract This work presents a fault detection and isolation methodology for the monitoring of Hybrid Dynamic Systems. This methodology rests on a mixed approach, which combines a model-based method for the fault detection and an approach based on data (pattern matching) for the identification of fault(s). This methodology is integrated within the simulation platform PrODHyS , through the development of the module PrODHySAEM . The goal of this paper is to underline the potentialities of our approach for the diagnosis of the system. This methodology is illustrated by the studies of diagnosis problems in the field of Chemical Process System Engineering.