Modelling a large scale system for risk assessment

Abstract

The aim of this communication is to present an earlier study of how to structure modelling process of complex and large scale systems for risk assessment and management purpose. The approach, in a first stage, uses ontology paradigm to determine variables (or concepts) characterizing system locally and the nature of relationships relating them and, in a second stage, object oriented Bayesian network (OOBN) to characterize the strength of these relationships in terms of conditional probabilities tables given that one of main feature of complex systems is the uncertainty that affect the relationships between different variables. A case study in the domain of risk assessment of flash floods effect on the infrastructures inoperability is considered to show potential applicability of the developed approach.