On an Application of Probabilistic Risk Assessment to Structural Health Monitoring

Abstract

A key motivation for implementing structural health monitoring is to facilitate decisionmaking regarding the operation of a structure throughout its life. The notion of risk has been used to inform decision-making under uncertainty in industries such as nuclear energy and aerospace - formalised in a procedure known as probabilistic risk assessment. The current paper aims to exploit methods used in probabilistic risk assessment to demonstrate a novel risk-based approach to structural health monitoring. The approach utilises a probabilistic graphical model framework in which information is passed from a probabilistic classifier to an influence diagram representing a decision-process via a Bayesian network representation of a fault tree. The risk-based approach is demonstrated on simulated data from a finite element model of a four bay truss.