On health-state transition models for risk-based structural health monitoring

A desire for informed decision-making regarding the operation and maintenance of structures provides motivation for the development and implementation of structural health monitoring (SHM) systems. One approach to decision-making in SHM is to adopt a risk-based framework in which failure events and decidable actions are attributed costs/utilities. Optimal maintenance strategies may be pursued by considering the probability of occurrence of future failure events in conjunction with associated costs. In order to forecast future failure events, a probabilistic model that describes the degradation of the structure over time is required; in the state-space formulation of risk-based SHM, this model is equivalent to the transition probabilities from possible current health-states of the structure to future health-states.

The current paper aims to demonstrate how such models may be determined using information gathered during the operational evaluation stage of the structural health monitoring paradigm. This information may include knowledge of the operational and environmental conditions under which the structure will operate, in addition to initial physics-based modelling of the structure. A probabilistic transition model describing the degradation of a four-bay truss is developed here, with finite element simulation used to yield knowledge of the load paths within the structure when it is in differing health-states. The paper concludes with a discussion of the importance of probabilistic degradation models within SHM decision-making. The discussion highlights the challenges that arise due to the lack of data available prior to the implementation of an SHM system and suggests for how these may be overcome.