ABSTRACT
Structural health monitoring is comprised of different stages, which range from the optimisation of a structural system to the provision of future predictions about its integrity. In this chapter, both the localisation of damage and the optimisation of a structural health monitoring system are rigorously addressed and illustrated for ultrasonic guided-waves. The complexities related to the measurement and post-processing process of the ultrasonic signals lead to uncertainties that need to be quantified for unbiased diagnostics and fully informed decision-making. The framework for damage localisation consists of two hierarchical Bayesian inverse problems, whereby both model class selection of several candidate signal post-processing models and the damage position inference are carried out simultaneously. Furthermore, the optimal sensor configuration (number and location) is addressed by maximising the value instead of the amount of information, whereby a trade-off between cost and damage-related information is obtained. Several examples show the rigour and accuracy of these Bayesian methods for structural health monitoring.
