This paper formulates the basis for a theory of optimising the structural health monitoring (SHM) system topologies based on maximising the value of information an SHM system can deliver. The value of SHM information is in overall reduction of the expected risk and is calculated using the pre-posterior Bayesian analysis. This requires, inter alia, the prior probabilities of system survival/failure and likelihoods of survival/failure state indication by the monitoring system. The paper thus starts with some basic concepts from the structural reliability theory which point out to the fact that system survival/failure probabilities are functions of local member- or cross-section-level survival/failure probabilities, and that the latter can be updated using SHM data. Then, the pre-posterior Bayesian decision tree analysis used to minimise the total risk is introduced including the probabilistic modelling of the SHM system performance and considering the probabilities of state mis-indication errors. Finally, the sensing system topology optimisation problem is stated mathematically. A simple analytical example using a truss structure with strain gauge-based SHM system rounds up the paper and illustrates the concepts discussed.
|Title of host publication||Proceedings of the 4th Conference on Smart Monitoring, Assessment and Rehabilitation of Civil Structures SMAR 2017|
|Number of pages||9|
|Publication status||Published - 2017|
|Event||SMAR 2017: 4th International Conference on Smart Monitoring, Assessment and Rehabilitation of Civil Structures - ETH Zurich, Zurich, Switzerland|
Duration: 13 Sep 2017 → 15 Sep 2017
|Period||13/09/17 → 15/09/17|