Implementation of Decision Analysis on a Structural Health Monitoring System Applied to a Bridge Benchmark Study

Abstract

Structural health monitoring (SHM) of bridges addresses the need for efficient and cost-effective monitoring of structures. The purpose of SHM is to detect structural damage and provide information related to maintenance, inspection, and repair. This information would then be supplied to asset managers in order to make improved maintenance decisions. This paper presents an investigation into the connection between SHM and decision-making via Bayesian decision theory and the value of information (VoI) obtained from SHM. The results and properties of an SHM damage classifier are used to update the probabilities of a tree-based decision model. Two analyses are included. Firstly, a sensitivity analysis of the value of information (VoI) of an SHM system for varying cost ratios is performed. The SHM system was applied to a benchmark study in which the dynamic behaviour of a steel bridge was measured for both undamaged and damaged structural conditions. Secondly, the relationship between the expected costs of the available actions and the size of the sensor network of the SHM system is obtained by selecting sub-sets from the complete measurement system utilized in the benchmark system. The applicability of SHM results for informing and updating damage probabilities is demonstrated.