This article proposes a novel framework for optimal sensor placement (OSP) that explicitly maximizes the probability of achieving structural health monitoring (SHM) objectives while accounting for modeling uncertainties and measurement noise. The framework consists of four main stages: (1) generating numerical models with varying structural parameters and performing statistical structural analysis using Monte Carlo simulations, (2) introducing simulated measurement noise to reflect real-world data contamination, (3) evaluating the probability of achieving the SHM objective for a subset of candidate sensor layouts and (4) training a regression model to predict the optimum sensor layout that maximizes the likelihood of a successful application. The optimal sensor layout selected thus ensures robust SHM performance under uncertainty. Unlike existing OSP methods, which typically rely on deterministic models and fixed optimization criteria, the proposed approach explicitly considers both modeling variability and measurement inaccuracies. By integrating uncertainty quantification into the optimization process, this framework provides a more reliable basis for sensor placement decisions. It is also adaptable to different SHM objectives, such as mode shape identification or damage detection, offering a flexible, data-driven methodology for real-world applications. Through its probabilistic approach, the proposed framework advances OSP strategies, enhancing the effectiveness of SHM systems in bridge monitoring and beyond.
Erduran et al. (Thu,) studied this question.