A Bayesian approach for controlling structural displacements

Bayesian Networks represent one of the most powerful and effective tools for knowledge acquisition in the observation of physical phenomena affected by randomness and uncertainties. The methodology is the result of several developments concerning the Bayesian statistical theory and permits, by inference, to update the statistics describing physical variables by the observation of experimental evidences. In general, Bayesian Networks have become a very popular and versatile approach in problem solving strategies because of their capability in enhancing the status of knowledge of a physical problem domain and to characterize expected outcomes. In particular, this work presents a strategy performing the Bayesian updating of the mechanical and geometrical properties of a steel structure. Based on high-precision topographical measurements, such a strategy has the purpose of accurately estimating the structural displacements expected during the structural life-cycle.