Bayesian dynamic linear models for structural health monitoring

In several countries, infrastructure is in poor condition, and this situation is bound to remain prevalent for the years to come. A promising solution for mitigating the risks posed by ageing infrastructure is to have arrays of sensors for performing, in real time, structural health monitoring across populations of structures. This paper presents a Bayesian dynamic linear model framework for modeling the time-dependent responses of structures and external effects by breaking it into components. The specific contributions of this paper are to provide (a) a formulation for simultaneously estimating the hidden states of structural responses as well as the external effects it depends on, for example, temperature and loading, (b) a state estimation formulation that is robust toward the errors caused by numerical inaccuracies, (c) an efficient way for learning the model parameters, and (d) a formulation for handling nonuniform time steps.

Uncontrolled Keywords

• Bayesian models
• bridge
• dynamic linear models
• infrastructure
• Kalman filter
• structural health monitoring (SHM)