Real-Time Bayesian Damage Detection for Uncertain Dynamical Systems

Abstract

This paper introduces a new Bayesian state-estimation methodology based on stochastic simulation of damage detection for nonlinear structural systems with non-Gaussian uncertainties. The new method uses a linear system with Gaussian uncertainties to build up an importance sampling probability density function (PDF). Samples are taken from the importance sampling PDF to estimate the state of the nonlinear system. The sampled system state can then be used to detect and assess structural and non-structural damage through fragility functions. We demonstrate the consistency of the new methodology using a numerical example and apply the new technique to a real-data case study for damage detection. It is concluded that the proposed method should be useful for real-time damage detection.