Probabilistic Damage Detection and Quantification Using a Simulation-Based Approach

This paper proposes a simulation-based approach for probabilistic damage detection and quantification using vibration-based measurements. The method uses an Approximate Bayesian Computation (ABC) algorithm in which a Nested Sampling (NS) technique is introduced, which has the added advantage of guiding the parameters towards region(s) of high posterior density, especially in the first ABC-NS round. The ABC-NS algorithm gives the whole probability density function (PDF) of structural damage severities and a full characterisation of the posterior uncertainties in a reasonable computational time. Two numerical examples with different complexities were proposed to demonstrate the efficiency and the robustness of the ABC-NS sampler in terms of damage localisation and severity quantification.