ABSTRACT
In this study, a probabilistic model updating approach is proposed for multiple cracks characterisation in beam-like structures. Although model-based approach is able to achieve crack characterisation, it has difficulty in determining the number of cracks for multiple cracks situation. The proposed method addresses this difficulty by dividing the damage identification process into two stages. In Stage 1, the number of cracks is determined using the Bayesian model class section method. The crack parameters are then identified in the Stage 2 using Bayesian statistical framework. Instead of pinpointing the crack location, depth and width, the Stage 2 of the proposed probabilistic approach also quantifies the uncertainties associated with the damage identification results, which is particularly important for making decisions about necessary remedial work.
