.png){kind=icon}
.png){kind=icon}
Return
Simulation-based method for probabilistic damage detection and quantification using modal measurements
.png){kind=icon}
OA
AI
DOI：10.1016/j.istruc.2026.111527.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a novel simulation-based method rooted in Bayesian statistics that can be used to detect and estimate structural damage based on changes in dynamic characteristics. The proposed method uses Approximate Bayesian Computation based on a Nested Sampling (ABC-NS) technique to make inferences. The ABC-NS algorithm relies on numerical simulations to minimise the discrepancy between the modal data determined experimentally and their theoretical counterparts obtained from a simulator. By definition, a simulator is a computer program which could be for instance a computational model built by using finite element method, an analytical model or an ordinary differential equations model, etc. To improve the sampling speed and to gain in efficiency, the minimum-volume enclosing ellipsoid (MVEE) is incorporated to enclose the “alive” particles in each iteration and to guide the sampler more efficiently towards the high probability region. The proposed inference scheme simultaneously provides the defective elements and a full probabilistic description of the damage severity. Different discrepancy functions measuring the similarity between the measured and predicted modal measurements could be employed to make Bayesian inference. The performance of the proposed probabilistic framework for damage detection and quantification was evaluated by means of numerical applications with different levels of complexity and a three-storey shear frame model using real data. It is demonstrated that the inclusion of the MVEE technique leads to extraordinary improvements in the computational efficiency of the ABC-NS sampler.
Keywords:
Structural damage detection
Approximate Bayesian computation
Modal measurements
Minimum-volume enclosing ellipsoid
Uncertainty quantification
Model updating
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.