A credible interval model updating method for structural population analysis and design stages based on small samples

In practical engineering, a persistent discrepancy exists between numerical simulations and real responses. This gap significantly undermines reliability in the established models and spurs the development of model updating. Yet, during the structural analysis and design phases, the focus of model updating often extends beyond the current structure to encompass the same type of structural population, so this paper proposes a credible interval model updating method for addressing the issue of uncertain model updating. This method divides the uncertain model updating problem into two subgoals: ensuring that the experimental responses credibly describe the real responses and that the simulation responses accurately fit experimental responses. For the first subgoal, the non-probabilistic credible convex sets for multi-type responses are established by introducing the concepts of multidimensional response space and credibility level. For the second subgoal, this paper categorizes model parameters into uncertain parameters and updating parameters, allowing the simulation model to fully consider prior information and be more generally applicable to the uncertain conditions of structural population. Particularly, the comparison between the predictions of the updated model and experimental results from other operating conditions highlights the robustness of the updated model and the advancement of the methodology.