A novel Bayesian model validation method based on bootstrap resampling and distance remapping

A Bayesian model validation method based on Bootstrap resampling and distance remapping techniques is proposed. Building upon a rigorous likelihood formulation and a judicious choice of priors, the proposed method performs posterior inference using a novel hybrid algorithm. This algorithm couples Transitional Markov Chain Monte Carlo (TMCMC) with Simulated Annealing (SA) and is further augmented by Bootstrap resampling and distance remapping strategies to effectively overcome the limitations imposed by strong hyperparameter dependence. The research develops a concentration index to systematically classify and assess the effectiveness of model calibration and the quality of samples participating in the calibration process. The effectiveness of the method is validated through the NASA Langley UQ challenge problem and an engineering wing model. This research provides a theoretical and methodological framework for model updating and validation in the engineering field.