Combined Bayesian and error assessment-based model calibration method for vehicle under-belly blast with uncertainty

The research of military vehicle protection against the under-belly blast is a particular topic that enormously relies on computer-aided engineering (CAE) simulation technology because of the high cost of relevant real physical tests. CAE model with high predictive performance is required to obtain accurate results. An efficient model calibration process is necessary to improve CAE modeling. Traditional model calibration method uses visual comparison; this cannot provide any quantitative, and cannot consider uncertainty. This paper would study an inverse model calibration process that combines error assessment of response time histories (EARTH) and Bayesian method. Experimental data and computational data would be compared, and the discrepancies between them are quantified by EARTH metric. And Bayesian method is applied to address the uncertainties of discrepancies quantization caused by model parameters uncertainties. Comparison of model confidence level can be quantified by Bayes factor. Then, the calibration process can be considered as an optimization problem solved by optimization-based method. The optimization objectives are minimizing quantified discrepancies between experimental data and computational data and maximizing model’s Bayes factor. The model parameters which need to be calibrated would be improved during this process. An example case of an occupant loaded in vehicle under-belly blast scenario is used to describe the raised method.