Fast Bayesian approach to model calibration of vehicle occupant restraint systems

Abstract

ABSTRACTVehicle occupant restraint systems play a crucial role in preventing and reducing occupant injuries. One of the difficulties in designing these systems by computer aided engineering (CAE) analysis is identifying and incorporating friction coefficients (shoulder belt–shoulder of dummy, shoulder belt–thorax of dummy, waist belt–abdomen of dummy, etc.) due to existence of inevitable uncertainty. This paper describes the use of fast Bayesian approach for efficient sampling of the posterior distributions of unknown friction coefficients. The adaptive densifying approximation technique accelerated Markov Chain Monte Carlo (MCMC) method is applied to quickly identify the means and confidence intervals of friction coefficients using the observed head acceleration response of a dummy.