Multi-parameter, Multi-objective Optimization of Injury Indexes of Vehicle Crash Models

This paper presents a method for optimizing occupant restraint system parameters in vehicle frontal crashes. Simulation models incorporating restraint systems and dummies are used for predicting injury indexes. A full-scale survey of all of the design parameters related to the injury indexes would require a vast number of simulations. Therefore, the Design of Experiments (DOE) method involving a minimum number of experiments is more realistic. However, dummy behavior often shows discontinuity if the dummy comes in contact with the steering wheel, so it is not predicted well with usual DOE methods. This paper shows how to incorporate such discontinuity in a DOE study and how to optimize the restraint system parameters to reduce occupant injury indexes. It also discusses the feasibility of this method for integrated optimization of 50th percentile and 5th percentile dummies.