Bayesian Approximation-Assisted Optimization Applied to Crashworthiness Design of a Pickup Truck

Direct application of most optimization techniques, especially Multi-Objective Evolutionary Algorithms (MOEAs) that require many response evaluations, is computationally prohibitive for most real-world engineering simulations. In this paper, an approximation-assisted approach to multi-objective optimization of computationally expensive response functions is presented. We employ a Bayesian approach, referred to as Sequential MAXimum Entropy Design (SMAXED), for design of experiments and global approximation of an expensive finite-element model, i.e., crash event simulation of front end of a pick-up truck. The approximation model is optimized using a multi-objective genetic algorithm. It is shown that while the approach dramatically reduces the computational costs, it also finds a good estimate to the Pareto-optimal solution set for such a complex problem.