A New Hybrid Stochastic Optimization Method for Vehicle Structural Design

With the continuous improvement of powerful computers, vehicle structural designs have been addressed using computational methods, resulting in more efficient development of new vehicles. Most simulation-based optimization generates deterministic optimal designs without considering variability effects in modeling, simulation, and/or manufacturing. This paper presents a new hybrid stochastic optimization method for vehicle side impact design. Nonlinear response surface models are employed as the ’real’ models for the side impact related performance functions to conduct this study. The main goal is to maintain or enhance the vehicle side impact performance while minimizing the vehicle weight under various uncertainties. The new method alleviates the computational burden of excessive model evaluations by estimating the objective and constraint functions during the optimization process through a reweighting approach. It also combines genetic algorithm and sequential quadratic programming method to achieve the global or a near global stochastic optimal solution for highly nonlinear nonconvex optimization problems. The efficiency and accuracy of this method are demonstrated by solving a vehicle safety design problem.