Abstract: Based on the condition of 40% overlap offset crash against deformable barrier in C-NCAP, the structural crashworthiness of a car is optimized in this paper. In order to enhance the predicted accuracy of output responses, the support vector regression model based on particle swarm optimization algorithm is employed to simulate the relationship between output responses and design variables, and the Pareto front of the proposed optimization problem is obtained by using non-dominated sorting multi-objective genetic algorithm II. On the basis of deterministic optimization with consideration of the fluctuation of product performance under the influence of uncertain factors, an optimization design for robustness is conducted. Finally, the results of optimization are verified by finite element simulation. The results show that after optimization the mass of structure is reduced, the crashworthiness is obviously enhanced, and the stable product performances are ensured

Key words: 40% overlap offset crash, lightweighting, PSO-SVR approximate model, robustness, optimization design