Abstract: A simplified finite element model of the dummy chest is established based on sensitivity analysis of the structural material properties. The model is characterized with coarse and fine meshes and also satisfies the experimental calibration. In order to improve the anthropomorphic precision of the simplified model, the engineering issue is abstracted as a multiobjective optimization issue, which aims to reduce errors of both chest displacement and pendulum force between the simplified dummy and the physical dummy simultaneously. Considering sternum material parameters as the optimization parameters, high precision surrogate models of objectives are constructed by polynomial response surface method. A MBNWS algorithm is proposed, which is based on BFGS method in conjunction with weighted sum technique and Frisch method. The Pareto optimal solution and its front of the optimization problem for dummy chest are achieved by the MBNWS algorithm. The optimized dummy chest structure can simulate the mechanical response of the physical dummy chest more accurately during vehicle collision. In addition, the MBNWS algorithm is proved fast and efficient, and it is suitable to solve multiobjective optimization problem in engineering practice.

Key words: mechanical design, vehicle crash, dummy chest modeling, multiobjective optimization, Pareto optimal solution, weighted sum method, BFGS algorithm