Abstract: Aiming at the problems of incomplete consideration of product design stage and structure optimization level in the optimization process of subway car body, unclear correspondence between the two, and the challenges of traditional finite element model optimization method such as large calculation amount, long time and difficult to meet the actual programming requirements, a full-stage multi-objective optimization design method of product design based on section performance parameterization was proposed. Firstly, the theoretical calculation formula of the section performance parameters which affect the quality and stiffness of the frame is derived. Then, the multi-objective optimization of the whole stage of product design was carried out for the aluminum alloy subway body underframe. In this optimization process, the minimum weighted flexibility (i.e. maximum stiffness) is taken as the optimization objective, and the variable density method is used to optimize the topology in the conceptual design stage of the frame section. In both the basic design stage and detailed design stage, combined with the deduced theoretical calculation formula of the frame section performance parameters,Non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ) is used to carry out shape-free dimension collaborative optimization. Finally, the topology optimization results and the whole stage optimization results are reconstructed respectively, and the detailed analysis and check are carried out. The results show that：This method can significantly improve the static rigidity of the bottom frame section and achieve the purpose of lightweight. The section area A (mass) of the three profiles is reduced by 4.06%, 1.31% and 5.17% respectively, and the bending moment of inertia I_x is increased by 9.50%, 5.21% and 9.00% respectively. The section performance of the designed profiles is excellent. It has guiding significance for systematic optimization design of products.

Key words: subway underframe, section design, full stage multi-objective optimization, section performance parameterization, topology optimization, Non-dominated sorting genetic algorithm Ⅱ