Abstract: On the basis of considering the manufacturing and assembling costs, a component divide method based on graph decomposition algorithm is proposed in this paper, with a technique of multilayer selection from Pareto solution set introduced to solve the convergence problem in optimization with thicknesses as design variables. With the underbody panel of a bodyinwhite as example, the structure is divided according to the minimum manufacturable size, and with the connection relation between divided units as design variables, a simulation is conducted by using NSGAII algorithm with optimum assembly schemes obtained. Then with the assembly schemes in Pareto solution set obtained in optimization as initial solutions, from which the solutions meeting stability conditions are selected by using the technique of selection from Pareto solution introduced. Next, constraints are applied to the thicknesses within subcomponents to perform thickness optimization and the thicknesses obtained are returned back to graph decomposition for iteration until the problem converges, and the concurrent optimization of welding locations and panel thicknesses is achieved in the end.

Key words: car body assembly, graph decomposition algorithm, selection from Pareto solution set, multi layerl optimization, multidisciplinary optimization