In this paper, the first-seal cross-section with bar structure of a typical automotive door  was analyzed. The sponge tube of the weatherstrip, which produces most of the compressive load, was divided into five sub-areas whose thicknesses and angles were selected as optimization variables, considering the compression space of the door sheet metal and nonlinear deformation process of the sealing system. Based on  the engineering practice, an optimization objective function using the demanded compression load deflection (CLD) criterion was established. The nonlinear mapping between cross-section parameters and compression load was built by BP neural network and the parallel intelligent optimization was realized for ideal cross-section structure parameter. The engineering application proved that 15% of cycle time can be reduced using this computer-aided design method.