Multi-objective optimization design method for rear seat backrest frame of passenger car based on hierarchical optimization and adaptive hybrid surrogate model

Addressing the curse of dimensionality in optimization design while simultaneously balancing optimization efficiency and computational cost has become a critical issue of concern for researchers. Based on sequential iterative optimization and multi-objective hierarchical optimization, this paper proposes an efficient optimization design strategy for automotive seat backrests.First, a finite element (FE) model of a passenger vehicle rear seat is established and validated against experimental results. Considering total cost, seat mass, and safety performance indicators, a comprehensive parameter sensitivity analysis is conducted to classify the structural parameters into primary and secondary sensitivity parameters. Subsequently, an integrated optimization framework is developed by combining optimal Latin hypercube design (OLHD), an RF–RBFNN adaptive ensemble surrogate model, the non-dominated sorting genetic algorithm III (NSGA-III), and the MPSI method. This framework is then applied to automotive seat frame optimization. Compared with the nonhierarchical multi-objective optimization strategy, the proposed method demonstrates significant advantages in optimization performance, computational efficiency, and convergence speed. Finally, the optimized rear seat backrest frame achieves a 15.52% reduction in mass and a 19.47% reduction in cost, while both comfort and occupant protection performance are improved to varying degrees.