A multi-objective optimization approach based on HBA-RF for lightweight design of automotive rear seats

In multi-objective optimization design of automotive seats, crashworthiness evaluation is often constrained by the high cost of physical experiments and the time-consuming nature of finite element simulations. Conventional single surrogate models frequently struggle to achieve satisfactory fitting accuracy when dealing with complex nonlinear responses. Moreover, complex engineering multi-objective optimization problems employing high-fidelity simulation models often suffer from excessive computational costs. Integrated surrogate models combining metaheuristic algorithms with surrogate modeling techniques can improve prediction accuracy and robustness; however, their global optimization capability remains challenging and requires efficient metaheuristic solvers. To address the imbalance between global exploration and local exploitation, susceptibility to local optima, and slow convergence of conventional Random Forest (RF) surrogate models, an optimized surrogate model (HBA-RF) is proposed. By incorporating the Honey Badger Algorithm (HBA), the optimization efficiency of the original RF is significantly enhanced, leading to improved global search capability, convergence stability, and overall performance. Based on this model, an integrated optimization framework is developed, incorporating optimal Latin hypercube design (OLHD), HBA-based RF surrogate modeling, the multi-objective grey wolf optimizer (MOGWO), and an MPSI-based decision-making method. To validate the effectiveness of the HBA-RF surrogate model, the trunk impact response of a passenger vehicle rear seat was selected as a test case, and comparative analyses were conducted against several modeling approaches reported in recent literature. For the passenger vehicle rear seat, headrest displacement ( Dis 1 ), seatback displacement ( Dis 2 ), cost, and mass were selected as the primary performance indicators. The proposed framework was further validated through a multi-objective lightweight optimization problem of the rear seat, demonstrating its feasibility and effectiveness in practical engineering applications. The optimization results indicate that, compared with the original design, the seatback mass is reduced by 3.46%, successfully achieving the lightweight design objective. In terms of performance, the headrest displacement ( Dis 1 ) is reduced by 5.05%, while the seatback displacement ( Dis 2 ) decreases slightly by 3.83%. Meanwhile, the cost decreases from 49.02 (¥) to 48.226 (¥), corresponding to a reduction of 1.62%. Simulation results confirm that all prediction errors remain within acceptable limits and that all performance indicators satisfy the predefined development targets for the rear seat. These findings demonstrate that the proposed method effectively balances lightweight design, cost, and safety performance, fulfilling the multi-objective optimization requirements of a passenger vehicle rear seat frame while highlighting its robustness and engineering practicality.