A novel optimization strategy combining hybrid surrogate modeling and hybrid optimization algorithms for multi-objective design of a rear seat backrest frame in a passenger car

Abstract

In the multi-objective optimization design of automotive seats, traditional single surrogate models often fail to achieve the required fitting accuracy when dealing with complex nonlinear responses, and the quality of optimal solutions obtained solely through global optimization algorithms remains suboptimal. To address these challenges, this paper proposes a novel hybrid surrogate model with enhanced predictive capability and integrates hybrid global optimization with gradient-based algorithms to further refine solution quality. Accordingly, an integrated optimization framework is developed, encompassing optimal Latin hypercube sampling, a hybrid surrogate model that combines the Pointer optimization algorithm with weighted prediction error reduction, and the archive-based micro genetic algorithm-Hooke-Jeeves combinatorial optimization algorithm, thereby establishing a comprehensive and systematic optimization strategy. The feasibility of this strategy is validated through an engineering case study on automotive seat optimization, demonstrating that the combination of the hybrid surrogate model and the combinatorial optimization algorithm offers notable advantages. Compared with classical multi-objective optimization strategies, the proposed method outperforms in terms of cost-effectiveness and lightweight performance, fully utilizes the design space for multi-objective optimization, and serves as a reliable reference case for future research. The research findings indicate that, compared with the original steel rear seat, the optimized version exhibits enhanced crashworthiness and a significant reduction in overall weight and material cost. Specifically, in terms of crash performance, the maximum displacement of the headrest frame is reduced by 3.81%, while that of the backrest frame remains well within the permissible range specified by relevant safety standards. Meanwhile, the seat’s weight and cost are reduced by 17.6% and 21.15%, respectively.