Multi-objective optimization strategy for rear seat skeleton of a passenger car based on MO-SHERPA optimization algorithm

Abstract

In order to reduce the complexity of multi-objective design for automotive seat structures under crash scenarios, this study applies the MO-SHERPA algorithm to the automotive seat frame optimization problem for the first time. Based on a validated high-fidelity finite element model of a passenger car rear seat backrest, key design variables, material alternatives, and thickness ranges are defined to formulate the multi-objective optimization problem. The adaptive hybrid search capability of MO-SHERPA within HEEDS is leveraged to explore the design space efficiently, reducing reliance on deep theoretical knowledge and extensive surrogate modeling. To balance prediction accuracy and computational effort, a genetic aggregation response surface surrogate model is constructed using limited finite element samples. The optimization process evaluates different iteration schemes, revealing that increasing the number of evaluations significantly improves the Pareto front quality. The modified technique for order preference by similarity to an ideal solution decision-making method is then applied to select the optimal compromise solution from the Pareto sets. Comparative results show that with only 200 evaluations, the MO-SHERPA approach achieves a weighted improvement of 19.3%, outperforming local optimization and achieving similar or better performance than the non-dominated sorting genetic algorithm-III algorithm, which requires more than five times the evaluations. This demonstrates that MO-SHERPA can deliver robust, high-quality solutions for nonlinear crashworthiness problems with reduced computational resources. The proposed strategy provides practical guidance for multi-objective lightweight design of automotive seat structures, highlighting the potential of adaptive intelligent search algorithms to simplify and enhance complex engineering optimizations.