Complex engineering multi-objective optimization problems that utilize high-fidelity simulation models often encounter challenges related to computational expense; however, surrogate-based optimization methods can effectively mitigate this limitation. This paper presents an integrated optimization framework that encompasses optimal Latin hypercube experimental design, a hybrid surrogate model combining a DBO optimization algorithm with weighted prediction error reduction, a non-dominated sorting genetic algorithm-II, and the MPSI method as integral components of a comprehensive systematic optimization strategy. Subsequently, the thickness and number of layers of the carbon fiber-reinforced polymer (CFRP) automotive front crash beam are established according to the principle of equal stiffness replacement. Ultimately, this optimization strategy is implemented in a multi-objective lightweight optimized design concerning the layup sequence of the CFRP bumper beam. The findings of this study demonstrate that the DBO-EWL surrogate model exhibits favorable predictive accuracy, which fully meets the precision requirements for subsequent optimization analyses. Furthermore, comparative evaluations reveal that the optimized carbon fiber-reinforced polymer (CFRP) front crash beam not only delivers superior crashworthiness performance relative to the original high-strength steel bumper beam but also achieves a substantial reduction in overall structural weight. The optimized CFRP bumper demonstrates superior crash performance compared to the original steel structure, reducing the maximum crash force by 10.26%, while maintaining stable intrusion and energy absorption. Additionally, the optimized bumper achieves a substantial 67.5% weight reduction.
Gao et al. (Fri,) studied this question.