With the rapid development of new energy vehicles (EVs), the battery enclosure’s design has become a critical bottleneck, requiring an optimal balance between extensive lightweighting and stringent crashworthiness standards. However, the high nonlinearity of crash responses and the excessive computational cost of high-fidelity finite element (FE) models pose significant challenges for structural optimization. This study proposes a comprehensive Simulation-Modeling-Optimization-Decision collaborative framework to address these issues. First, a high-precision adaptive surrogate model is developed by fusing the Kriging model with the K-Nearest Neighbor (KNN) algorithm (K-Kriging) to capture complex nonlinear responses efficiently. Second, a multi-objective optimization (MOO) problem is formulated, considering 16 component groups with both thickness and material selection variables (totaling 32 design variables). The objectives are to minimize material cost and structural mass while maximizing the first-order natural frequency, subject to stringent crashworthiness constraints derived from nonlinear crushing and impact simulations. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) is employed to search for the Pareto frontier. To identify the optimal compromise solution, a Modified VIKOR (M-VIKOR) method, integrating Grey Relational Analysis (GRA), is implemented to mitigate the subjectivity of weight assignments. Results demonstrate that the optimized design achieves a 17.03% reduction in material cost, a 16.91% reduction in structural mass, and a 19.87% increase in the first-order natural frequency compared to the baseline design. The proposed framework not only ensures superior predictive accuracy (relative errors < 5%) but also provides a robust methodology for the multidisciplinary design of complex EV structural systems.
Li et al. (Thu,) studied this question.