基于空间收缩回归的汽车耐撞性优化

选择汽车正面碰撞耐撞性为优化目标，结合最优拉丁超立方样本点设计、RBF模型以及NSGA-Ⅱ多目标优化对设计变量进行优化。为了加快目标值收敛速度并寻求可能存在的初始设计域之外的最优解，引入空间收缩回归法在迭代过程中对设计域进行动态更新。优化迭代过程最终收敛到一组最优解，使得在未增加车身总质量的条件下提升了车身正面碰撞耐撞性。通过与有限元模型计算结果进行对比分析，验证了该方法具有收敛速度快和寻优精度高的特点。

Optimization of Vehicle Crashworthiness Based on Shrinking Space Regression Method

The frontal crashworthiness of a vehicle was chosen to be optimized herein. Opt LHD method, RBF model and NSGA-Ⅱ multi-objective optimization method were combined to optimize design variables. To accelerate the convergences of objectives in the process and seek latent solutions out of the initial design space, the shrinking space regression method was introduced to refresh the design region dynamically. The convergence conditions were finally met after several rounds of iteration. An optimal solution was reached, which promoted the frontal crashworthiness shapely without any mass increased. The fast convergence and high accuracy of this method was proved by the comparison between predicted and FEA results.