基于响应面法的汽车吸能部件优化问题研究

对汽车的前纵梁中不同构件板厚的优化,将其总体重量控制在一定范围内的前提下,对其吸能效果以及最大加速度进行优化。对于目前国际上流行的遗传算法,神经网络算法等优化手段求解问题收敛性是建立在对正问题多次迭代求解的基础上,导致其计算量庞大。为了打破这个瓶颈,建立了基于调整权函数策略的移动最小二乘法的响应面方法,通过均匀拉丁方试验设计方法,选择了适当的设计参数样本,快速建立了精确的多维响应面模型,并采用Pareto遗传算法对该模型进行了多目标优化,得到了较好的结果。

Study of Optimization for Absorbing Energy Member of Automobile Based on Response Surface Method

The optimization for thickness of plates of front side member of vehicles in the crashworthiness problems for maximizing absorbing energy and minimizing maximum acceleration was proposed based on the limitation of weight. Due to computational complexity of forward problems, the popular optimization methods are time-consuming ways, such as neural net and genetic algorithms. For breaking this bottleneck, the response surfaces by using moving least squares approximations constructed with a moving region of interested based on the updated weight function method was suggested. The uniform Latin square method designed the rational experimental samples. The accurate response surface for crashworthiness problem was rapidly built and the better results optimized by Pareto genetic algorithm were obtained.