中文摘要：

为得到具有理想耐撞性能的铁道车辆承载吸能结构，分别基于多项式响应面法的二次响应面模型、四次响应面模型和Kriging法的响应面模型等3种代理模型，构造承载吸能结构的比吸能SEA及比吸能与撞击力峰值之比REAr关于设计参数的二次、四次和Kriging法响应曲面，结合遗传算法整体寻优分别得到这3种代理模型的SEA和R队r的最优值。对比分析结果表明：Kriging法响应面模型的拟合精度高于多项式的二次和四次响应面模型，四次响应面模型的拟合精度次之；但是Kriging法响应面模型拟合曲面没有多项式响应面模型的光滑，其原因是Kriging模型采用的局部插值方法虽能提高模型拟合精度、却不利于降低模型拟合过程中的数字噪声；整车车体碰撞仿真表明，承载吸能结构优化后的整车车体具有更好的耐撞性能。

英文摘要：

In order to obtain the ideal crashworthy energy-absorbing bearing structure of railway vehicle, the quadratic, quartic and Kriging response surfaces of specific energy absorption （SEA）, ratio of SEA to initial peak force （R EAF） with respect to the design parameters were respectively constructed based on three kinds of surrogate models, namely, the quadratic and quartic response surface models of the polynomial response surface method （PRSM） and response surface model of Kriging method （KM）. Then the S EA. and R EAF optimal results of three kinds of surrogate models were obtained by genetic algorithm （GA）. Comparative analysis results show that the fitting accuracy of KM response surface model is higher than that of the PRSM quadratic and quartic response surface models, and the fitting accuracy of the quartic response sur- face model is higher than that of the quadratic model. However, the fitting surface of KM response surface model is not as smooth as that of PRSM response surface model. Although the local interpolation method of Kriging model can improve the fitting precision, it does not help to reduce the digital noise during the model fitting process. Finally, through the collision simulation analysis of vehicle bodies, it indicates that the vehicle body after energy-absorbing structure optimization has better crashworthiness.