基于全局敏感性分析和Kriging代理模型的汽车乘员约束系统多参数优化研究

针对汽车碰撞乘员约束系统优化设计中多参数与高度非线性的问题，提出了基于Kriging模型的设计参数全局敏感性分析方法，通过计算各设计参数对系统响应方差的贡献来评估各参数的敏感性。在此基础上，选取对响应影响大的设计参数作为优化变量，运用多岛遗传算法进行优化求解。优化结果表明，人体综合损伤WIC值相比初始设计值降低11.9%。研究结果为实际工程设计提供了参考。

Study on Multi-parameter Optimization of Occupant Restraint System Based on Global Sensitivity Analysis and Kriging Model

Aiming at the problems of multi-parameter and high nonlinearity in the optimal design of occupant restraint system, a method of parameters global sensitivity analysis based on Kriging model was raised. This method evaluated the sensitivity of each parameter through calculating the contribution of design parameters to the variance of response. On the basis of these data, the design parameters which were more important to response were chosen as optimization variables, additionally. Multi-island genetic algorithm was used to realize optimal design. The results of optimal design indicate that in contradistinction to initial design values, the value of human injury WIC(weighted injury criterion) is reduced by 11.9%, it provides a good reference for actual engineering design.