基于HAM的汽车乘员约束系统多参数优化

针对汽车乘员约束系统高度非线性且难于求解最优值的特点,提出全局敏感性分析结合混合元模型的优化方法,通过蒙特卡罗模拟在整个设计空间内采样,以元模型代替仿真模型来完成设计参数的敏感性分析,并将分析获得的信息用于混合元模型优化（hybrid and adaptive metamodeling method,HAM）,将二阶多项式响应面、Kriging模型、径向基函数三种元模型有机结合,自适应选择最佳的元模型进行寻优.搜索过程中元模型不断更新与重建,逐渐提高关键区域的精度,从而快速寻找到全局最优解.对某工程实例的优化结果表明该方法是有效的.

Multi-parameter Optimization of Occupant Restraint System Based on Hybrid Metamodel Method

The occupant restraint system is highly nonlinear and difficult to get the optimal solution.In view of the above problem,a new method was proposed that combines global sensitivity with hybrid and adaptive metamodeling method（HAM）.First the Monte Carlo simulation was applied to sampling in the entire design region,the sensitivity of design variables were analyzed by using the metamodel instead of using the simulation model.Then the results of the analysis were applied back to HAM.This method integrated three different metamodels,including quadratic function model,Kriging model and radial basis function model,among which the appropriate one could be selected automatically according to the specific problem and adaptively updated in the optimization process Therefore,the accuracy in the core region was gradually improved and the optimal solution was found quickly.A satisfying result is achieved in an example by using this method.