中文摘要：

针对工程中复杂耗时的黑匣子问题,提出一种基于多组混合元模型的全局最优化方法。该方法同时应用三组元模型进行搜索,每组都包含克里金、二阶多项式及径向基函数的元模型。应用该优化方法对某车型的后车架进行轻量化设计后,后车架子系统的质量减少7.2kg,并提高了该子系统的刚度。研究结果表明,与HAM法相比,新方法在搜索效率和精度上都有显著提高。

英文摘要：

Aiming at the complex and expensive black-box problems in engineering,a multiple hybrid meta-model based global optimization method was proposed.In the proposed method,three sets of meta-models were employed in the search of the design space simultaneously,each of which included the well-known Kriging,quadratic function and radial basis functions.After the proposed method was applied to the rear frame of a vehide,the mass of the rear frame is reduced by 7.2kg,and subsystem stiffness is also improved.The study shows that both of the search efficiency and accuracy are noticeably improved compared with hybrid and adaptive meta-modeling method.