Abstract: Lightweight design of mechanical parts plays an important role in improving equipment performance and reducing cost.To improve convergence,optimization efficiency for the parts lightweight design of traditional Kriging method,an adaptively dynamic Kriging model was proposed.The multiple correlation coefficient was utilized to achieve the goal of overall update of Kriging model,and the minimums of each objective from the Pareto solutions were utilized to search for the optimal characteristic point.Then the local update of Kriging model was achieved by the feasible solutions near the optimal characteristic point.Numerical test cases and a platen of FT(N)-2800 injection machine were given to demonstrate the effectiveness of the proposed method.

Key words: lightweight design, adaptively dynamic Kriging model, multiple correlation coefficient, optimal characteristic point

摘要： 机械零件轻量化对提高装备运行性能、降低成本具有重要作用。针对传统基于Kriging模型的全局优化方法在处理零件轻量化问题时,收敛速度慢、优化效率低、精度差等问题,提出一种自适应动态Kriging模型。采用复相关系数R2检验实现Kriging模型的全局动态更新,以Pareto解中各目标函数最小值为特征点,通过特征点加权计算理想特征点位置,在理想点附近选取可行解完成样本局部更新,实现Kriging模型的动态自适应性。以数值算例和FT(N)-2800注塑机模板为例,验证了所提方法的有效性。

关键词: 轻量化设计, 自适应动态Kriging模型, 复相关系数, 理想特征点