基于Voronoi序列采样的加筋壁板优化设计

Optimum Design of Stiffened Panels Based on Voronoi Sequence Sampling Method

为提高加筋壁板结构轻质优化效率,提出一种基于Voronoi序列采样的加筋壁板优化设计方法。建立了该结构的参数化模型,分析了加筋壁板蒙皮与桁条腹板的网格规模对结构承载能力及失稳模式的影响规律,并依据桁条腹板高厚比进行了网格划分,进而平衡后屈曲分析精度与效率;然后,提出了基于探索策略和开发策略的序列近似优化方法,其中Voronoi加点侧重全空间探索,近似最优解侧重局部开发,通过动态加点迭代至收敛,实现综合探索和开发性能的并行采样,提高了算法的收敛精度和计算效率;最后,应用该算法进行加筋壁板轻量化设计,并与传统序列近似优化方法进行对比。结果表明,本文方法利用更少的初始样本点,通过更少的迭代次数达到更高的优化精度,并获得了相对初始设计减重32.6%的优化结构。

To improve the efficiency of lightweight optimization for stiffened panels,an efficient dynamic sequential lightweight optimization method based on Voronoi is proposed.Above all,the influences of the mesh size of the stiffened panel skin and strings on load-carrying capacity and the buckling modes of the structure are analyzed based on the parametric model.Furthermore,the dynamic mesh partition is realized according to the height⁃thickness of the web to balance the accuracy and efficiency of post-buckling analysis.Then,a dynamic sequential approximate optimization incorporating an exploration strategy and an exploitation strategy is proposed.Sampling by Voronoi focuses on global space exploration,while sampling by approximate optimal solution focuses on local exploitation.The parallel sampling with both explorative and exploitative performance is realized by dynamic sampling iteration to convergence,which improved the convergence precision and computational efficiency of the algorithm.Ultimately,the proposed algorithm is applied to the lightweight design of stiffened panels,and compared with the traditional sequential approximation optimization method.The results reveal that the proposed method needs fewer initial sample points and less iteration times to achieve higher optimization accuracy,and obtains an optimized structure with a weight reduction of 32.6%compared with the initial design,which indicates that the proposed method is validated for the prospect of engineering applications.