基于GPU的嵌套网格装配方法  

作　　者：杨克龙 

机构地区：[1]南京航空航天大学,南京210016

出　　处：《科技创新与应用》2025年第1期177-179,184,共4页Technology Innovation and Application

摘　　要：采用嵌套网格可以有效地处理大幅运动问题,但随着网格规模的增大和流动问题复杂度的提高,传统的基于CPU的嵌套网格装配方法越来越难以满足当前的计算需求。针对上述问题,该文基于CUDA平台,发展一种基于GPU的k-d树嵌套网格装配方法,并对k-d树构建过程和搜索过程进行优化,大大提升贡献单元搜索效率和物面距计算效率,进而加快嵌套网格装配速度。Using overset grid can effectively handle problems with large-scale motions.However,as the scale of the grid and the complexity of the flow problem increase,traditional CPU-based overset grid assembly methods are becoming increasingly difficult to meet current computational demands.To address these issues,this paper develops a GPU-based k-d tree nested grid assembly method based on the CUDA platform,and optimizes the k-d tree construction process and search process,which greatly improves the contribution unit search efficiency and object-surface distance calculation efficiency,thereby accelerating the nested grid assembly speed.

关 键 词：图形处理器 嵌套网格 K-D树 装配方法 流场计算域 

分 类 号：V211.3[航空宇航科学与技术—航空宇航推进理论与工程]