基于可编程图形处理器的可视化技术在中子学分析中的应用研究  

GPU-Based Visualization in Neutronics Analysis

作　　者：龙鹏程[1] 罗月童[2] 邹俊[1] 黄善清[1] 李佳[1] 闫锋[2] 范言昌[2] 吴宜灿[1,3] 无 

机构地区：[1]中国科学院等离子体物理研究所,安徽合肥230031 [2]合肥工业大学计算机与信息学院,安徽合肥230009 [3]中国科学技术大学核科学技术学院,安徽合肥230027 [4]FDS团队

出　　处：《核电子学与探测技术》2010年第8期1042-1045,共4页Nuclear Electronics & Detection Technology

摘　　要：随着计算模型规模与复杂性的不断提高、中子学自动建模程序的发展,中子学计算后处理已逐渐成为中子学分析工作的瓶颈。FDS团队为提高中子学计算后处理效率,利用科学计算可视化等相关技术,自主研发了集数据管理、预处理与可视化分析于一体的中子学可视化系统SVIP-N;并基于可编程图形处理(programm able Graphics Processing Unit,GPU)实现了体数据的裁剪可视化等高级功能。以国际热核实验堆ITER的中子通量密度场为测试用例进行的测试与初步应用实践表明,SVIP-N显著简化了中子学分析工作的流程,提高了分析效率,为中子学计算后处理提供了直观、高效、实用的可视化分析环境。The post - processing of neutronics calculations becomes the bottleneck of neutronics analysis with the increasingly scale and complexity of problems, and the development of automatic modeling system. SVIP - N is a 3D visualization platform with data management, pre -processing and various visualization functions, which self - developed by the FDS Team, designed to enhance the post - processing efficiency of neutronics analysis. In the study, the advanced visualization supporting complicated volume cutting was provided by using Graphics Processing Unit （GPU） programming technique. Based on the density field of neutron flux of International Thermonuclear Experimental Reactor （ ITER）, it was proved that SVIP - N apparently simplified the analysis process and greatly enhanced the efficiency. In conclusion, the SVIP - N may become an intuitive and efficient analysis platform for neutronics analysis.

关 键 词：中子学分析 科学计算可视化 体数据裁剪 可编程图形处理器 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]