| According to the CFD-2030 vision of the renowned CFD research community,the number of grids in aerospace flow fields will grow to the scale of billions or even tens of billions around 2030.Traditional postprocessing methods are no longer capable of handling such large-scale flow field data.The current challenge is how to effectively utilize a single high-precision CFD simulation data.Moreover,there is a lack of effective processing methods for real-time automated analysis,processing,and display of such ultra-large-scale and high-precision CFD simulation data in both existing commercial software and aerospace self-developed software.In response to the aforementioned issues and challenges,this project has conducted research and implementation of an aerospace big data analysis system based on in-situ computing.Its main features include:(1)building an aerospace in-situ computing cluster system based on selfdeveloped aerospace solvers,which has 48 high-performance computing nodes to provide the capability of large-scale flow field data in-situ analysis;(2)proposing a mesh partitioning method based on topological reconstruction,enabling the system to handle complex large-scale flow field meshes;(3)implementing automated in-situ computing analysis functions based on decision trees,improving the efficiency and real-time analysis capability of aerospace big data analysis;(4)implementing offline in-situ computing of flow field mesh partitioning and synthesis based on parallel image synthesis methods,thereby supporting the visualization and analysis of flow field data with grid scales in the billions.The in-situ computing aerospace big data analysis system implemented in this project has been applied in the field of aerospace fluid mechanics research.By achieving quasi-real-time analysis and interactive visualization capabilities of ultra-large-scale grid flow field data,it greatly improves the productivity of aerospace researchers. |
