飞行器大数据量CAD模型并行预处理

摘要

In order to improve the pre-processing efficiency of massive aircraft CAD models, a parallel processing system based on PC-cluster instead of a single PC is constructed by using the message passing interface (MPI). With the parts and components from the CAD model assembly tree as objects, a parallel pre-processing algorithm is presented. By simplifying the geometrical models, generating LOD and partitioning the models in a parallel way, meanwhile, balancing the dynamic load and optimizing the data transmission and storage, the pre-processing efficiency can be improved drastically. An example, in which the aircraft CAD models with ten millions of triangles and storage up to GB level can be processed in 60 seconds, is given to demonstrate the efficiency of this approach. The experimental result shows that the processing speed with the method can be improved more than 10 times over that with the single PC.%针对虚拟仿真中飞行器大数据量CAD模型在单台计算机执行预处理非常耗时的问题,采用消息传递接口(MPI)并行开发技术搭建了机群并行处理软硬件平台.以CAD模型装配树的零部件为处理对象,设计了并行预处理算法；通过并行执行几何模型简化、LOD生成、模型分割等预处理,以及动态负载均衡与数据传输存储优化来提高并行处理的效率.实验结果表明:与单台计算机执行相比,其预处理速度提高10倍以上.包含千万级三角面片、GB级数据量的飞行器CAD模型采用该并行处理系统,可以在60 s左右完成预处理.