基于GPU的弹道快速计算方法

摘要

For the problem that the calculation in aircraft routine planning system is huge, and time-consuming, and associating with the GPU (Graphics Processing Units) large-scale parallel computing feature, a fast calculation method for trajectory base on GPU is raised here. This method uses the Fourth-order Runge-Kutta to resolve the trajectory differential equations, makes the trajectory computing taskrnthe control ability in CPU + GPU heterogeneous platform in data stream , Lastly, compute thousands of trajectories in the parallel mode. The experimental results show that the calculation accuracy of the method can meet the requirements, and the speed of the method is several decades faster than the traditional method computing on the CPU. This method is also offering some new ideas for the other related applications which have large-scale parallel calculation in aircraft route planning system.%针对目前飞行器航迹规划中弹道计算量大、耗时长的问题,结合GPU大规模并行计算特性,提出了一种基于GPU的弹道快速计算方法.该方法以4阶Runge-Kutta (R-K)法为基础求解弹道微分方程组,通过计算资源的分配、数据流分段将弹道计算任务映射到GPU线程模型,利用CPU加GPU异构模型的数据流控制能力,实现上千条弹道的并行计算;实验表明该方法计算精度满足要求,并获得几十倍理想加速效果;也为航迹规划系统其他大规模并行计算提供了新的解决思路.