论文部分内容阅读
针对目前飞行器航迹规划中航迹计算量大、耗时长的问题,结合图形图像处理器(GPU)大规模并行计算特性,提出了一种基于GPU的航迹快速计算方法。该方法以4阶Runge-Kutta(R-K)法为基础求解航迹微分方程组,通过计算资源的分配、数据流分段将航迹计算任务映射到GPU线程模型,利用CPU+GPU异构模型的数据流控制能力,实现多条航迹的并行计算。试验表明,该方法的计算精度满足要求,并获得了几十倍的理想加速效果,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。
In view of the large computational load and long time-consuming trajectory of the current aircraft trajectory planning, a GPU-based rapid trajectory calculation method is proposed in combination with the large-scale parallel computing characteristics of the graphics image processor (GPU). The method solves the system of differential equations on the basis of the fourth-order Runge-Kutta (RK) method. By computing the resource allocation, the data flow segmentation maps the trajectory calculation task to the GPU thread model. By using the CPU + GPU heterogeneous model Data flow control capabilities, to achieve parallel calculation of multiple tracks. Experiments show that the method can meet the requirements of calculation accuracy and obtain the ideal acceleration of several tens of times, which provides a new solution to other large-scale parallel computation of flight path planning system.