基于GPU并行的液货船破损浮态计算

为了提高液货船破损浮态计算性能,提出一种基于图形处理器(graphics processing unit,GPU)的船舶破损浮态并行计算方法。通过对船舶三维模型切片,然后对所有切割肋位横截面做等距偏移和简化,得到船舶外壳和破损舱室的外板离线数据。在GPU中进行船舶外壳、破损舱室和水线面的求交计算。然后对雅可比矩阵系数分类进行并行计算,将计算得出的雅可比系数送回中央处理器(center processing unit,CPU)中求解船舶破损浮态方程。通过计算两种载况下的破损组合,与直接使用CPU计算相比,在保证计算精度的前提下,计算速度可提高9~14倍,有效提高了液货船破损浮态浮态计算的实时性。

GPU Based Parallel Computation for Damaged Floating Conditions of Tankers

For a better computation performance of damaged floating conditions of tankers,a parallel computation method based on Graphics Processing Unit(GPU)was proposed.By slicing 3D model,all the sliced middle sections were equidistantly offset and simplified,resulting in the offline data of the shell and damaged compartments.The intersection calculation of shell,damaged compartments and waterline plane was carried out in GPU.The parallel computation was performed for Jacobi matrix coefficient classification.The obtained Jacobi coefficient was sent back to CPU in order to solve the equation of damaged floating conditions.The damage cases under two different loading conditions were calculated.Under the prerequisite of ensuring the calculation accuracy,the computation speed could be increased 9 to 14 times compared with the method to use CPU directly,improving the real-time performance of the damaged floating condition computation of tankers.