论文部分内容阅读
为提高复杂条件下时间域航空电磁数据解释精度,本文开展了时间域航空电磁三维并行反演算法研究。该算法中的三维正演是基于有限差分技术,并采用“移动脚印”技术来减小实际计算模型尺寸;三维反演基于Gauss-Newton反演方法,并采用显式灵敏度矩阵计算技术减少反演过程中的正演次数。为提高三维反演的效率,本文基于OpenMP并行库实现了三维反演的并行化。从理论和实测数据的三维并行反演结果可以看出本文的并行化策略明显地提高了三维反演的速度,能够胜任大数据量时间域航空电磁实测资料三维反演解释任务。
In order to improve the accuracy of aeronautical electromagnetic data interpretation in the time domain under complex conditions, this paper studies the three-dimensional aeromagnetic parallel algorithm in time domain. The three-dimensional forward modeling in this algorithm is based on the finite difference technique and adopts the “moving footprints” technique to reduce the size of the actual calculation model. The three-dimensional inversion is based on the Gauss-Newton inversion method and uses the explicit sensitivity matrix calculation technique to reduce The number of forwards in the inversion process. In order to improve the efficiency of 3D inversion, this paper implements the parallelization of 3D inversion based on OpenMP parallel library. From the results of three-dimensional parallel inversion of the theoretical and measured data, it can be seen that the parallelization strategy of this paper obviously improves the speed of 3D inversion and is capable of performing the three-dimensional inversion and interpretation tasks of the measured data of the aeronautical and electromagnetism in the large data volume time domain.