随着大地电磁法三维勘探及处理技术的逐步推广,导致数据计算量成倍增长,需要具备高性能计算能力的系统才能满足计算需要.提出基于边缘计算的理念,采用容器技术构建具备大规模计算资源池的分布式计算环境,实现了大地电磁法三维反演的分布式并行计算.基于松耦合的软件架构设计研发了大地电磁法三维反演分布式计算软件,实现了人机交互操作和反演结果可视化.

宽线地震采集原始数据相比二维数据具有更丰富的信息.为充分挖掘宽线数据本身的优势和潜力,提升宽线数据处理效果,本文根据宽线数据和二维及三维数据的异同,借鉴了目前"两宽一高"三维地震数据处理流程,主要研究了宽线有源噪音压制及浅海拖缆宽线大羽角数据共中心点离散处理技术,提出了观测系统定义方式需要根据不同处理模块的要求进行实时变换的建议,总结完善了针对复杂区宽线处理技术策略,提升了宽线的处理成果剖面质量,为复杂区宽线地震资料处理总结了一些经验.

提出一种基于多频多单站体制的均匀地下介质中二维目标自聚焦成像算法.在Born近似下采用半空间谱域格林函数和目标的爆炸点模型推导出地下目标反演成像公式,借助快速Fourier变换实现地下目标的快速实时成像,避开了耗时较多的大型病态矩阵的正则化求解.针对地下介质电磁参数未知的情况,根据时间反转成像的原理和最小熵准则,进一步提出了目标的自聚焦算法.通过熵计算选取最优聚焦时间确定最优的聚焦图像,修正由于地下介质电磁参数估计偏差引起的成像的目标位置偏移及图像退化.文中用数值仿真的结果证明了在地下介质电磁参数未知情况下,提出的算法能在较短的计算时间内得到好的成像结果.

The problem of imaging of 2-D dielectric objects embedded in a lossy earth is considered with unknown permittivity and conductivity of the lossy background. Under Born approximation, the spectral form of the Green's function in half-space is employed to formulate the half-space imaging algorithm with multi-frequency and multi-monostatic data. Hence the fast Fourier transform can be used to achieve real-time imaging in a very short computation time. Inspired by the principle of time reversed imaging, a refocusing time factor has been introduced into the refocusing formulation. The optimal focusing time can be determined according to minimum entropy criterion. At the optimal focusing time a satisfactory image can be obtained regardless of the unknown characteristics of the earth. Numerical results have shown that the proposed algorithm can provide a high quality focused image in a short time despite the inaccurate estimation of earth electric parameters.

The singular integral in the calculation of rough surface scattering is transferred into nonsingular function integral.It can improve the calculation accuracy.Comparison with other approximate algorithms is given.And the error of another common used approximate algorithm is analyzed.