摘要
基于匹配追踪的平滑伪Wigner-Ville分布具有良好的时频分辨率,但由于计算量过大,影响其在实际地震数据处理中的应用。在深入研究基于OpenMP的并行算法的基础上,本文将该并行计算方式应用于匹配追踪法对Morlet子波的寻优过程以及计算多个Morlet子波平滑伪Wigner-Ville分布。实际数据计算结果表明:该并行算法加速性能良好,加速比随线程数增加呈近线性增长,即在确保计算精度的情况下,能显著提高计算效率;而且计算量越大其加速性能体现越充分。
The smoothed-pseudo Wigner-Ville distribution(SPWVD)based on matching pursuit has a high time-frequency resolution,but the calculation is extremely large in the matching pursuit algorithm and SPWVD calculation of all the wavelets.To improve the efficiency of this method,we propose a parallel computing method for the matching pursuit time-frequency analysis based on OpenMP.The searching process in the matching pursuit algorithm and the SPWVD calculation of all the wavelets are achieved in parallel.Field seismic data processing shows that the efficiency of the proposed parallel algorithm is significantly improved,and its computation precision is the same as the serial algorithm.Parallel calculation performance shows that the speedup grows linearly as the number of processors increases,and larger computation amount can fully exploit the advantage of multicore processors.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2018年第3期454-461,共8页
Oil Geophysical Prospecting
基金
中国地质科学院基本科研业务费项目"深反射地震数据处理技术研究"(YYWF201506)
国家重点研发计划课题"年尺度和短临强震发震紧迫程度判定技术研究"(2017YFC1500502)
川滇国家地震监测预报实验场项目"四川长宁地震窗注水影响和机理研究"(2016CESE0101)联合资助
