Please wait a minute...
E-mail Alert Rss
 
物探与化探  2020, Vol. 44 Issue (2): 290-299    DOI: 10.11720/wtyht.2020.1460
  方法研究·信息处理·仪器研制 本期目录 | 过刊浏览 | 高级检索 |
基于MPI+OpenMP的时间域航空电磁快速正演算法
任运通, 李貅, 齐彦福, 曹华科
长安大学 地质工程与测绘学院,陕西 西安 710054
Research on time domain airborne electromagnetic fast forward algorithm based on MPI+OpenMP
Yun-Tong REN, Xiu LI, Yan-Fu QI, Hua-Ke CAO
College of Geological Engineering and Geomatics, Chang'an University, Xi'an 710054, China
全文: PDF(5724 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 

时间域有限元算法已被广泛应用于航空电磁三维正演模拟当中,然而由于航空电磁测区面积大,且采样密集,造成正演计算量巨大,传统的串行算法已经无法满足计算效率要求,为此,开展了并行加速算法研究以解决计算效率不足的问题。基于航空电磁系统的影响范围有限,采用局部网格技术将计算任务划分成多个子网格,即每个发射源一套网格,各网格的正演计算相互独立,不存在数据依赖性,具有很好的可并行性;利用MPI技术对多个子网格正演任务进行分配,在各个进程上进行并行计算;针对每个正演子网格,在进行时间域有限元算法正演模拟过程中,采用OpenMP技术对单元矩阵进行并行计算。典型地电模型的数值模拟结果表明本文开发的MPI+OpenMP并行正演算法可以有效提高正演速度,最高加速比可达10倍。

服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
任运通
李貅
齐彦福
曹华科
关键词 时间域航空电磁MPIOpenMP有限元    
Abstract

The time domain finite element algorithm has been widely used in airborne electromagnetic three-dimensional forward modeling. However, due to the large airborne electromagnetic measurement area and dense sampling, the forward calculation is huge, and the traditional serial algorithm cannot meet the calculation efficiency requirements. In view of such a situation, the authors carried out parallel acceleration algorithm research to solve the problem of insufficient computational efficiency. Based on the limited range of influence of aviation electromagnetic system, the local grid technology is used to divide the computing task into multiple sub-grids, that is, a set of grids for each source, and the forward calculations of each grid are independent of each other, and hence no data dependency is existent and there is good parallelism. In this paper, MPI technology is used to allocate multiple sub-grid forward tasks, and parallel computing is performed on each process. For each forward subgrid, in the forward modeling of the time domain finite element algorithm, the parallel matrix is calculated by OpenMP technology. The numerical simulation results of the typical electrical model show that the MPI+OpenMP parallel forward algorithm developed in this paper can effectively improve the forward speed, and the maximum acceleration ratio can reach 10 times.

Key wordstime-domain airborne EM    MPI    OpenMP    finite element
收稿日期: 2019-09-24      出版日期: 2020-04-22
ZTFLH:  P631  
基金资助:国家重点研发计划项目(2017YFC1502605);国家自然科学基金重点项目(41830101)
通讯作者: 李貅   
作者简介: 任运通(1995-),男,长安大学硕士,研究方向为瞬变电磁探测
引用本文:   
任运通, 李貅, 齐彦福, 曹华科. 基于MPI+OpenMP的时间域航空电磁快速正演算法[J]. 物探与化探, 2020, 44(2): 290-299.
Yun-Tong REN, Xiu LI, Yan-Fu QI, Hua-Ke CAO. Research on time domain airborne electromagnetic fast forward algorithm based on MPI+OpenMP. Geophysical and Geochemical Exploration, 2020, 44(2): 290-299.
链接本文:  
http://www.wutanyuhuatan.com/CN/10.11720/wtyht.2020.1460      或      http://www.wutanyuhuatan.com/CN/Y2020/V44/I2/290
Fig.1  航空系统影响范围示意
Fig.2  全局网格与局部网格对比
(a为全局网格;b、c、d为局部网格)
Fig.3  OpenMP并行模式示意
Fig.4  MPI并行模式示意
Fig.5  航空瞬变电磁三维正演并行计算流程
Fig.6  水平板状体模型示意
Fig.7  全局网格串行计算和局部网格并行计算电磁响应结果对比
程序类型 进程数 计算时间/s 加速比 并行效率
串行 1 2136.4303 1
1 2139.7464 0.998 99.85%
3 745.2973 2.867 95.55%
并行 5 499.5326 4.277 85.54%
7 372.3317 5.738 81.97%
12 266.8348 8.007 66.72%
16 210.9645 10.127 63.29%
Table 1  基于MPI+OpenMP并行加速策略的行正演计算统计
Fig.8  不同进程数的加速比(a)和并行效率(b)
Fig.9  复杂模型计算
网格单元数 消耗内存 分解次数 回代次数 计算总耗时/h
串行
并行
约90 000 约5 350M
5.2G
9 350 327 250 32.8
3.24
Table 2  计算情况统计
Fig.10  复杂模型主剖面(x=0测线)多测道曲线
Fig.11  复杂地形响应曲面
[1] Pemberton R H . Airborne electromagnetics in review[J]. Geophysics, 1962,27(5):691-713.
[2] Dobrin M . Introduction to Geophysical Prospecting[M]. 3ed.New York: McGraw-Hill, 1976.
[3] Palacky G J . The airborne electromagnetic method as a tool of geological mapping[J]. Geophysical Prospecting, 2006,29(1):60-88.
[4] Wynn J C . Evaluating groundwater in arid lands using airborne magnetic/EM methods: An example in the southwestern U. S. and northern Mexico[J]. The Leading Edge, 2002,21(1):62-64.
[5] Smith R. Airborne electromagnetic methods: applications to minerals, water and hydrocarbon exploration [C]// CSEG 2010 Distinguished Lecture, 2010: 7-10.
[6] 殷长春, 张博, 刘云鹤 , 等. 航空电磁勘查技术发展现状及展望[J]. 地球物理学报, 2015,58(8):2637-2653.
[6] Yin C C, Zhang B, Liu Y H , et al. Development status and prospects of aviation electromagnetic exploration technology[J]. Geophysics, 2015,58(8):2637-2653.
[7] Mogi T . Three-dimensional modeling of magnetotelluric data using finite element method[J]. J. Appl. Geophys., 1996,35(2-3):185-189.
[8] Knight J H, Raich A P . Transient electromagnetic calculations using the Gaver-Stehfest inverse Laplace transform method[J]. Geophysics, 1982,47(1):47-50, doi: 10.1190/1.1441280.
[9] 考夫曼 A A, 凯勒 G V . 频率域和时间域电磁测深[M]. 王建谋译. 北京: 地质出版社, 1987.
[9] Kaufman A A, Keller G V. Frequency domain and time domain electromagnetic sounding[M].Wang J M translation. Beijing: Geological Publishing House, 1987.
[10] 殷长春, 张博, 刘云鹤 , 等. 2.5维起伏地表条件下时间域航空电磁正演模拟[J]. 地球物理学报, 2015,58(4):1411-1424.
[10] Yin C C, Zhang B, Liu Y H , et al. Simulation of time domain aeromagnetic forward modeling under 2.5-dimensional undulating surface conditions[J]. Chinese Journal of Geophysics, 2015,58(4):1411-1424.
[11] Um E S, Harris J M, Alumbaugh D L . 3D time-domain simulation of electromagnetic diffusion phenomena: A finite-element electric-field approach[J]. Geophysics, 2010,75(4):F115-F126, doi: 10.1190/1.3473694.
[12] 李贺 . 直接时间域矢量有限元瞬变电磁三维正演模拟[D]. 西安:长安大学, 2016.
[12] Li H . Direct time domain vector finite element transient electromagnetic three-dimensional forward modeling [D]. Xi'an: Chang’an University, 2016.
[13] 齐彦福, 殷长春, 刘云鹤 , 等. 基于瞬时电流脉冲的三维时间域航空电磁全波形正演模拟[J]. 地球物理学报, 2017,60(1):369-382.
[13] Qi Y F, Yin C C, Liu Y H , et al. A three-dimensional time domain aeromagnetic electromagnetic full waveform forward modeling based on instantaneous current pulse[J]. Chinese Journal of Geophysics, 2017,60(1):369-382.
[14] 齐彦福 . 复杂介质中时间域航空电磁数据仿真技术研究[D]. 长春:吉林大学, 2017.
[14] Qi Y F . Research on time domain aeronautical electromagnetic data simulation technology in complex media[D]. Changchun:Jilin University, 2017.
[15] Newman G A, Alumbaugh D L . Three-dimensional massively parallel electromagnetic inversion[J]. Geophysics, 1997,128:345-354.
[16] 荣莹, 曹俊兴 . 基于MPI的机群并行计算系统平台构建[J]. 物探化探计算技术, 2005,27(1):89-91,100.
[16] Rong Y, Cao J X . The Construction of MPI-based cluster parallel computing system platform[J]. Computing Geophysical and Geochemical Exploration Technology, 2005,27(1):89-91,100.
[17] Tan H D, Yan T, Lin C H . Research on parallel algorithm of magnetotelluric 3D forward modeling[J].Applied Geophysics, 2006(4):197-202+261.
[18] 李小康 . 基于MPI的频率域航空电磁法有限元二维正演并行计算研究[D]. 北京:中国地质大学, 2011.
[18] Li X K . Research on finite element two-dimensional forward modeling parallel computing based on MPI in frequency domain aeromagnetic method[D]. Beijing: China University of Geosciences, 2011.
[19] 陈辉 . 基于MPI的航空瞬变电磁一维正反演[D]. 成都:成都理工大学, 2015.
[19] Chen H . MPI-based aeronautical transient electromagnetic one-dimensional forward and inversion[D]. Changdu:Chengdu University of Technology, 2015.
[20] Yang D, Oldenburg D W, Haber E . 3-D inversion of airborne electromagnetic data parallelized and accelerated by local mesh and adaptive soundings[J]. Geophysical Journal International, 2014,196(3):1492-1507.
[21] Ward S H, Hohmann G W. Electromagnetic theory for geophysical applications [C]//Electromagnetic Methods in Applied Geophysics, 1988,1(3):131-311.
[22] Nédélec J C . Mixed finite elements in 3[J]. Numerische Mathematik, 1980,35(3):315-341.
[23] Jin J M . The finite element method in electromagnetics, 2nd edn[M]. John Wiler and Sons, 2002.
[24] Jahandari H, Farquharson C G . A finite-volume solution to the geophysical electromagnetic forward problem using unstructured grids[J]. Geophysics, 2014,79(6):E287-E302.
[1] 贲放,黄威,路宁,韩飞,郑红闪,丁志强,李军峰. 时间域航空电磁的天电噪声去除研究[J]. 物探与化探, 2020, 44(2): 388-393.
[2] 黄威,贲放,吴珊,孙思源,廖桂香,西永在. 正交多项式法在航空电磁运动噪声去除中的应用[J]. 物探与化探, 2019, 43(4): 892-898.
[3] 孙大利,李貅,齐彦福,孙乃泉,李文忠,周建美,孙卫民. 基于非结构网格三维有限元堤坝隐患时移特征分析[J]. 物探与化探, 2019, 43(4): 804-814.
[4] 俞岱, 孙渊, 路婧, 王颖, 边瑞峰. 层初至波旅行时层析并行算法及在地裂缝调查中的应用[J]. 物探与化探, 2017, 41(5): 977-985.
[5] 孙栋华, 李怀渊, 江民忠, 王培建. 利用时间域航空电磁资料再论华北地台北界的划分[J]. 物探与化探, 2017, 41(3): 478-483.
[6] 任志平, 李貅, 戚志鹏, 赵威, 智庆全, 刘磊. 地面核磁共振三维响应影响因素[J]. 物探与化探, 2017, 41(1): 92-97.
[7] 骆燕, 江民忠, 宁媛丽, 彭莉红, 朱琳. 不同类型低阻异常航电时间常数的特征分析[J]. 物探与化探, 2016, 40(5): 991-997.
[8] 李军峰, 李文杰, 刘俊杰. 一种新型的组合波航空电磁脉冲发射系统[J]. 物探与化探, 2014, 38(6): 1186-1189,1199.
[9] 西永在, 吴珊, 廖桂香, 李文杰. 时间域航空电磁响应数值模拟的精度分析[J]. 物探与化探, 2014, 38(6): 1190-1194.
[10] 严波, 刘颖, 叶益信. 基于对偶加权后验误差估计的2.5维直流电阻率自适应有限元正演[J]. 物探与化探, 2014, 38(1): 145-150.
[11] 张斌, 谭捍东. 带地形的可控源音频大地电磁法二维正演[J]. 物探与化探, 2014, 38(1): 151-156.
[12] 田培培, 冯雪, 关珊珊, 万玲, 尹炳琪, 嵇艳鞠. 时域航空中心回线源 三维异常体的电磁响应探测分辨率[J]. 物探与化探, 2013, 37(3): 538-542.
[13] 苏洲, 胡文宝. 二维大地电磁正演中的无网格算法[J]. 物探与化探, 2012, 36(6): 1024-1028,1039.
[14] 李长伟, 熊彬, 吕玉增. 电法测井的三维有限元模拟[J]. 物探与化探, 2012, 36(4): 585-590.
[15] 鄂国庆, 徐英哲, 李文杰. 固定翼时间域航空电磁系统的飞机选型问题[J]. 物探与化探, 2012, 36(4): 595-597.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
京ICP备05055290号-3
版权所有 © 2017《物探与化探》编辑部
通讯地址:北京市学院路29号航遥中心 邮编:100083
电话:010-62060192;62060193 E-mail:whtbjb@sina.com