赵鹏飞^1,2,3,,
刘财^1,2,3,4,,,
冯晅^1,4,5,
郭智奇¹,
阮庆丰¹
1. 吉林大学地球探测科学与技术学院, 长春 130026
2. 吉林大学应用地球物理实验教学中心, 长春 130026
3. 吉林大学地质资源立体探测虚拟仿真实验教学示范中心, 长春 130026
4. 国土资源部应用地球物理重点实验室, 长春 130026
5. 油页岩地下原位转化与钻采技术国家地方联合工程实验室, 长春 130026

基金项目: 国家自然科学基金重点项目（41430322），国家重点研发计划项目（2016YFC0600505），吉林大学高层次科技创新团队建设项目，中央高校基本科研业务费专项，国家自然科学基金青年项目（41304087）资助


详细信息

作者简介: 赵鹏飞, 男, 1981年生, 博士, 副教授, 主要从事计算地球物理方面的研究.E-mail:zhaopf@jlu.edu.cn

通讯作者: 刘财, 男, 1963年生, 教授, 博士生导师, 主要从事信号处理与分析和勘探地震学研究.E-mail:liucai@jlu.edu.cn

中图分类号: P631

收稿日期:2018-02-01
修回日期:2019-01-08
上线日期:2019-03-05

Stochastic seismic inversion based on neural network

ZHAO PengFei^1,2,3,,
LIU Cai^1,2,3,4,,,
FENG Xuan^1,4,5,
GUO ZhiQi¹,
RUAN QingFeng¹
1. College of Geo-exploration Science and Technology, Jilin University, Changchun 130026, China
2. Central Lab of Applied Geophysics, Changchun 130026, China
3. Virtual Simulation Experiment Teaching Center for Stereoscopic Exploration of Geological Resources, Changchun 130026, China
4. Key Laboratory of Applied Geophysics, Ministry of Land and Resources, Changchun 130026, China
5. National-Local Joint Engineering Laboratory of In-situ Conversion, Drilling and Exploitation Technology for Oil Shale, Changchun 130026, China


More Information

Corresponding author: LIU Cai,E-mail:liucai@jlu.edu.cn

MSC: P631

--> Received Date: 01 February 2018
Revised Date: 08 January 2019
Available Online: 05 March 2019


摘要
摘要:针对随机地震反演中存在的两个主要问题，随机实现含有噪声和难以从大量随机实现中挖掘有效信息，提出了一种基于神经网络的随机地震反演方法.通过对多组随机实现及其正演地震数据的计算，构建了基于序贯高斯模拟的训练集.这也为应用神经网络求解地球物理反问题，提供了一种有效建立训练集的方法.较之传统的神经网络反演，这种训练集不仅保证了学习样本具有多样性，同时还引入了空间相关性.数值模拟结果表明，该方法只需要通过单层前馈神经网络，就可以比较有效的解决一个500个阻抗参数的反演问题.
关键词: 随机地震反演/
序贯高斯模拟/
神经网络/
训练集

Abstract:In view of two principal problems in stochastic seismic inversion, stochastic realizations with massive noise and being difficult to dig effective information out of mass of realizations, we propose a novel stochastic seismic inversion method based on neural network.Through the simulation of stochastic realizations and the corresponding seismic forward modeling, we build the training sets based on Sequential Gaussian Simulation (SGS).It provides an effective method to establish training sets for neural network in solving geophysical inverse problems.Compared with traditional neural network algorithms, such training sets not only have the diversity of the learning samples, but also possess the spatial correlation.The numerical simulation results show that with the aid of the proposed method, we can solve an impedance inversion problem with 500 parameters by using a feed forward neural network with only one hidden layer.
Key words:Stochastic seismic inversion/
Sequential Gaussian Simulation/
Neural network/
Training set



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=dqwlxb_14919