王峣钧^1,,
邢凯²,
厍斌¹,
刘宇¹,
陈挺¹,
胡光岷¹,
吴秋波³
1. 电子科技大学资源与环境学院, 成都 611731
2. 太原理工大学求实学院, 太原 030024
3. 东方地球物理公司物探技术研究中心, 河北 涿州 072751

基金项目: 国家自然科学基金(41804126)及国家重大专项(2017ZX05018001-006)联合资助


详细信息

作者简介: 王峣钧, 男, 1987年生, 博士, 副教授, 主要从事地球物理人工智能方法和地球物理反演方法研究.E-mail: yaojun.wang@uestc.edu.cn

中图分类号: P631

收稿日期:2020-06-05
修回日期:2021-04-12
上线日期:2021-07-10

Mixed Gaussian stochastic inversion based on hybrid of cuckoo algorithm and Markov chain Monte Carlo

WANG YaoJun^1,,
XING Kai²,
SHE Bin¹,
LIU Yu¹,
CHEN Ting¹,
HU GuangMin¹,
WU QiuBo³
1. School of Resources and Environment, University of Electronic Science and Technology of China, Chengdu 611731, China
2. Qiushi College, Taiyuan University of Technology, Taiyuan 030024, China
3. Geophysical Exploration Technology R & D Center, BGP, Hebei Zhuozhou 072751, China


MSC: P631

--> Received Date: 05 June 2020
Revised Date: 12 April 2021
Available Online: 10 July 2021


摘要
摘要:地质统计学随机反演可以获得比常规反演更高分辨率的结果，目前已成为储层高分辨率预测的主流方法.地下不同岩相储层参数存在明显差异，本文在地质统计学反演框架下构建了岩相和储层参数同步反演目标函数，实现不同岩相条件下储层参数分布精细描述.在求解该高维数据多参数同步反演问题时，本文将可以动态调节搜索步长的布谷鸟算法与马尔科夫链蒙特卡洛方法融合，采用多条马尔科夫链进行Levy飞行产生新解的策略扩大解的空间范围，通过适应度最佳选择输出最优解实现全局优化迭代，有效提升了反演方法的稳定性和全局最优性，避免了传统马尔科夫链蒙特卡洛方法因抽样随机性而陷入局部最优的问题.通过含噪声模型和实际数据分析验证了本文方法的有效性.
关键词: 随机反演/
混合高斯/
MCMC/
布谷鸟算法/
全局优化

Abstract:Geostatistical stochastic inversion can obtain higher resolution results than conventional inversion. Considering that there are obvious differences in the parameters of different lithofacies in the underground, in this paper, we propose a new method to inverse the petrographic proportion, lithofacies classification and elastic parameters simultaneously, achieving a detailed description of the reservoir parameter distribution under different lithofacies. When considering the multi-parameter simultaneous inversion problem of high-dimensional data, the paper realizes the fusion of the cuckoo algorithm and the Markov chain Monte Carlo approach to solve this inversion problem. Our new algorithm uses multiple Markov chains to carry out Levy flight to generate a new solution strategy to expand the solution spatial range. Selecting the optimal solution through the best fitness to achieve the global optimization iteration. Through algorithm integration, the stability and global optimality of the inversion method are effectively improved. The effectiveness of the new method is verified by synthetic data and the field data.
Key words:Geostatistical stochastic inversion/
Mixed Gaussian model/
MCMC/
Cuckoo algorithm/
Global optimization



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=51b3d011-ecd9-4a77-8af4-38622f3f394f