Delineation of bed boundaries of array induction logging curves using deep learning
Zhang Lei 1,4, Wang Jian♦1,3,4, Jiao RuiLi 2, Hao Chen 1,3,4, Wang Xiu-Ming 1,3,4, and Ji You-Ming 2
1. State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China
2. Beijing Information Science & Technology University, Beijing 100192, China
3. University of Chinese Academy of Sciences, Beijing 100049, China
4. Beijing Engineering Research Center of Sea Deep Drilling and Exploration, Beijing 100190, China
Abstract Delineation of bed boundaries based on resistivity logging curves is important prior information for the inversion and interpretation of resistivity logging data. Traditionally, the layering algorithm mainly use the derivatives of resistivity curves or other logging methods as reference. However, measurement error or resolution mismatch may lead to misjudgment of the boundary. In view of the shortcomings of traditional methods, this paper presents an automatic layering algorithm of array induction logging curves based on deep learning. In this algorithm, a locally connected convolution neural network is used, and the generalization ability of the network is improved by enlarging the training set, optimizing the window length and threshold, and strengthening the layering effect. Simulation and field data show the effectiveness of the proposed algorithm.
This study was supported by the National Nature Science Foundation of China (No. 41604123)
Corresponding Authors: Wang Jian (Email: wangjian@mail.ioa.ac.cn).
E-mail: wangjian@mail.ioa.ac.cn
About author: Lei Zhang. Ph.D. graduated from the Institute of Acoustics, Chinese Academy of Sciences in 2012. His main research interests are electrical logging methods and principles, forward inversion algorithms, and machine learning algorithms. Email: Zhanglei@mail.ioa.ac.cn
Cite this article:
. Delineation of bed boundaries of array induction logging curves using deep learning[J]. APPLIED GEOPHYSICS, 2021, 18(1): 45-53.
2021, Vol. 18 
