Deep prestack seismic wavelets extraction in Tarim based on generative adversarial network
-
摘要:
深层超深层叠前地震子波的准确提取可有效提高全波形反演及叠前偏移成像等方法的准确性，对储层预测和油气分析具有重要意义.针对塔里木低信噪比叠前地震记录，本文提出一种基于生成对抗网络（GAN）的塔里木深层超深层叠前地震子波提取方法.结合柯东地区地质构造特点及噪声特性，采用有限差分法正演叠前地震记录，构建了塔里木柯东地区叠前地震记录及对应子波的训练样本集，利用地震子波、反射系数及噪声特征的区别训练生成对抗网络，实现了从叠前地震记录中有效分离地震子波.合成数据和实际地震资料处理结果表明，本文所提方法对塔里木深层超深层叠前地震子波提取具有有效性和较高的准确性，较传统方法更具优越性.
Abstract:The accurate extraction of deep and ultra-deep underground prestack seismic wavelets can effectively improve the accuracy of seismic data processing methods such as full waveform inversion and prestack migration imaging. It has great significance to reservoir prediction and oil and gas analysis. Aiming at the low signal-to-noise ratio pre-stack seismic records of Tarim, we propose a method for extracting the deep and ultra-deep underground prestack seismic wavelets from Tarim based on the Generative Adversarial Network (GAN). Combining the characteristics of geological structure and noise in the Kedong area, we used the finite difference method to forward pre-stack seismic traces to construct the training sets of pre-stack seismic traces and corresponding seismic wavelets in the Kedong area of Tarim. We train the GAN by using seismic wavelets, reflection coefficients and noise distinguishing features. Finally we achieved effective separation of seismic wavelets from pre-stack seismic traces. The processing results of synthetic data and actual seismic data show that the method proposed in this paper is effective and accurate for prestack seismic wavelets extraction in deep and ultra-deep underground in Tarim. And our method is more superior to traditional methods.
-
-
图 3
生成对抗网络的学习过程图
Figure 3.
Diagram of the learning process of generating a confrontation network
图 4
生成对抗网络的训练流程图
Figure 4.
The training flowchart of the confrontation generation network
图 9
叠前地震记录反褶积结果对比图
Figure 9.
Comparison of deconvolution results of pre-stack seismic records
表 1
传统谱模拟算法与生成对抗网络算法的叠前地震子波提取速度对比
Table 1.
Comparison of pre-stack seismic wavelet extraction speed between traditional spectrum simulation algorithm and GAN
子波提取方法 子波提取速度/(1000道·s-1) 传统谱模拟算法 23088.22 变分模态分解算法 39587.12 生成对抗网络算法 1225.66 
下载: 导出CSV
-
Chen D W, Yang W Y, Wei X J, et al. 2021. Research on first-break automatic picking based on an improved U-Net network. Progress in Geophysics (in Chinese), 36(4): 1493-1503, doi:10.6038/pg2021EE0235.
Chen D Y, Gao J H, Hou Y P, et al. 2019. High resolution inversion of seismic wavelet and reflectivity using iterative deep neural networks. //89th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts, 2538-2542.
Côrte G, Dramsch J, Amini H, et al. 2020. Deep neural network application for 4D seismic inversion to changes in pressure and saturation: Optimizing the use of synthetic training datasets. Geophysical Prospecting, 68(7): 2164-2185. doi: 10.1111/1365-2478.12982
Das V, Pollack A, Wollner U, et al. 2018. Convolutional neural network for seismic impedance inversion. //88th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts, 869-880.
Du Z L, Liang H, Shi J, et al. 2013. Cenozoic structural deformation and hydrocarbon exploration of Kedong structure in the piedmont of western Kunlun mountain. Acta Petrolei Sinica (in Chinese), 34(1): 23-29.
Goodfellow I J, Pouget-Abadie J, Mirza M, et al. 2014. Generative adversarial nets. //Proceedings of the 27th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2672-2680.
Grubas S I, Loginov G N, Duchkov A A. 2020. Traveltime-table compression using artificial neural networks for Kirchhoff-migration processing of microseismic data. Geophysics, 85(5): U121-U128.
Kaur H, Pham N, Fomel S. 2019. Seismic data interpolation using CycleGAN. //89th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts, 2202-2206.
Kim Y, Nakata N. 2018. Geophysical inversion versus machine learning in inverse problems. The Leading Edge, 37(12): 894-901. doi: 10.1190/tle37120894.1
Li Q X, Luo Y N. 2019. Using GAN priors for ultrahigh resolution seismic inversion. //89th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts, 2453-2457.
Lin N T, Zhang D, Zhang K, et al. 2018. Predicting distribution of hydrocarbon reservoirs with seismic data based on learning of the small-sample convolution neural network. Chinese Journal of Geophysics (in Chinese), 61(10): 4110-4125, doi:10.6038/cjg2018J0775.
Lü Y, Shan X C, Huo S D, et al. 2020. Local SNR estimation of seismic data based on deep convolutional neural network. Chinese Journal of Geophysics (in Chinese), 63(1): 320-328, doi:10.6038/cjg2020N0058.
Mao W T, Liu Y M, Ding L, et al. 2019. Imbalanced fault diagnosis of rolling bearing based on generative adversarial network: a comparative study. IEEE Access, 7: 9519-9530.
Nair V, Hinton G E. 2010. Rectified linear units improve restricted boltzmann machines. //Proceedings of the 27th International Conference on International Conference on Machine Learning. Madison: Omnipress, 807-814.
Rosa A L R, Ulrych T J. 1991. Processing via spectral modeling. Geophysics, 56(8): 1244-1251. doi: 10.1190/1.1443144
Siahkoohi A, Louboutin M, Kumar R, et al. 2018. Deep-Convolutional Neural Networks in prestack seismic-two exploratory examples. //88th Ann. Internat Mtg., Soc. Expi. Geophys. . Expanded Abstracts, 2196-2200.
Wang H, Yan J Y, Fu G M, et al. 2020. Current status and application prospect of deep learning in geophysics. Progress in Geophysics (in Chinese), 35(2): 642-655, doi:10.6038/pg2020CC0476.
Wang L X, Mendel J M. 1991. Adaptive minimum prediction-error deconvolution and wavelet estimation using Hopfield neural networks. //1991 IEEE International Conference on Acoustics, Speech, and Signal Processing. Toronto, ON, Canada: IEEE, 2969-2972.
Wang S X, Yuan S Y, Ma M, et al. 2015. Wavelet phase estimation using ant colony optimization algorithm. Journal of Applied Geophysics, 122: 159-166. doi: 10.1016/j.jappgeo.2015.09.013
Wang W L, Ma J W. 2020. Velocity model building in acrosswell acquisition geometry with image-trained artificial neural networks. Geophysics, 85(2): U31-U46. doi: 10.1190/geo2018-0591.1
Zeng C M, Du Z L, Zhou X H, et al. 2011. Structural Characteristics and Deformation Mechanism of Kedong Thrust Belt in Piedmont of Kunlun Mountain. Xinjiang Petroleum Geology (in Chinese), 32(2): 133-136.
Zhang F C, Liu H Q, Niu X M, et al. 2014. High resolution seismic inversion by convolutional neural network. Oil Geophysical Prospecting (in Chinese), 49(6): 1165-1169.
陈德武, 杨午阳, 魏新建等. 2021. 一种基于改进的U-Net网络的初至自动拾取研究. 地球物理学进展, 36(4): 1493-1503, doi:10.6038/pg2021EE0235.
杜治利, 梁瀚, 师骏等. 2013. 西昆仑山前柯东构造新生代构造变形及油气意义. 石油学报, 34(1): 22-29. https://www.cnki.com.cn/Article/CJFDTOTAL-SYXB201301002.htm
林年添, 张栋, 张凯等. 2018. 地震油气储层的小样本卷积神经网络学习与预测. 地球物理学报, 61(10): 4110-4125, doi:10.6038/cjg2018J0775. http://www.geophy.cn/article/doi/10.6038/cjg2018J0775
吕尧, 单小彩, 霍守东等. 2020. 基于深度卷积神经网络的地震数据局部信噪比估计. 地球物理学报, 63(1): 320-328, doi:10.6038/cjg2020N0058. http://www.geophy.cn/article/doi/10.6038/cjg2020N0058
王昊, 严加永, 付光明等. 2020. 深度学习在地球物理中的应用现状与前景. 地球物理学进展, 35(2): 642-655, doi:10.6038/pg2020CC0476.
曾昌民, 杜治利, 周学慧等. 2011. 昆仑山前柯东构造带构造解析及形成机制. 新疆石油地质, 32(2): 133-136. https://www.cnki.com.cn/Article/CJFDTOTAL-XJSD201102012.htm
张繁昌, 刘汉卿, 钮学民等. 2014. 褶积神经网络高分辨率地震反演. 石油地球物理勘探, 49(6): 1165-1169. https://www.cnki.com.cn/Article/CJFDTOTAL-SYDQ201406025.htm
-
图(9)
表(1)
计量
- 文章访问数:
- PDF下载数:
- 施引文献: 0