Abstract:For an effective deep learning based seismic impedance inversion strategy, a deep convolutional network is trained by massive data-driven models to obtain the mapping between seismic data and impedance. After the network is pre-trained by substantial synthetic data, a small amount of real data is required for transfer learning of the network. In this paper, we propose a new method based on data augmentation and active learning for seismic wave impedance inversion. First, the original single-trace wave impedance data is augmented by resampling at the same frequency, and then the reflectivity and random kernel are calculated to generate the seismic data after augmentation. The augmented seismic and wave impedance data is taken as training sets, and the maximum-error samples are selected to train the network iteratively considering active learning. The proposed method can avoid seismic wavelet estimation, while training the network with higher accuracy using a small amount of label data. The test results from the Marmousi 2 model demonstrate that this method only needs one tenth of label data and iteration times to achieve the prediction accuracy similar to that of iterative random training, with the prediction errors distributed more evenly on the profile.
伊小蝶, 吴帮玉, 孟德林, 曹相湧. 数据增广和主动学习在波阻抗反演中的应用[J]. 石油地球物理勘探, 2021, 56(4): 707-715.
YI Xiaodie, WU Bangyu, MENG Delin, CAO Xiangyong. Application of data augmentation and active learning to seismic wave impedance inversion. Oil Geophysical Prospecting, 2021, 56(4): 707-715.
Russell B H. Introduction to Seismic Inversion Me-thods[M]. SEG,1988.
[2]
Araya-Polo M,Jennings J,Adler A,et al. Deep-lear-ning tomography[J]. The Leading Edge,2018,37(1):58-66.
[3]
Bergen K J,Johnson P A,De Hoop M V,et al. Machine learning for data-driven discovery in solid Earth geoscience[J]. Science,2019,363(6433):eaau0323.
[4]
Karpatne A,Ebert-Uphoff I,Ravela S,et al. Machine learning for the geosciences:Challenges and opportunities[J]. IEEE Transactions on Knowledge & Data Engineering,2017,31(8):1544-1554.
[5]
Yang F,Ma J. Deep-learning inversion:a next generation seismic velocity-model building method[J]. Geophysies,2019,84(4):R583-R589.
[6]
Wang Y,Ge Q,Lu W,et al. Well-logging constrained seismic inversion based on closed-loop convolutional neural network[J]. IEEE Transactions on Geosience and Remote Sensing,2020,58(8):5564-5574.
[7]
Fabien-Ouellet G,Sarkar R. Seismic velocity estimation:A deep recurrent neural-network approach[J]. Geophysics,2020,85(1):U21-U29.
[8]
Park M J,Sacchi M D. Automatic velocity analysis using convolutional neural network and transfer lear-ning[J]. Geophysics,2020,85(1):V33-V43.
[9]
Das V,Pollack A,Wollner U,et al. Convolutional neural network for seismic impedance inversion[J]. Geophysics,2019,84(6):R869-R880.
[10]
Mustafa A,Alfarraj M,Al Regib G. Estimation of acoustic impedance from seismic data using temporal convolutional network[C]. SEG Technical Program Expanded Abstracts,2019,38:2554-2557.
[11]
Du J,Liu J,Zhang G,et al. Pre-stack seismic inversion using SeisInv-ResNet[C]. SEG Technical Program Expanded Abstracts, 2019,38:2338-2342.
[12]
孙宇航,刘洋. 利用GRU神经网络预测横波速度[J]. 石油地球物理勘探,2020,55(3):484-492.SUN Yuhang,LIU Yang. Prediction of S-wave velo-city based on GRU neural network[J]. Oil Geophy-sical Prospecting,2020,55(3):484-492.
[13]
王俊,曹俊兴,尤加春. 基于GRU神经网络的测井曲线重构[J]. 石油地球物理勘探,2020,55(3):510-520.WANG Jun,CAO Junxing,YOU Jiachun. Reconstruction of logging traces based on GRU neural network[J]. Oil Geophysical Prospecting,2020,55(3):510-520.
[14]
Ali S,Mathias L,Felix J H. The importance of transfer learning in seismic modeling and imaging[J]. Geo-physics,2019,84(6):A47-A54.
[15]
张政,严哲,顾汉明. 基于残差网络与迁移学习的断层自动识别[J].石油地球物理勘探,2020,55(5):950-956.ZHANG Zheng,YAN Zhe,GU Hanming. Automatic fault recognition with residual network and transfer learning[J].Oil Geophysical Prospecting,2020,55(5):950-956.
[16]
Park M J,Sacchi M D. Automatic velocity analysis using convolutional neural network and transfer lear-ning[J]. Geophysics,2020,85(1):V33-V43.
[17]
Wu D,Lin C T,Huang J. Active learning for regression using greedy sampling[J]. Information Sciences,2019,474:90-105.
[18]
Cao X,Yao J,Xu Z,et al. Hyperspectral image classification with convolutional neural network and active learning[J]. IEEE Transactions on Geoscience and Remote Sensing,2020,58(7):4604-4616.
[19]
McKinley S,Levine M. Cubic spline interpolation[J]. College of the Redwoods,1998,45(1):1049-1060.
[20]
Gray R M. Toeplitz and circulant matrices:A review[J]. Foundations & Trends in Communications & Information Theory,2006,2(3):155-239.
[21]
汪玲玲,高静怀,赵谦,等. 基于矩阵Toeplitz稀疏分解的相对波阻抗反演方法[J]. 地球物理学报,2017,60(2):639-654.WANG Lingling,GAO Jinghuai,ZHAO Qian,et al. Relative acoustic impedance inversion via Toeplitz-sparse matrix factorization[J]. Chinese Journal of Geophysics,2017,60(2):639-654.
[22]
Wang K,Bandura L,Bevc D,et al. End-to-end deep neural network for seismic inversion[C]. SEG Technical Program Expanded Abstracts,2019,38:4983-4986.
[23]
梁光河. 地震子波提取方法研究[J]. 石油物探,1998,37(1):31-39.LIANG Guanghe. On the methods of seismic wavelet extraction[J]. Geophysical Prospecting for Petro-leum,1998,37(1):31-39.
[24]
Long J,Shelhamer E,Darrell T. Fully convolutional networks for semantic segmentation[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2015,3431-3440.
[25]
He K,Zhang X,Ren S,et al. Deep residual learning for image recognition[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016,770-778.
[26]
Wu B,Meng D,Wang L,et al. Seismic impedance inversion using fully convolutional residual network and transfer learning[J]. IEEE Geoscience and Remote Sensing Letters,2020,17(12):2140-2144
[27]
Xu B,Wang N,Chen T,et al. Empirical evaluation of rectified activations in convolutional network[CS/ML]. arxiv.org/abs/1505.00853,2015-05-05.
[28]
Ioffe S,Szegedy C. Batch normalization:Accelerating deep network training by reducing internal covariate shift[C]. Proceedings of the 32nd International Conference on Machine Learning,2015,448-456.
[29]
Keskar N S,Mudigere D,Nocedal J,et al. On large-batch training for deep learning:Generalization gap and sharp minima[C]. International Conference on Learning Representations,2017,1-16.
[30]
Kingma D P,Ba J. Adam:A method for stochastic optimization[CS/ML]. arxiv.org/abs/1412.6980,2014-12-22.
[31]
Martin G S,Wiley R,Marfurt K J. Marmousi 2:An elastic upgrade for Marmousi[J]. The Leading Edge,2006,25(2):156-166.