基于深度学习的台站远震波形规则化研究

由于地表条件和经济成本等因素的限制，地震台站的空间分布具有稀疏和不规则的特征。相比于偏移成像算法的要求，台站间距太大，无法满足成像精度要求。因此，进行偏移成像前需要对不规则台站数据进行规则化处理。

本论文提出了一种基于深度学习（deep learning，DL）的方法，用于二维测线观测的远震台站波形数据规则化。本文设计了一个基于深度生成模型和残差网络（residual network，ResNet）的地震波形规则化网络（earthquake waveform regularization network，EWR-Net），该网络由转置卷积块、卷积块和全连接块组成。卷积块中的ResNet层数是可变的，可以根据不同的数据复杂度进行调整，以提高网络适应性。该网络能够正确捕捉台站位置和对应波形之间的复杂映射，用于随机缺失和规则缺失数据的规则化处理，且无需进行空间平滑。它使用不规则台站记录的远震波形进行训练，然后用于预测缺失台站的波形。为避免模型过拟合，本文采用了L2正则化和早停法，并在全连接块中添加了dropout层。

本文首先使用区域地壳速度模型进行了正演模拟，利用合成数据训练模型并评价规则化效果。信噪比被用于评价整体的规则化准确度，对波形的时频变换分析用于确定预测波形在走时和振幅上的预测精度。随后，规则化数据被用于转换波逆时偏移成像，以检验方法的有效性。本文使用长测线下的台站实测数据进行了规则化应用，以检验方法的实用性。合成数据和实测数据的测试结果表明，EWR-Net能够在缺失台站生成比Curvelet方法更准确的地震波形，且逆时偏移成像结果具有更高的分辨率和准确度。本文使用不同台站数据类型确定了网络结构调整策略，并讨论了模型的适应性和适用范围。该研究证明了这种新的DL网络设计思想用于不规则台站数据的规则化是可行的，这对于应用先进转换波逆时偏移成像方法，基于越来越多的台阵远震数据，确定地球深部复杂速度界面具有重要意义。

[1] 仝中飞. Curvelet阈值迭代法在地震数据去噪和插值中的应用研究[D]. 吉林大学, 2009.
[2] 丁超. 被动源转换波逆时偏移和角度域共成像点道集的提取[D/OL]. 中国科学技术大学, 2019.
[3] CHEN Y, BAI M, CHEN Y. Obtaining free USArray data by multi-dimensional seismic reconstruction[J/OL]. Nature Communications, 2019, 10(1): 4434. DOI:10.1038/s41467-019-12405-0.
[4] SHANG X, DE HOOP M V, VAN DER HILST R D. Common conversion point stacking of receiver functions versus passive-source reverse time migration and wavefield regularization[J/OL]. Geophysical Journal International, 2017, 209(2): 923-934. DOI:10.1093/gji/ggx069.
[5] NAGHIZADEH M, SACCHI M D. f-x adaptive seismic-trace interpolation[J/OL]. GEOPHYSICS, 2009, 74(1): V9-V16. DOI:10.1190/1.3008547.
[6] SPITZ S. Seismic trace interpolation in the F-X domain[J/OL]. GEOPHYSICS, 1991, 56(6): 785-794. DOI:10.1190/1.1443096.
[7] WANG Y. Seismic trace interpolation in the f‐x‐y domain[J/OL]. GEOPHYSICS, 2002, 67(4): 1232-1239. DOI:10.1190/1.1500385.
[8] LIU G, CHEN X. Seismic data interpolation using frequency-domain complex nonstationary autoregression: f-x RNA seismic interpolation[J/OL]. Geophysical Prospecting, 2018, 66(3): 478-497. DOI:10.1111/1365-2478.12499.
[9] FOMEL S. Seismic reflection data interpolation with differential offset and shot continuation[J/OL]. GEOPHYSICS, 2003, 68(2): 733-744. DOI:10.1190/1.1567243.
[10] RONEN J. Wave‐equation trace interpolation[J/OL]. GEOPHYSICS, 1987, 52(7): 973-984. DOI:10.1190/1.1442366.
[11] KAPLAN S T, NAGHIZADEH M, SACCHI M D. Data reconstruction with shot-profile least-squares migration[J/OL]. GEOPHYSICS, 2010, 75(6): WB121-WB136. DOI:10.1190/1.3478375.
[12] 张雪敏. 基于低秩补全的天然地震数据重建[D]. 中国地质大学, 2019.
[13] CHEN Y, HUANG W, ZHANG D, 等. An open-source Matlab code package for improved rank-reduction 3D seismic data denoising and reconstruction[J/OL]. Computers & Geosciences, 2016, 95: 59-66. DOI:10.1016/j.cageo.2016.06.017.
[14] OROPEZA V, SACCHI M. Simultaneous seismic data denoising and reconstruction via multichannel singular spectrum analysis[J/OL]. GEOPHYSICS, 2011, 76(3): V25-V32. DOI:10.1190/1.3552706.
[15] CHEN Y, ZHANG D, JIN Z, 等. Simultaneous denoising and reconstruction of 5-D seismic data via damped rank-reduction method[J/OL]. Geophysical Journal International, 2016, 206(3): 1695-1717. DOI:10.1093/gji/ggw230.
[16] GAO J, SACCHI M D, CHEN X. A fast reduced-rank interpolation method for prestack seismic volumes that depend on four spatial dimensions[J/OL]. GEOPHYSICS, 2013, 78(1): V21-V30. DOI:10.1190/geo2012-0038.1.
[17] TRICKETT S, BURROUGHS L, MILTON A, 等. Rank‐reduction‐based trace interpolation[C/OL]//SEG Technical Program Expanded Abstracts 2010. Society of Exploration Geophysicists, 2010: 3829-3833. DOI:10.1190/1.3513645.
[18] GAO J, STANTON A, SACCHI M D. Parallel matrix factorization algorithm and its application to 5D seismic reconstruction and denoising[J/OL]. GEOPHYSICS, 2015, 80(6): V173-V187. DOI:10.1190/geo2014-0594.1.
[19] KREIMER N, SACCHI M D. A tensor higher-order singular value decomposition for prestack seismic data noise reduction and interpolation[J/OL]. GEOPHYSICS, 2012, 77(3): V113-V122. DOI:10.1190/geo2011-0399.1.
[20] DUIJNDAM A J W, SCHONEWILLE M A, HINDRIKS C O H. Reconstruction of band‐limited signals, irregularly sampled along one spatial direction[J/OL]. GEOPHYSICS, 1999, 64(2): 524-538. DOI:10.1190/1.1444559.
[21] TRAD D. Five-dimensional interpolation: Recovering from acquisition constraints[J/OL]. GEOPHYSICS, 2009, 74(6): V123-V132. DOI:10.1190/1. 3245216.
[22] HENNENFENT G, FENELON L, HERRMANN F J. Nonequispaced Curvelet transform for seismic data reconstruction: A sparsity-promoting approach[J/OL]. GEOPHYSICS, 2010, 75(6): WB203-WB210. DOI:10.1190/1.3494032.
[23] HERRMANN F J, HENNENFENT G. Non-parametric seismic data recovery with Curvelet frames[J/OL]. Geophysical Journal International, 2008, 173(1): 233-248. DOI:10.1111/j.1365-246X.2007.03698.x.
[24] NAGHIZADEH M, SACCHI M D. Beyond alias hierarchical scale Curvelet interpolation of regularly and irregularly sampled seismic data[J/OL]. Geophysics, 2010, 75(6): WB189-WB202. DOI:10.1190/1.3509468.
[25] SHAHIDI R, TANG G, MA J, 等. Application of randomized sampling schemes to Curvelet-based sparsity-promoting seismic data recovery: Application of randomized sampling schemes to seismic data recovery[J/OL]. Geophysical Prospecting, 2013, 61(5): 973-997. DOI:10.1111/1365-2478.12050.
[26] CHEN Y, FOMEL S, HU J. Iterative deblending of simultaneous-source seismic data using seislet-domain shaping regularization[J/OL]. GEOPHYSICS, 2014, 79(5): V179-V189. DOI:10.1190/geo2013-0449.1.
[27] GAN S, WANG S, CHEN Y, 等. Seismic data reconstruction via fast projection onto convex sets in the seislet transform domain[C/OL]//SEG Technical Program Expanded Abstracts 2015. New Orleans, Louisiana: Society of Exploration Geophysicists, 2015: 3814-3819. DOI:10.1190/segam2015-5744782.1.
[28] GAN S, WANG S, CHEN Y, 等. Compressive sensing for seismic data reconstruction via fast projection onto convex sets based on seislet transform[J/OL]. Journal of Applied Geophysics, 2016, 130: 194-208. DOI:10.1016/j.jappgeo.2016. 03.033.
[29] TRAD D O, ULRYCH T J, SACCHI M D. Accurate interpolation with high‐resolution time‐variant Radon transforms[J/OL]. GEOPHYSICS, 2002, 67(2): 644-656. DOI:10.1190/1.1468626.
[30] WANG J, NG M, PERZ M. Seismic data interpolation by greedy local Radon transform[J/OL]. GEOPHYSICS, 2010, 75(6): WB225-WB234. DOI:10.1190/1. 3484195.
[31] YU Z, FERGUSON J, MCMECHAN G, 等. Wavelet-Radon domain dealiasing and interpolation of seismic data[J/OL]. GEOPHYSICS, 2007, 72(2): V41-V49. DOI:10.1190/1.2422797.
[32] ABMA R, KABIR N. 3D interpolation of irregular data with a POCS algorithm[J/OL]. GEOPHYSICS, 2006, 71(6): E91-E97. DOI:10.1190/1.2356088.
[33] JAHANJOOY S, NIKROUZ R, MOHAMMED N. A faster method to reconstruct seismic data using anti-leakage Fourier transform[J/OL]. Journal of Geophysics and Engineering, 2016, 13(1): 86-95. DOI:10.1088/1742-2132/13/1/86.
[34] XU S, ZHANG Y, PHAM D, 等. Antileakage Fourier transform for seismic data regularization[J/OL]. GEOPHYSICS, 2005, 70(4): V87-V95. DOI:10.1190/1. 1993713.
[35] XU S, ZHANG Y, LAMBARÉ G. Antileakage Fourier transform for seismic data regularization in higher dimensions[J/OL]. GEOPHYSICS, 2010, 75(6): WB113-WB120. DOI:10.1190/1.3507248.
[36] CHAI X, GU H, LI F, 等. Deep learning for irregularly and regularly missing data reconstruction[J/OL]. Scientific Reports, 2020, 10(1): 3302. DOI:10.1038/s41598-020-59801-x.
[37] RONNEBERGER O, FISCHER P, BROX T. U-Net: Convolutional Networks for Biomedical Image Segmentation[J/OL]. International Conference on Medical Image Computing and Computer-Assisted Intervention, 2015. DOI:10.1007/978-3-319-24574-4_28.
[38] LIU Q, FU L, ZHANG M. Deep-seismic-prior-based reconstruction of seismic data using convolutional neural networks[J/OL]. GEOPHYSICS, 2021, 86(2): V131-V142. DOI:10.1190/geo2019-0570.1.
[39] WANG B, ZHANG N, LU W, 等. Deep-learning-based seismic data interpolation: A preliminary result[J/OL]. GEOPHYSICS, 2019, 84(1): V11-V20. DOI:10.1190/ geo2017-0495.1.
[40] GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, 等. Generative Adversarial Networks[J/OL]. arXiv:1406.2661 [cs, stat], 2014. http://arxiv.org/abs/1406.2661.
[41] KAUR H, PHAM N, FOMEL S. Seismic data interpolation using deep learning with generative adversarial networks[J/OL]. Geophysical Prospecting, 2021, 69(2): 307-326. DOI:10.1111/1365-2478.13055.
[42] XIE C, FANG Y, ZHANG J. Regularizing the 3-D teleseismic wavefield for receiver function imaging using a radial basis function[J/OL]. Geophysical Journal International, 2022, 229(2): 1255-1267. DOI:10.1093/gji/ggab536.
[43] HU S, JIANG X, ZHU L, 等. Wavefield Reconstruction of Teleseismic Receiver Function with the Stretching‐and‐Squeezing Interpolation Method[J/OL]. Seismological Research Letters, 2019, 90(2A): 716-726. DOI:10.1785/0220180197.
[44] JIANG X, ZHU L, HU S, 等. Three-dimensional reverse-time migration of teleseismic receiver functions using the phase-shift-plus-interpolation method[J/OL]. Geophysical Journal International, 2019, 217(2): 1047-1057. DOI:10.1093/gji/ggz066.
[45] YEEH Z, SONG Y, BYUN J, 等. Regularization of multidimensional sparse seismic data using Delaunay tessellation[J/OL]. Journal of Applied Geophysics, 2020, 174: 103877. DOI:10.1016/j.jappgeo.2019.103877.
[46] NEAL S L, PAVLIS G L. Imaging P-to-S conversions with multichannel receiver functions[J/OL]. Geophysical Research Letters, 1999, 26(16): 2581-2584. DOI:10.1029/1999GL900566.
[47] NEAL S L, PAVLIS G L. Imaging P-to-S conversions with broad-band seismic arrays using multichannel time-domain deconvolution[J/OL]. Geophysical Journal International, 2001, 147(1): 57-67. DOI:10.1046/j.1365-246X.2001.00507.x.
[48] POPPELIERS C, PAVLIS G L. Three-dimensional, prestack, plane wave migration of teleseismic P-to-S converted phases: 1. Theory: PLANE WAVE MIGRATION[J/OL]. Journal of Geophysical Research: Solid Earth, 2003, 108(B2). DOI:10.1029/2001JB000216.
[49] PAVLIS G L. Three-dimensional, wavefield imaging of broadband seismic array data[J/OL]. Computers & Geosciences, 2011, 37(8): 1054-1066. DOI:10.1016/j. cageo.2010.11.015.
[50] ZHANG J, ZHENG T. Receiver Function Imaging with Reconstructed Wavefields from Sparsely Scattered Stations[J/OL]. Seismological Research Letters, 2015, 86(1): 165-172. DOI:10.1785/0220140028.
[51] SHELDRAKE K P. Regional Wavefield reconstruction for teleseismic P-waves and Surface waves[J/OL]. Geophysical Research Letters, 2002, 29(11): 1544. DOI:10.1029/2001GL013721.
[52] WILSON C, GUITTON A. Teleseismic wavefield interpolation and signal extraction using high-resolution linear radon transforms[J/OL]. Geophysical Journal International, 2007, 168(1): 171-181. DOI:10.1111/j.1365-246X.2006.03163.x.
[53] SONG P, ZHANG X, LIU Y, 等. Moho imaging based on receiver function analysis with teleseismic wavefield reconstruction: Application to South China[J/OL]. Tectonophysics, 2017, 718: 118-131. DOI:10.1016/j.tecto.2017.05.031.
[54] CHAI C, AMMON C J, MACEIRA M, 等. Inverting interpolated receiver functions with surface wave dispersion and gravity: Application to the western U.S. and adjacent Canada and Mexico[J/OL]. Geophysical Research Letters, 2015, 42(11): 4359-4366. DOI:10.1002/2015GL063733.
[55] OUSSIDI A, ELHASSOUNY A. Deep generative models: Survey[C/OL]//2018 International Conference on Intelligent Systems and Computer Vision (ISCV). Fez: IEEE, 2018: 1-8.DOI:10.1109/ISACV.2018.8354080.
[56] SALAKHUTDINOV R, HINTON G E. Deep Boltzmann Machines[J/OL]. Journal of Machine Learning Research, 2009, 5(2): 1967-2006. DOI:10.1162/NECO_a_00311.
[57] REZENDE D J, MOHAMED S, WIERSTRA D. Stochastic Backpropagation and Approximate Inference in Deep Generative Models[J/OL]. arXiv:1401.4082 [cs, stat], 2014. http://arxiv.org/abs/1401.4082.
[58] OORD A van den, KALCHBRENNER N, KAVUKCUOGLU K. Pixel Recurrent Neural Networks: arXiv:1601.06759[R/OL]. arXiv, 2016.DOI:10.48550/arXiv. 1601.06759.
[59] URIA B, CÔTÉ M A, GREGOR K, 等. Neural Autoregressive Distribution Estimation[J/OL]. arXiv:1605.02226 [cs], 2016. http://arxiv.org/abs/ 1605.02226.
[60] RADFORD A, NARASIMHAN K, SALIMANS T, 等. Improving Language Understanding by Generative Pre-Training[J].
[61] BROWN T B, MANN B, RYDER N, 等. Language Models are Few-Shot Learners[M/OL]. arXiv, 2020. http://arxiv.org/abs/2005.14165.
[62] RADFORD A, WU J, CHILD R, 等. Language Models are Unsupervised Multitask Learners[J].
[63] OUYANG L, WU J, JIANG X, 等. Training language models to follow instructions with human feedback[M/OL]. arXiv, 2022. http://arxiv.org/abs/ 2203.02155.
[64] LIU M, LI W, JERVIS M, 等. 3D seismic facies classification using convolutional neural network and semi-supervised generative adversarial network[C]//SEG Technical Program Expanded Abstracts 2019. 2019.
[65] FENG R, BALLING N, GRANA D, 等. Bayesian Convolutional Neural Networks for Seismic Facies Classification[J/OL]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(10): 8933-8940. DOI:10.1109/TGRS.2020.3049012.
[66] LIU M, JERVIS M, LI W, 等. Seismic facies classification using supervised convolutional neural network and semi-supervised generative adversarial networks[J]. Geophysics, 2020, 85(4): 1-57.
[67] CHEN G, LIU Y, ZHANG M, 等. Dropout-Based Robust Self-Supervised Deep Learning for Seismic Data Denoising[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19.
[68] RICHARDSON A, FELLER C. Seismic data denoising and deblending using deep learning[J]. 2019.
[69] HAN W, ZHOU Y, YUE C. Deep learning convolutional neural networks for random noise attenuation in seismic data[J]. Geophysical Prospecting for Petroleum, 2018.
[70] GAO, HAN, ZHANG, 等. Simultaneous Denoising and Interpolation of Seismic Data via the Deep Learning Method[J]. Earthquake Research in China, 2019.
[71] ZHU W, TAI K S, MOUSAVI S M, 等. An End-to-End Earthquake Detection Method for Joint Phase Picking and Association using Deep Learning[J]. 2021.
[72] GUO C, ZHU T, GAO Y, 等. AEnet: Automatic Picking of P-Wave First Arrivals Using Deep Learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020.
[73] HE Z, PENG P, WANG L, 等. PickCapsNet: Capsule Network for Automatic P-Wave Arrival Picking[J]. IEEE Geoscience and Remote Sensing Letters, 2020, PP(99): 1-5.
[74] HUGO S, BERND S. DeepPhasePick: A method for detecting and picking seismic phases from local earthquakes based on highly optimized convolutional and recurrent deep neural networks[J]. Geophysical Journal International, 2021.
[75] JIN P, ZHANG X, CHEN Y, 等. Unsupervised Learning of Full-Waveform Inversion: Connecting CNN and Partial Differential Equation in a Loop[J]. arXiv e-prints, 2021.
[76] WU Y, LIN Y, ZHOU Z. Inversionet: Accurate and efficient seismic-waveform inversion with convolutional neural networks[C]//SEG Technical Program Expanded Abstracts 2018. 2018.
[77] ZHANG W, GAO J. Deep-Learning Full-Waveform Inversion Using Seismic Migration Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, PP(99): 1-18.
[78] YANG L, SUN S Z. Seismic horizon tracking using a deep convolutional neural network[J/OL]. Journal of Petroleum Science and Engineering, 2020, 187: 106709. DOI:10.1016/j.petrol.2019.106709.
[79] WU H, ZHANG B, LIN T, 等. Semi-automated seismic horizon interpretation using encoder-decoder convolutional neural network[J]. Geophysics, 2019, 84(6): 1-56.
[80] BAI T, PEJMAN T. Attention-based LSTM-FCN for earthquake detection and location[J]. Geophysical Journal International, 2021(3): 3.
[81] NGUYEN V Q, YANG H J, KIM K, 等. Real-Time Earthquake Detection Using Convolutional Neural Network and Social Data[C]//2017 IEEE Third International Conference on Multimedia Big Data (BigMM). 2017.
[82] MCCULLOCH W S, PITTS W H. A logical Calculus of Ideas Immanent in Nervous Activity[J]. The Bulletin of Mathematical Biophysics, 1942, 5: 115-133.
[83] RUMELHART D E, HINTON G E, WILLIAMS R J. Learning Representations by Back Propagating Errors[J]. Nature, 1986, 323(6088): 533-536.
[84] HINTON G E, OSINDERO S, TEH Y W. A Fast Learning Algorithm for Deep Belief Nets[J/OL]. Neural Computation, 2006, 18(7): 1527-1554. DOI:10.1162/neco.2006. 18.7.1527.
[85] NAIR V, HINTON G E. Rectified Linear Units Improve Restricted Boltzmann Machines Vinod Nair[C]//International Conference on International Conference on Machine Learning. 2010.
[86] MAAS A L, HANNUN A Y, NG A Y. Rectifier nonlinearities improve neural network acoustic models[J/OL]. in ICML Workshop on Deep Learning for Audio, Speech and Language Processing, 2013.
[87] LECUN Y, BOSER B, DENKER J, 等. Backpropagation Applied to Handwritten Zip Code Recognition[J]. Neural Computation, 1989, 1(4): 541-551.
[88] SZEGEDY C, LIU W, JIA Y, 等. Going Deeper with Convolutions[J]. IEEE Computer Society, 2014.
[89] SIMONYAN K, ZISSERMAN A. Very Deep Convolutional Networks for Large-Scale Image Recognition[M/OL]. arXiv, 2015. http://arxiv.org/abs/1409.1556. DOI:10.48550/arXiv.1409.1556.
[90] KRIZHEVSKY A, SUTSKEVER I, HINTON G. ImageNet Classification with Deep Convolutional Neural Networks[J]. Advances in neural information processing systems, 2012, 25(2).
[91] GIRSHICK R, DONAHUE J, DARRELL T, 等. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[J]. IEEE Computer Society, 2013.
[92] GIRSHICK R. Fast R-CNN[J]. Computer Science, 2015.
[93] HE K, GKIOXARI G, DOLLAR P, 等. Mask R-CNN[C]//International Conference on Computer Vision. 2017.
[94] ZHAO Z, WU Y. Attention-Based Convolutional Neural Networks for Sentence Classification[C]//Interspeech 2016. 2016.
[95] MILLETARI F, NAVAB N, AHMADI S A. V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation[M/OL]. arXiv, 2016. http://arxiv.org/abs/1606.04797. DOI:10.48550/arXiv.1606.04797.
[96] TRAN D, BOURDEV L, FERGUS R, 等. Learning Spatiotemporal Features with 3D Convolutional Networks[J]. IEEE, 2015.
[97] SIMONYAN K, ZISSERMAN A. Two-Stream Convolutional Networks for Action Recognition in Videos[J]. Advances in neural information processing systems, 2014, 1.
[98] NOH H, HONG S, HAN B. Learning Deconvolution Network for Semantic Segmentation[J]. IEEE, 2016.
[99] BENGIO Y, SIMARD P, FRASCONI P. Learning long-term dependencies with gradient descent is difficult[J/OL]. IEEE Transactions on Neural Networks, 1994, 5(2): 157-166. DOI:10.1109/72.279181.
[100] GLOROT X, BENGIO Y. Understanding the difficulty of training deep feedforward neural networks[J]. Journal of Machine Learning Research - Proceedings Track, 2010, 9: 249-256.
[101] HE K, ZHANG X, REN S, 等. Deep Residual Learning for Image Recognition[J/OL]. arXiv:1512.03385 [cs], 2015. http://arxiv.org/abs/ 1512.03385.
[102] HE K, ZHANG X, REN S, 等. Identity Mappings in Deep Residual Networks[J/OL]. European Conference on Computer Vision, 2016. DOI:10.1007/978-3-319-46493-0_38.
[103] 张敏, 李贤均, 王瑞琦, 等. 一种基于时空融合特征的间歇过程弱故障识别方法[J]. 工业工程与管理, 2022, 27(3): 10.
[104] IOFFE S, SZEGEDY C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift[J]. JMLR.org, 2015.
[105] GRAVES A, FERNÁNDEZ S, GOMEZ F. Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks[C]//ACM. 2006.
[106] LECUN Y, BOTTOU L, ORR G B, 等. Efficient BackProp[J]. neural networks tricks of the trade, 1998.
[107] TIBSHIRANI R. Regression shrinkage and selection via the lasso: a retrospective[J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 2011, 73(3): 267-288.
[108] TIKHONOV A N. On the stability of inverse problems[J]. dolk.akad.nauk sssr, 1943.
[109] HINTON G E, CAMP D V. Keeping the neural networks simple by minimizing the description length of the weights[J]. ACM, 1999.
[110] HINTON G E, SRIVASTAVA N, KRIZHEVSKY A, 等. Improving neural networks by preventing co-adaptation of feature detectors[M]. 2012: págs. 212-223.
[111] KINGMA D P, BA J. Adam: A Method for Stochastic Optimization: arXiv:1412.6980[R/OL]. arXiv, 2017. DOI:10.48550/arXiv.1412.6980.
[112] HE K, ZHANG X, REN S, 等. Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification[J/OL]. arXiv:1502.01852 [cs], 2015. http://arxiv.org/abs/1502.01852.
[113] XIN H, ZHANG H, KANG M, 等. High‐Resolution Lithospheric Velocity Structure of Continental China by Double‐Difference Seismic Travel‐Time Tomography[J/OL]. Seismological Research Letters, 2019, 90(1): 229-241. DOI:10.1785/0220180209.
[114] SUN Y, ZHANG W, CHEN X. Seismic‐Wave Modeling in the Presence of Surface Topography in 2D General Anisotropic Media by a Curvilinear Grid Finite‐Difference Method[J/OL]. Bulletin of the Seismological Society of America, 2016, 106(3): 1036-1054. DOI:10.1785/0120150285.
[115] SUN Y, ZHANG W, CHEN X. 3D Seismic Wavefield Modeling in Generally Anisotropic Media with a Topographic Free Surface by the Curvilinear Grid Finite‐Difference Method[J/OL]. Bulletin of the Seismological Society of America, 2018, 108(3A): 1287-1301. DOI:10.1785/0120170154.
[116] ZHANG W, CHEN X. Traction image method for irregular free surface boundaries in finite difference seismic wave simulation[J/OL]. Geophysical Journal International, 2006, 167(1): 337-353. DOI:10.1111/j.1365-246X.2006.03113.x.
[117] ZHU H, ZHANG W, CHEN X. Two-dimensional seismic wave simulation in anisotropic media by non-staggered finite difference method[J/OL]. Chinese Journal of Geophysics, 2009, 52(6): 1536-1546. DOI:10.3969/j.issn.0001-5733.2009. 06.015.
[118] KRISTEKOVÁ M, KRISTEK J, MOCZO P. Time-frequency misfit and goodness-of-fit criteria for quantitative comparison of time signals[J/OL]. Geophysical Journal International, 2009, 178(2): 813-825. DOI:10.1111/j.1365-246X.2009. 04177.x.
[119] KRISTEKOVÁ M, KRISTEK J, MOCZO P, 等. Misfit Criteria for Quantitative Comparison of Seismograms[J/OL]. Bulletin of the Seismological Society of America, 2006, 96(5): 1836-1850. DOI:10.1785/0120060012.
[120] CHEN Y, CHEN X, WANG Y, 等. The Interpolation of Sparse Geophysical Data[J/OL]. Surveys in Geophysics, 2019, 40(1): 73-105. DOI:10.1007/s10712-018-9501-3.
[121] LI J, SHEN Y, ZHANG W. Three‐Dimensional Passive‐Source Reverse‐Time Migration of Converted Waves: The Method[J/OL]. Journal of Geophysical Research: Solid Earth, 2018, 123(2): 1419-1434. DOI:10.1002/2017JB014817.
[122] DZIEWONSKI A M, ANDERSON D L. Preliminary reference Earth model[J/OL]. Physics of the Earth and Planetary Interiors, 1981, 25(4): 297-356. DOI:10.1016/0031-9201(81)90046-7.
[123] 张桂诚. 第一视角行为识别算法研究[D/OL]. 西安理工大学, 2021. DOI:10.27398/d.cnki.gxalu.2021.001464.
[124] 陈九辉. 远震体波接收函数方法：理论与应用[D/OL]. 中国地震局地质研究所, 2007.