基于深度学习的表面多次波智能压制研究进展

Research advances in intelligent suppression of surface-related multiples based on deep learning

摘要

表面多次波的存在会严重影响地震资料的成像质量与解释精度, 其有效压制一直是地震数据处理中的关键环节。传统方法主要包括基于信号处理的滤波法和基于波动方程的预测相减法, 但对于复杂构造或低信噪比条件的压制效果有限。近年来, 深度学习技术凭借其强大的特征提取与模式识别能力, 为表面多次波压制提供了新的解决思路与方法。本研究调研并分析了多种表面多次波智能压制方法, 涵盖监督学习、无监督学习、自监督学习以及融合物理先验等多种范式, 并系统介绍了其发展历程、基础原理、适用性优缺点以及表面多次波压制效果。最后结合当下基于深度学习的表面多次波智能压制的研究趋势, 展望未来表面多次波智能压制技术的发展方向。

Abstract

The presence of surface-related multiples severely degrades the imaging quality and interpretational accuracy of seismic data, making their effective suppression a critical task in seismic data processing. Traditional methods, primarily including filtering techniques based on signal processing and predictive subtraction methods based on wave equation, often exhibit limited performance when dealing with complex geological structures or low signal-to-noise ratio conditions. In recent years, deep learning technology, with its powerful feature extraction and pattern recognition capabilities, has provided novel solutions and methodologies for suppressing surface-related multiples. This paper investigates and analyzes a variety of intelligent suppression methods for surface-related multiples, covering multiple paradigms such as supervised learning, unsupervised learning, self-supervised learning, and physics-informed integration. It also systematically reviews their development, fundamental principles, applicability, advantages, limitations, and suppression effects. Based on the current research trends in deep learning-based multiple suppression, it finally discusses the future development directions for intelligent suppression techniques of surface-related multiples.

关键词

表面多次波 / 深度学习 / 监督学习 / 无监督学习 / 自监督学习 / 物理融合

Key words

surface-related multiples / deep learning / supervised learning / unsupervised learning / self-supervised learning / physics-informed methods

参考文献

[1] 胡嘉晨, 童思友, 尚新民, 等 . 基于注意力机制的 MWCNN 网络的海洋自由表面多次波压制研究[J]. 中国海洋大学学报(自然科学版), 2025, 55(2): 89-102.

[2] HU Jiachen , TONG Siyou , SHANG Xinmin , et al. Research on ocean free surface multiple suppression based on attention mechanism in MWCNN[J]. Periodical of Ocean University of China, 2025, 55(2): 89-102.

[3] WEGLEIN A B . Multiple attenuation: An overview of recent advances and the road ahead[J]. The Leading Edge, 1999, 18(1): 40-44.

[4] 曹艳玲, 赵波, 吴波, 等 . 基于多次反射等效界面模型的海上三维多次波预测与剔除[J]. 中国海洋大学学报(自然科学版), 2021, 51(11): 66-73.

[5] CAO Yanling , ZHAO Bo , WU Bo , et al. Prediction and elimination of marine 3D multiple based on multiple reflection equivalent interface model[J]. Periodical of Ocean University of China, 2021, 51(11): 66-73.

[6] 黄柱富, 刘剑锋, 方文倩, 等 . 动校正域 CNN 法压制自由表面多次波[J]. 煤田地质与勘探, 2024, 52(11): 160-170.

[7] HUANG Zhufu , LIU Jianfeng , FANG Wenqian , et al. Surfaced—related multiple attenuation using the CNN method in the NMO domain[J]. Coal Geology and Exploration, 2024, 52(11): 160-170.

[8] 徐嘉亮, 张冰, 王维红, 等 . 基于 AVO 正演模拟的自由表面多次波压制方法[J]. 地球物理学报, 2021, 64(7): 2471-2479.

[9] XU Jialiang , ZHANG Bing , WANG Weihong , et al. Free—surface multiple wave suppression method based on AVO forward simulation[J]. Chinese Journal of Geophysics, 2021, 64(7): 2471-2479.

[10] 马继涛, 赵凯歌, 廖震 . 应用主频带约束高精度 Radon 变换的多次波压制方法[J]. 石油地球物理勘探, 2024, 59(6): 1286-1296.

[11] MA Jitao , ZHAO Kaige , LIAO Zhen . Multiple attenuation by high precision Radon transform with dominant frequency band constraint[J]. Oil Geophysical Prospecting, 2024, 59(6): 1286-1296.

[12] 徐嘉亮, 陈海峰, 于波, 等 . 基于虚同相轴和高精度拉东变换联合压制层间多次波方法[J]. 地球物理学进展, 2025, 40(4): 1417-1427.

[13] XU Jialiang , CHEN Haifeng , YU Bo , et al. Internal multiples suppression method based on virtual event and high—precision Radon transform[J]. Progress in Geophysics, 2025, 40(4): 1417-1427.

[14] 薛亚茹, 张程, 冯璐瑜, 等 . 基于 L1/L2 正则化的高分辨率 Radon 变换反演方法[J]. 地球物理学报, 2025, 68(6): 2348-2363.

[15] XUE Yaru , ZHANG Cheng , FENG Luyu , et al. High—resolution Radon transform inversion based on L1/L2 regularization[J]. Chinese Journal of Geophysics, 2025, 68(6): 2348-2363.

[16] 陈祖传 . 衰减多次波的剩余时差分析法[J]. 中国海上油气(地质), 1999, 13(1): 50-56.

[17] CHEN Zuchuan . Residual NMO analysis method for multiple attenuation[J]. China Offshore Oil and Gas(Geology), 1999, 13(1): 50-56.

[18] 胡天跃, 王润秋, 温书亮 . 聚束滤波法消除海上地震资料的多次波[J]. 石油地球物理勘探, 2002, 37(1): 18-23.

[19] HU Tianyue , WANG Runqiu , WEN Shuliang . Multiple attenuation of seismic data from South China Sea by using beam—forming filtering method[J]. Oil Geophysical Prospecting, 2002, 37(1): 18-23.

[20] 洪菲, 胡天跃, 张文坡, 等 . 用优化聚束滤波方法消除低信噪比地震资料中的多次波[J]. 地球物理学报, 2004, 47(6): 1106-1110.

[21] HONG Fei , HU Tianyue , ZHANG Wenpo , et al. Attenuating multiples for low signal—to—noise ratio seismic data using optimal beamforming[J]. Chinese Journal of Geophysics, 2004, 47(6): 1106-1110.

[22] ROBINSON E A , TREITEL S . Principles of digital Wiener filtering[J]. Geophysical Prospecting, 1967, 15(3): 311-332.

[23] PEACOCK K L , TREITEL S . Predictive deconvolution: Theory and practice[J]. Geophysics, 1969, 34(2): 155-169.

[24] 贾连凯, 吴时国, 赵昌垒 . 多次波压制技术在南海北部陆缘深水区的应用[J]. 地球物理学进展, 2014, 29(2): 920-930.

[25] JIA Liankai , WU Shiguo , ZHAO Changlei . Application of multiple elimination techniques in the deepwater continental margin of northern South China Sea[J]. Progress in Geophysics, 2014, 29(2): 920-930.

[26] MORLEY L , CLAERBOUT J . Predictive deconvolution in short—receiver space[J]. Geophysics, 1983, 48(5): 515-531.

[27] WIGGINS J W . Attenuation of complex water—bottom multiples by wave—equation—based prediction and subtraction[J]. Geophysics, 1988, 53(12): 1527-1539.

[28] 张广利, 郝重涛, 姚陈 . 海洋地震资料多次波衰减方法综述[J]. 地球物理学进展, 2016, 31(6): 2777-2787.

[29] ZHANG Guangli , HAO Chongtao , YAO Chen . Summary of multiples attenuation approaches in marine seismic data processing[J]. Progress in Geophysics, 2016, 31(6): 2777-2787.

[30] WEGLEIN A B , GASPAROTTO F A , CARVALHO P M , et al. An inverse scattering series method for attenuating multiples in seismic reflection data[J]. Geophysics, 1997, 62(6): 1975-1989.

[31] 李翔, 胡天跃 . 逆散射级数法去除自由表面多次波[J]. 地球物理学报, 2009, 52(6): 1633-1640.

[32] LI Xiang , HU Tianyue . Surface—related multiple removal with inverse scattering series method[J]. Chinese Journal of Geophysics, 2009, 52(6): 1633-1640.

[33] 赵锐锐, 陈新哲, 向平奥, 等 . 地震资料层间多次波压制方法研究进展[J]. 石油物探, 2024, 63(6): 1163-1176.

[34] ZHAO Ruirui , CHEN Xinzhe , XIANG Ping’ao , et al. Research progress of interbed multiples suppression[J]. Geophysical Prospecting for Petroleum, 2024, 63(6): 1163-1176.

[35] KENNETT B L N . The suppression of surface multiples on seismic records[J]. Geophysical Prospecting, 1979, 27(3): 584-600.

[36] BERKHOUT A J , VERSCHUUR D J . Estimation of multiple scattering by iterative inversion, Part I: Theoretical considerations[J]. Geophysics, 1997, 62(5): 1586-1595.

[37] VERSCHUUR D J , BERKHOUT A J . Estimation of multiple scattering by iterative inversion, Part II: Practical aspects and examples[J]. Geophysics, 1997, 62(5): 1596-1611.

[38] WANG Y H . Multiple subtraction using an expanded multichannel matching filter[J]. Geophysics, 2003, 68(1): 346-354.

[39] GUITTON A , VERSCHUUR D J . Adaptive subtraction of multiples using the L1—norm[J]. Geophysical Prospecting, 2004, 52(1): 27-38.

[40] 陈习峰, 薛永安, 黄新武 . 自适应加权混合 L1/L2 范数匹配相减多次波压制方法[J]. 石油物探, 2019, 58(4): 524-532.

[41] CHEN Xifeng , XUE Yongan , HUANG Xinwu . Adaptive subtraction for multiples suppression using hybrid L1/L2 norm[J]. Geophysical Prospecting for Petroleum, 2019, 58(4): 524-532.

[42] WANG Z S , LI Y X , SU W Q , et al. Comparisons of five least—squares adaptive matched filtering approaches for multiple attenuation[J]. Journal of Geophysical Engineering, 2022, 19(5): 1046-1063.

[43] 李志娜, 张敏, 李振春, 等 . 基于伪多道匹配的时移地震互均化方法[J]. 石油地球物理勘探, 2015, 50(6): 1083-1088.

[44] LI Zhina , ZHANG Min , LI Zhenchun , et al. Cross—equalization based on pseudo—multichannel matching in time—lapse seismic[J]. Oil Geophysical Prospecting, 2015, 50(6): 1083-1088.

[45] 刘瀚晖, 曾庆田, 宋戈, 等 . 矿井作业视频图像的轻量级自适应面部疲劳检测算法[J]. 山东科技大学学报(自然科学版), 2024, 43(5): 97-109.

[46] LIU Hanhui , ZENG Qingtian , SONG Ge , et al. Lightweight adaptive facial fatigue detection algorithm for mine operation video images[J]. Journal of Shandong University of Science and Technology(Natural Science), 2024, 43(5): 97-109.

[47] 张浚丰, 许明, 余小越, 等 . 挡墙异常体雷达图像的深度学习智能判译方法[J]. 地球物理学报, 2025, 68(5): 1970-1983.

[48] ZHANG Junfeng , XU Ming , YU Xiaoyue , et al. Intelligent interpretation method of retaining wall anomaly radar images based on deep learning[J]. Chinese Journal of Geophysics, 2025, 68(5): 1970-1983.

[49] 李盼盼, 李洋冰, 税蕾蕾, 等 . 基于深度学习的致密砂岩显微图像识别[J]. 岩矿测试, 2026, 45(1): 220-230.

[50] LI Panpan , LI Yangbing , SHUI Leilei , et al. Microscopic recognition of tight sandstone based on deep learning[J]. Rock and Mineral Analysis, 2026, 45(1): 220-230.

[51] 张旭, 王旭强, 田雨婷, 等 . 基于主题感知的跨模态序列到序列生成模型[J]. 山东科技大学学报(自然科学版), 2021, 40(3): 71-79.

[52] ZHANG Xu , WANG Xuqiang , TIAN Yuting , et al. Topic—aware based cross—modal sequence—to—sequence generation model[J]. Journal of Shandong University of Science and Technology(Natural Science), 2021, 40(3): 71-79.

[53] 何雪锋, 周洁, 陈德光, 等 . 自然语言处理的深度学习模型综述[J]. 计算机应用与软件, 2025, 42(2): 1-19.

[54] HE Xuefeng , ZHOU Jie , CHEN Deguang , et al. Overview of deep learning models in natural language processing[J]. Computer Applications and Software, 2025, 42(2): 1-19.

[55] 钱增志, 孙玉龙, 张捷, 等 . 基于 YOLO 和自然语言处理的施工方案智能审核方法[J]. 图学学报, 2025, 46(5): 1113-1122.

[56] QIAN Zengzhi , SUN Yulong , ZHANG Jie , et al. A construction plan intelligent review method based on YOLO and natural language processing[J]. Journal of Graphics, 2025, 46(5): 1113-1122.

[57] 王瑞, 花嵘, 仪秀龙, 等 . 自动生成影像学报告的混合特征提取无卷积结构深度学习模型[J]. 山东科技大学学报(自然科学版), 2023, 42(3): 85-93.

[58] WANG Rui , HUA Rong , YI Xiulong , et al. A hybrid feature extraction deep learning model without convolutional architecture for automatic imaging report generation[J]. Journal of Shandong University of Science and Technology(Natural Science), 2023, 42(3): 85-93.

[59] 李智唯, 刘静, 张俊忠, 等 . 深度学习在肋骨骨折辅助诊断中的应用[J]. 计算机工程与应用, 2025, 61(19): 74-91.

[60] LI Zhiwei , LIU Jing , ZHANG Junzhong , et al. Application of deep learning in auxiliary diagnosis of rib fractures[J]. Computer Engineering and Applications, 2025, 61(19): 74-91.

[61] 陶春昊, 王路欣, 乔爱科 . 基于深度学习的血流动力学计算研究进展[J]. 医用生物力学, 2025, 40(5): 1354-1359.

[62] TAO Chunhao , WANG Luxin , QIAO Aike . Advances in hemodynamic computation based on deep learning[J]. Journal of Medical Biomechanics, 2025, 40(5): 1354-1359.

[63] 崔家豪, 杨平, 王洪强, 等 . 基于深度学习的地震速度谱自动拾取研究[J]. 地球物理学报, 2022, 65(12): 4832-4845.

[64] CUI Jiahao , YANG Ping , WANG Hongqiang , et al. Research on automatic picking of seismic velocity spectrum based on deep learning[J]. Chinese Journal of Geophysics, 2022, 65(12): 4832-4845.

[65] 潘海侠, 耿伟峰, 崔家豪, 等 . 面向地震速度谱智能拾取的深度学习方法[J]. 地球物理学进展, 2023, 38(6): 2553-2564.

[66] PAN Haixia , GENG Weifeng , CUI Jiahao , et al. Deep learning method for intelligent picking of seismic velocity spectrum[J]. Progress in Geophysics, 2023, 38(6): 2553-2564.

[67] 许冲 . 基于深度学习的智能速度谱拾取技术及应用[J]. 油气藏评价与开发, 2025, 15(5): 788-795.

[68] XU Chong . Deep learning—based intelligent velocity spectrum picking technology and its application[J]. Petroleum Reservoir Evaluation and Development, 2025, 15(5): 788-795.

[69] 武庭润, 高建虎, 常德宽, 等 . 基于 Transformer 的地震数据断层识别[J]. 石油地球物理勘探, 2024, 59(6): 1217-1224.

[70] WU Tingrun , GAO Jianhu , CHANG Dekuan , et al. Faults identification of seismic data based on Transformer[J]. Oil Geophysical Prospecting, 2024, 59(6): 1217-1224.

[71] 唐金良 . 基于迁移学习的断层识别方法与应用[J]. 石油物探, 2025, 64(3): 543-552.

[72] TANG Jinliang . Fault detection method and application based on transfer learning[J]. Geophysical Prospecting for Petroleum, 2025, 64(3): 543-552.

[73] 安虹伊, 文馨, 李居正, 等 . 基于 U—CNNformer 网络的地震断层智能识别方法[J]. 海相油气地质, 2025, 30(3): 277-288.

[74] AN Hongyi , WEN Xin , LI Juzheng , et al. Intelligent seismic fault identification method of based on U—CNNformer network[J]. Marine Origin Petroleum Geology, 2025, 30(3): 277-288.

[75] 吴迪, 文武, 门哲, 等 . 基于扩散模型的地震数据随机噪声压制方法[J]. 石油地球物理勘探, 2024, 59(6): 1252-1259.

[76] WU Di , WEN Wu , MEN Zhe , et al. A random noise suppression method for seismic data based on diffusion model[J]. Oil Geophysical Prospecting, 2024, 59(6): 1252-1259.

[77] 韦秀娟, 刘兴业, 周怀来 . 卷积 Mamba 模型驱动的地震随机噪声压制方法[J]. 煤田地质与勘探, 2025, 53(5): 196-206.

[78] WEI Xiujuan , LIU Xingye , ZHOU Huailai . A seismic random noise suppression method based on CNN—Mamba[J]. Coal Geology and Exploration, 2025, 53(5): 196-206.

[79] 罗仁泽, 李治岐 . 基于 MSIP GAN 的地震数据随机噪声压制方法[J]. 石油物探, 2025, 64(4): 653-666.

[80] LUO Renze , LI Zhiqi . Random noise suppression method of seismic data based on MSIP GAN[J]. Geophysical Prospecting for Petroleum, 2025, 64(4): 653-666.

[81] WANG K , HU T , WANG S , et al. Seismic multiple suppression based on a deep neural network method for marine data[J]. Geophysics, 2022, 87(4): 341-365.

[82] LECUN Y , BOTTOU L , BENGIO Y , et al. Gradient—based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

[83] RONNEBERGER O , FISCHER P , BROX T . U—Net: Convolutional networks for biomedical image segmentation[C]// Medical Image Computing and Computer—Assisted Intervention—MICCAI 2015, LNCS 9351.Cham: Springer, 2015: 234-241.

[84] GOODFELLOW I , POUGET—ABADIE J , MIRZA M , et al. Generative adversarial nets[C]// Advances in neural information processing systems.Montréal: Curran Associates Inc., 2014: 2672-2680.

[85] LIU L C , HU T Y , HUANG J D , et al. Adaptive surface—related multiple subtraction based on convolutional neural network[J/OL]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 8021905.DOI: 10.1109/LGRS.2021.3122227.

[86] KIRAZ M S R , SNIEDER R , SHEIMAN J . Attenuating free—surface multiples and ghost reflection from seismic data using a trace—by—trace convolutional neural network approach[J]. Geophysical Prospecting, 2024, 72(3): 908-937.

[87] KIRAZ M S R , SNIEDER R , SHEIMAN J . Free—surface multiple attenuation and seismic deghosting for blended data using convolutional neural networks[J]. Geophysics, 2024, 89(3): 281-290.

[88] FALK T , MAI D , BENSCH R , et al. U—Net: Deep learning for cell counting, detection, and morphometry[J]. Nature Methods, 2019, 16(1): 67-70.

[89] QU S , VERSCHUUR E , ZHANG D , et al. Training deep networks with only synthetic data: Deep—learning—based near—offset reconstruction for(closed—loop)surface—related multiple estimation on shallow—water field data[J]. Geophysics, 2021, 86(3): 39-43.

[90] ZHANG D, DE LEEUW M, VERSCHUUR E. Deep learning—based seismic surface—related multiple adaptive subtraction with synthetic primary labels[C]// First International Meeting for Applied Geoscience and Energy.Houston: Society of Exploration Geophysicists, 2021: 2844-2848.

[91] BAO P N , SHI Y , WANG W H , et al. Surface—related and internal multiple elimination using deep learning[J/OL]. Energies, 2022, 15(11).DOI: 10.3390/en15113883.

[92] 齐娇, 曹思远 . 基于先验知识的深度学习表面相关多次波压制方法[J]. 吉林大学学报(地球科学版), 2025, 55(5): 1702-1714.

[93] QI Jiao , CAO Siyuan . Surface—related multiple attenuation based on deep learning with prior knowledge[J]. Journal of Jilin University(Earth Science Edition), 2025, 55(5): 1702-1714.

[94] 王正龙, 张保稳 . 生成对抗网络研究综述[J]. 网络与信息安全学报, 2021, 7(4): 68-85.

[95] WANG Zhenglong , ZHANG Baowen . Survey of generative adversarial network[J]. Chinese Journal of Network and Information Security, 2021, 7(4): 68-85.

[96] SIAHKOOHI A , VERSCHUUR D J , HERRMANN F J . Surface—related multiple elimination with deep learning[C]// SEG Technical Program Expanded Abstracts 2019.San Antonio: Society of Exploration Geophysicists, 2019: 4629-4634.

[97] TAO L R, REN H R, YE Y M, et al. Seismic surface—related multiples suppression based on SAGAN[J/OL]. IEEE Geoscience and Remote Sensing Letters, 2022, 19.DOI: 10.1109/LGRS.2022.3168143.

[98] 崔宁城, 张伟 . 利用深度学习图像转译方法压制自由表面多次波[J]. 石油地球物理勘探, 2025, 60(2): 283-291.

[99] CUI Ningcheng , ZHANG Wei . Free—surface—related multiple attenuation through image translation based on deep learning[J]. Oil Geophysical Prospecting, 2025, 60(2): 283-291.

[100] LECUN Y , BENGIO Y , HINTON G . Deep learning[J]. Nature, 2015, 521(7553): 436-444.

[101] WANG K X , HU T Y , ZHAO B L . An unsupervised deep neural network approach based on ensemble learning to suppress seismic surface—related multiples[J/OL]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60.DOI: 10.1109/TGRS.2022.3225809.

[102] 刘立超, 胡天跃, 李徯徯, 等 . 基于无监督神经网络匹配算法的叠前表面多次波压制方法[J]. 北京大学学报(自然科学版), 2024, 60(3): 453-463.

[103] LIU Lichao , HU Tianyue , LI Xixi , et al. Prestack surface multiple suppression method based on matching algorithm with unsupervised neural network[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2024, 60(3): 453-463.

[104] WANG K X , HU T Y , ZHAO B L , et al. Surface—related multiple attenuation based on a self—supervised deep neural network with local wavefield characteristics[J]. Geophysics, 2023, 88(5): 387-402.

基金资助

国家自然科学基金项目(42406077)

山东省自然科学基金项目(ZR2022QD087,ZR2023MD093,ZR2025MS527)