岩性油气藏 ›› 2025, Vol. 37 ›› Issue (6): 99–106.doi: 10.12108/yxyqc.20250609

• 地质勘探 • 上一篇    下一篇

基于加权K均值聚类的全自动速度拾取方法

刘正文1,2,3,4, 赵锐锐1,2,3,4, 陈阳阳1,2,3,4, 谢俊法5, 臧胜涛5, 秦龙1,2,3,4   

  1. 1. 中国石油天然气集团有限公司 超深层复杂油气藏勘探开发技术研发中心, 新疆 库尔勒 841000;
    2. 新疆维吾尔自治区超深层复杂油气藏勘探开发工程研究中心, 新疆 库尔勒 841000;
    3. 新疆超深油气重点实验室, 新疆 库尔勒 841000;
    4. 中国石油塔里木油田公司 勘探开发研究院, 新疆 库尔勒 841000;
    5. 中国石油勘探开发研究院 西北分院, 兰州 730020
  • 收稿日期:2025-05-29 修回日期:2025-06-30 出版日期:2025-11-01 发布日期:2025-11-07
  • 第一作者:刘正文（1983—），男，高级工程师，主要从事地震资料处理领域的科研及生产工作。地址：（841000）新疆库尔勒市塔指东路塔里木油田研发中心。Email：liuzhengwen@petrochina.com.cn。
  • 通信作者: 谢俊法(1987—),男,博士,高级工程师,主要从事地震数据规则化、层间多次波识别与压制等地震资料处理的方法研究。Email:xiejunfa@petrochina.com.cn。
  • 基金资助:
    中国石油天然气股份有限公司科技专项“超深层勘探开发地球物理关键技术研究与应用”（编号：2023ZZ14YJ05）资助。

A fully automatic velocity picking method based on weighted K-means clustering

LIU Zhengwen1,2,3,4, ZHAO Ruirui1,2,3,4, CHEN Yangyang1,2,3,4, XIE Junfa5, ZANG Shengtao5, QIN Long1,2,3,4   

  1. 1. R&D Center for Ultra-Deep Complex Reservoir Exploration and Development, CNPC, Korla 841000, Xinjiang, China;
    2. Engineering Research Center for Ultra-Deep Complex Reservoir Exploration and Development, Xinjiang Uygur Autonomous Region, Korla 841000, Xinjiang, China;
    3. Xinjiang Key Laboratory of Ultra-Deep Oil and Gas, Korla 841000, Xinjiang, China;
    4. Research Institute of Exploration and Development, PetroChina Tarim Oilfield Company, Korla 841000, Xinjiang, China;
    5. PetroChina Research Institute of Petroleum Exploration & Development-Northwest, Lanzhou 730020, China
  • Received:2025-05-29 Revised:2025-06-30 Online:2025-11-01 Published:2025-11-07

PDF (PC)

38

摘要： 为了提高地震速度的自动拾取效率，在常规K均值聚类算法基础上，提出了一种改进的加权K均值聚类方法，将该方法与常规K均值聚类在模型数据和塔里木盆地TD区块进行了应用，对拾取结果进行了分析。研究结果表明：①加权K均值聚类方法是通过设置速度谱幅值阈值剔除弱幅值速度点，设置比例系数，剔除离群速度点，在时间方向上设置时窗，寻找时窗内幅值最大的若干个速度点，以其平均速度与平均时间作为聚类中心的初值，根据初值剔除或者合并部分聚类中心，对不同的速度点赋予不同的权值，剔除每个聚类中心最边缘的速度点，使自动拾取的聚类中心逼近能量团中心，对存在速度反转的拾取结果，利用前2个拾取结果进行多次波的判断与剔除。②加权K均值聚类方法无需提供先验速度信息，实现了全自动的速度拾取，每次迭代均剔除部分速度点，大幅减少所需的迭代次数，加快了速度的计算，同时提高了拾取精度；应用加权K均值聚类方法对模型数据和塔里木盆地TD区块的实际资料进行了全自动速度拾取，比常规K均值聚类法的计算效率约提升了7倍，精度提高了1.7%。

关键词: 聚类中心, 加权K均值聚类, 速度拾取, 全自动, 加权, 无监督, 时间域, 速度建模

Abstract: To improve the efficiency of seismic velocity picking, an improved weighted K-means clustering method was proposed based on the conventional K-means clustering algorithm. Both of the improved method and the conventional K-means clustering were applied to model data and actual data of TD block in Tarim Basin, and the picking results were analyzed. The results show that: (1) The weighted K-means clustering method eliminates weak amplitude velocity points by setting velocity spectrum amplitude threshold, then set the proportional coefficient to eliminate the outlier velocity points. Set a time window in the time direction and search for several velocity points with the largest amplitudes within the time window. The average velocity and average time of the found velocity points are taken as the initial values of the clustering centers. Based on the initial values, remove or merge some clustering centers, assign different weights to different velocity points, remove the velocity points at the edge of each cluster center, make the automatically picked cluster centers approach the energy group center. For picking results with velocity reversal, the first two picking results are used to make multiple wave judgments and eliminations. (2) The weighted K-means clustering method does not require prior velocity information, and achieves fully automatic velocity picking. In each iteration, some velocity points are eliminated, significantly reducing the number of required iterations. It not only speeds up the calculation but also enhances the picking accuracy. The weighted K-means clustering method was applied to automatically velocity picking on the model data and the actual data of TD block in Tarim Basin, compared with the conventional K-means clustering method, its computational efficiency was increased by approximately 7 times and the accuracy was improved by 1.7%.

Key words: cluster center, weighted K-means clustering, velocity picking, fully automatic, weighted, unsupervised, time domain, velocity modeling

中图分类号: 

  • TE319
[1] SHI Wenwu, YONG Yundong, WU Kailong, et al. Velocity modeling and imaging of volcanic rocks in Laoyemiao area, Bohai Bay Basin[J]. Lithologic Reservoirs, 2021, 33(4): 101-110. 石文武, 雍运动, 吴开龙, 等. 渤海湾盆地老爷庙地区火山岩速度建模与成像[J]. 岩性油气藏, 2021, 33(4): 101-110.
[2] WANG Di, YUAN Sanyi, YUAN Huan, et al. Intelligent velocity picking based on unsupervised clustering with the adaptive threshold constraint[J]. Chinese Journal of Geophysics, 2021, 64(3): 1048-1060. 王迪, 袁三一, 袁焕, 等. 基于自适应阈值约束的无监督聚类智能速度拾取[J]. 地球物理学报, 2021, 64(3): 1048-1060.
[3] LI Yufeng, SUN Wei, HE Weiwei, et al. Prediction method of shale formation pressure based on pre-stack inversion[J]. Lithologic Reservoirs, 2019, 31(1): 113-121. 李玉凤, 孙炜, 何巍巍, 等. 基于叠前反演的泥页岩地层压力预测方法[J]. 岩性油气藏, 2019, 31(1): 113-121.
[4] ZHANG Tianze, WANG Hongjun, ZHANG Liangjie, et al. Application of ray-path elastic impedance inversion in carbonate gas reservoir prediction of the right bank of Amu Darya River [J]. Lithologic Reservoirs, 2024, 36(6): 56-65. 张天择, 王红军, 张良杰, 等. 射线域弹性阻抗反演在阿姆河右岸碳酸盐岩气藏储层预测中的应用[J]. 岩性油气藏, 2024, 36(6): 56-65.
[5] ZHAO Liang, SUN Xiaodong, LI Zhenchun, et al. Automatic velocity picking using dual-path convolutional neural network [J]. Oil Geophysical Prospecting, 2024, 59(6): 1206-1216. 赵亮, 孙小东, 李振春, 等. 利用双路卷积神经网络的速度自动拾取方法[J]. 石油地球物理勘探, 2024, 59(6): 1206-1216.
[6] TOLDI J L. Velocity analysis without picking[J]. Geophysics, 1989, 54(2): 191-199.
[7] ZHANG Jianbin, LIN Niantian, ZHANG Dong, et al. Intelligent picking of velocity spectrum based on nonlinear function [J]. Progress in Geophysics, 2016, 31(2): 856-860. 张建彬, 林年添, 张栋, 等. 基于非线性函数的速度谱智能拾取[J]. 地球物理学进展, 2016, 31(2): 856-860.
[8] HOU Bin, GUI Zhixian, XU Huiqun, et al. Application of multiattribute and neural network method to hydrocarbon reservoir prediction[J]. Lithologic Reservoirs, 2010, 22(3): 118-120. 侯斌, 桂志先, 许辉群, 等. 应用多属性神经网络方法预测油气[J]. 岩性油气藏, 2010, 22(3): 118-120.
[9] WANG Xiaoguang. Application of self-adaptive BP neural network to the prediction of shear wave velocity[J]. Lithologic Reservoirs, 2013, 25(5): 86-88. 王晓光. 自适应BP神经网络在横波速度预测中的应用[J]. 岩性油气藏, 2013, 25(5): 86-88.
[10] ZHANG Hao, ZHU Peimin, GU Yuan, et al. Automatic velocity picking based on deep learning[R]. San Antonio, 89th Society of Exploration Geophysicists International Exposition and Annual Meeting, 2019.
[11] WANG Wenlong, MCMECHAN G A, MA Jianwei, et al. Automatic velocity picking from semblances with a new deep-learning regression strategy: Comparison with a classification approach [J]. Geophysics, 2021, 86(2): 1-13.
[12] PARK M J, SACCHI M. Automatic velocity analysis using convolutional neural network and transfer learning[J]. Geophysics, 2020, 85(1): 33-43.
[13] ZHANG Hao, ZHU Peimin, GU Yuan, et al. Velocity auto-picking from seismic velocity spectra based on deep learning[J]. Geophysical Prospecting for Petroleum, 2019, 58(5): 724-733. 张昊, 朱培民, 顾元, 等. 基于深度学习的地震速度谱自动拾取方法[J]. 石油物探, 2019, 58(5): 724-733.
[14] 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. 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.
[15] WANG Hongtao, ZHANG Jiangshe, ZHAO Zixiang, et al. Automatic velocity picking using a multi-information fusion deep semantic segmentation network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-10.
[16] FABIEN-OUELLET G, SARKAR R. Seismic velocity estimation: A deep recurrent neural-network approach[J]. Geophysics, 2020, 85(1): 21-25.
[17] XUE Zhiwen, WU Xinming. Automatic velocity analysis with physics-constrained optimal surface picking[J]. Geophysics, 2023, 88(3): U71-U80.
[18] 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. 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.
[19] ZHANG Yang, ZHAO Pingqi, LU Fengming, et al. Recognition of single well reservoir architecture in alluvial fan based on K-means clustering and Bayes discrimination[J]. Oil Geophysical Prospecting, 2020, 55(4): 873-883. 张阳, 赵平起, 卢凤明, 等. 基于K均值聚类和贝叶斯判别的冲积扇单井储层构型识别[J]. 石油地球物理勘探, 2020, 55 (4): 873-883.
[20] ZHAO Xuan, YAN Jiabin, HU Tao. Application of cluster analysis in geophysics[J]. China Science and Technology Information, 2018, 30(15): 103-105. 赵玄, 严家斌, 胡涛. 聚类分析在地球物理中的应用进展[J]. 中国科技信息, 2018, 30(15): 103-105.
[21] GALVIS I S, VILLA Y, DUARTE C, et al. Seismic attribute selection and clustering to detect and classify surface waves in multicomponent seismic data by using K-means algorithm[J]. The Leading Edge, 2017, 36(3): 239-248.
[22] WANG Lide, WU Jie, XU Xingrong, et al. Intelligent velocity picking considering an expert experience based on the ChanVese model and mean-shift clustering[J]. Frontiers in Digital Humanities, 2023, 11: 1-16.
[23] ZHANG Peng, LU Wenkai. Automatic time-domain velocity estimation based on an accelerated clustering method[J]. Geophysics, 2016, 81(4): 13-23.
[24] CHEN Yuqing. Automatic semblance picking by a bottom-up clustering method[R]. Beijing, SEG 2018 Workshop: SEG Maximizing Asset Value Through Artificial Intelligence and Machine Learning, 2018.
[25] XIE Junfa, XU Xingrong, LAN Yan, et al. Automatic velocity picking with restricted weighted k-means clustering using prior information[J]. Frontiers in Digital Humanities, 2023, 10: 1-14.
[26] ZHOU Zhusheng, ZENG Weizu, LIU Siqin, et al. First arrival picking method by seismic multi-attribute based on weighted Kmeans clustering algorithm[J]. Acta Seismologica Sinica, 2020, 42(2): 177-186. 周竹生, 曾维祖, 刘思琴, 等. 基于加权K均值聚类的多属性初至拾取方法[J]. 地震学报, 2020, 42(2): 177-186.
[27] XIE Junfa, SUN Chengyu, WANG Xingmou, et al. The multicriteria velocity analysis of seismic data[J]. Geophysical and Geochemical Exploration, 2017, 41(3): 513-520. 谢俊法, 孙成禹, 王兴谋, 等. 地震资料的多准则速度分析方法[J]. 物探与化探, 2017, 41(3): 513-520.
[28] LIU Guochang, LI Chao, LIU Xingye, et al. Automatic stackingvelocity estimation using similarity-weighted clustering[J]. Geophysical Prospecting, 2018, 66(4): 649-663.
[29] SUN Mengyao, ZHANG Jie. The near-surface velocity reversal and its detection via unsupervised machine learning[J]. Geophysics, 2020, 85(3): 55-63.
[1] 王立德, 王小卫, 周辉, 吴杰, 张志强, 王建乐, 王德英, 冯刚. 一种基于改进共轭梯度法的弹性波全波形反演速度分层建模方法[J]. 岩性油气藏, 2023, 35(4): 61-69.
[2] 王建君, 李井亮, 李林, 马光春, 杜悦, 姜逸明, 刘晓, 于银华. 基于叠后地震数据的裂缝预测与建模——以太阳—大寨地区浅层页岩气储层为例[J]. 岩性油气藏, 2020, 32(5): 122-132.
[3] 徐兴荣, 苏勤, 王劲松, 寇龙江, 文一华, 王靖. 加权MPFI方法及其在三维连片处理中的应用[J]. 岩性油气藏, 2019, 31(1): 122-129.
[4] 刘文卿, 王孝, 胡书华, 张涛, 金保中. 测井与全方位道集联合各向异性参数建模及成像[J]. 岩性油气藏, 2018, 30(6): 83-88.
[5] 韩令贺, 胡自多, 冯会元, 刘威, 杨哲, 王艳香. 井震联合网格层析各向异性速度建模研究及应用[J]. 岩性油气藏, 2018, 30(4): 91-97.
[6] 楚翠金, 夏志林, 杨志强. 延川南区块致密砂岩气测井识别与评价技术[J]. 岩性油气藏, 2017, 29(2): 131-138.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!

陇ICP备17006252号
版权所有© 2018 中国石油集团西北地质研究所有限公司《岩性油气藏》编辑部

地址：甘肃省兰州市城关区雁儿湾路535号(陇ICP备17006252号)    电话：0931-8686158;8686058;8686288    邮箱：yxyqc@petrochina.com.cn

甘公网安备 62010202002827号

本系统由北京玛格泰克科技发展有限公司设计开发