面向深层干热岩体的全波形速度反演建模方法  

Velocity building using full waveform inversion for deep hot dry rock

作　　者：杨继东 于由财 刘朋 高建明 黄建平[1] 杨永红 YANG Jidong;YU Youcai;LIU Peng;GAO Jianming;HUANG Jianping;YANG Yonghong(School of Geosciences in China University of Petroleum(East China),Qingdao 266580,China;Shandong Energy Group South America Limited,Qingdao 266580,China;Shengli Oilfield Exploration and Development Research Institute,SINOPEC,Dongying 257000,China)

机构地区：[1]中国石油大学(华东)地球科学与技术学院,山东青岛266580 [2]山东能源集团南美有限公司,山东青岛266580 [3]中石化胜利油田勘探开发研究院,山东东营257000

出　　处：《中国石油大学学报（自然科学版）》2024年第1期70-76,共7页Journal of China University of Petroleum(Edition of Natural Science)

基　　金：国家自然科学基金优秀青年项目(海外)(ZX20230152);山东能源集团深层高温地热重大科技项目(SNKJ2022A06-R23);山东省海洋科技基金青岛海洋科学与技术国家试点实验室项目(2021QNLM020001);创新研究群体项目(41821002);中国石油重大科技专项(ZD2019-183-003)。

摘　　要：随着勘探深度增加,地震信号衰减严重,常规速度建模方法精度较低,无法满足深层干热岩体高精度勘探的需求。为提高速度建模的精度,采用全波形反演方法,但受局部优化算法的限制,其应用存在收敛速度慢、反演深度浅以及容易陷入局部极值等问题。为此,对梯度应用预条件处理和平滑处理,并使用共轭梯度优化算法,以解决地下照明不均匀问题,提高反演深度、精度和收敛速度。同时,为缓解局部极小值问题,引入局部相似性全波形反演方法来更新初始模型,并对构建的干热岩模型进行数值测试。结果表明:即使在初始模型极不准确的情况下,该方法仍能避免周波跳跃的不利影响,实现稳健的迭代更新;该方法能够显著增加反演深度和精度,并且对高速岩体有较好的刻画,最终能获得浅、中、深全层系高精度速度模型;提出的全波形反演方法为深层高温花岗岩体的勘探提供了一个切实可行的建模流程。Severe seismic signal attenuation with increasing exploration depth renders conventional velocity building methods less accurate for high-precision exploration of deep hot dry rock(HDR)bodies.To address this challenge,the full waveform inversion(FWI)method is used to enhance the accuracy of velocity building.However,FWI encounters issues like slow convergence,limited inversion depth,and susceptibility to local optimal solutions due to constraints imposed by local optimization algorithms.To mitigate these challenges,we introduce a novel approach that utilizes preconditioning and smoothing gradients in conjunction with the conjugate gradient optimization algorithm.This strategy aims to rectify unbalanced illumination issues and speed up convergence.Additionally,to alleviate the issue of local minima,a local coherence misfit is integrated into FWI to update the velocity model.Numerical simulations conducted on typical HDR models show that the proposed method is capable of circumventing the negative effects of cycle skipping issues,ensuring stable iterative updating even with highly inaccurate initial model.Furthermore,the method significantly improves the depth and accuracy of inversion,providing a more precise depiction of high-velocity rock bodies,eventually obtaining high-accuracy velocity models in both shallow and deep layers.The FWI method proposed in this study provides a practical and efficient velocity building tool for detecting deep high-temperature granite bodies.

关 键 词：干热岩 高温花岗岩体 全波形反演 深层速度建模 局部相似性 

分 类 号：P631.4[天文地球—地质矿产勘探]