基于机器学习属性建模地震响应特征数值模拟及分析  

Numerical simulation and analysis of seismic response characteristics based on machine learning attribute modeling

作　　者：陈榆桂 阎建国[1] 邓儒炳 王钰鹏 李辉 CHEN Yugui;YAN Jianguo;DENG Rubing;WANG Yupeng;LI Hui(Chengdu University of Technology College of Geophysics,Chengdu 610059,China)

机构地区：[1]成都理工大学地球物理学院,成都610059

出　　处：《物探化探计算技术》2023年第1期28-35,共8页Computing Techniques For Geophysical and Geochemical Exploration

基　　金：中国石油油气地球物理前沿理论与技术课题(RIPOED-2020-JS-121)。

摘　　要：基于属性建模的数值模拟方法正在成为复杂地震响应特征分析的一种有力手段,这里在二维数值模拟方法研究中,提出和采用了基于多种信息的属性建模方法,包括将机器学习算法应用到属性建模中,利用U-net网络改进了低信噪比资料的断缝系统识别精度,并将这种识别结果用于潜山内幕复杂断缝系统的建模中,再通过有限差分数值模拟方法,得到了较为符合地下实际条件的地震响应特征数值模拟结果,并且与实际资料上的复杂波场特征对应度高,实现了几何和物理参数的空间变化模拟,提高了数值模拟的精度和可靠度。应用提出的属性建模地震响应特征分析方法及流程,证实了研究区潜山内幕断缝系统的地震响应特征主要呈现为“高陡网状反射”和“短轴不连续反射”的特征及相关结论,为研究区复杂潜山储层地震响应特征认识和裂缝预测提供了借鉴。Numerical simulation method based on attribute modeling is becoming complex seismic response analysis of a powerful means,based on the two-dimensional numerical simulation method in the study,put forward and used the property modeling method based on a variety of information,including the applications of machine learning algorithms to attribute modeling,the use of U-net network to improve the low SNR data of broken seam system identification accuracy,And used for the identification results are similar in the modeling of complex faulted seam system,again through finite difference numerical simulation method,the more accord with the actual conditions of the underground seismic response characteristics of the results of numerical simulation,and the characteristics of complex wave field and the actual data of the corresponding degree is high,the geometrical and physical parameters of spatial variation simulation,improved the accuracy and reliability of the numerical simulation.Application of modeling in this paper,the properties of the seismic response characteristics analysis method and process,confirmed that the buried hill in the study area inside broken seam seismic response characteristics of the system are mainly presented as"high and steep mesh reflection"and"short axis discontinuous reflection"characteristics and relevant conclusions,as the research area of complex buried hill reservoir seismic response characteristics and fracture prediction provides a scientific basis.

关 键 词：属性建模 数值模拟 机器学习 地震响应特征 裂缝预测 

分 类 号：P631.4[天文地球—地质矿产勘探]