物探化探计算技术2024，Vol.46Issue(3)：284-291,8.DOI:10.3969/j.issn.1001-1749.2024.03.04
基于3D U-Net++卷积神经网络的断层识别方法及应用
Fault recognition method and application based on 3D U-Net++convolution neural network
李卿武 1王兴建 2张永恒 1文雪梅 1陈阳 1王崇名 1廖万平1
作者信息
- 1. 成都理工大学地球物理学院,成都 610059
- 2. 成都理工大学地球物理学院,成都 610059||成都理工大学油气藏地质及开发工程国家重点实验室,成都 610059
- 折叠
摘要
Abstract
Fault interpretation is the basis and key to seismic data interpretation,and accurate and reasonable fault identifi-cation plays a vital role in oil and gas exploitation.With the increasing demand of oil fields for fault interpretation accuracy,the accuracy of traditional fault interpretation methods based solely on artificial attributes such as coherence,curvature,etc.,can-not meet the requirements.Based on the U-Net convolution neural network model,this paper proposes an automatic fault recognition method,which can automatically extract faults from any 3D seismic image.In this paper,the model carries out au-tomatic fault identification on the actual seismic data of two blocks under the training of sufficient sample sets and analyzes and compares the identification results.The experimental results show that the model can automatically recognize faults from arbitrary 3D seismic data,and the fault recognition results based on the 3D U-Net++network model have significantly improved the accuracy of the recognition results compared with the traditional U-Net network.It also shows a good effect on the recognition of minor faults in-side the buried hill and significantly improves the efficiency of conventional and complex fault recognition.关键词
断层识别/三维地震数据/卷积神经网络/3D U-Net++Key words
fault identification/3D seismic data/convolution neural network/3D U-Net++分类
地质学引用本文复制引用
李卿武,王兴建,张永恒,文雪梅,陈阳,王崇名,廖万平..基于3D U-Net++卷积神经网络的断层识别方法及应用[J].物探化探计算技术,2024,46(3):284-291,8.
