摘要
顺北地区发育受走滑断裂控制的断控缝洞型油气藏,断裂具控储、控藏、控富的作用。断裂识别和评价是顺北超深层领域油气勘探部署工作的核心,随着勘探开发的深入,主干断裂带之间的低序级断裂成为近期勘探的重点领域,但低序级断裂尺度小,目的层埋深大,地震响应特征不清,识别难度大,常规相干等断裂检测方法不适用,严重制约了勘探开发进程。针对以上难点,在深入研究低序级断裂地质地震特征基础上,从低序级断裂样式和地震响应特征出发,研发低序级断裂人工智能检测技术,对中小尺度断裂建立训练样本集,训练并优化用于断裂识别的深度神经网络,提升低序级断裂识别精度。应用以上方法在顺北中部地区检测出多条低序级断裂,同时,结合构造动力学背景及断裂发育特征,对检测结果进行二次解释,实现低序级断裂带的描述评价和目标优选。
The Shunbei area has developed strike-slip fault-controlled fractured-vuggy reservoirs,where faults play critical roles in reservoir formation,hydrocarbon accumulation,and enrichment distribution.Fault identification and evaluation form the core of oil and gas exploration in the ultra-deep formations of Shunbei.With the advancement of exploration,low-order faults between major fault zones have become key exploration targets.However,these low-order faults present significant identification challenges due to their small scale,great burial depth of target formations,ambiguous seismic responses,and the ineffectiveness of conventional fault detection methods like coherence analysis,which severely constrains exploration progress.To address these challenges,we developed an AI-based detection technology for low-order faults through in-depth research on their geological and seismic characteristics.Starting from the fault patterns and seismic response features of low-order faults,we established a training dataset for medium-small scale faults,trained and optimized deep neural networks for fault identification,thereby significantly improving the recognition accuracy of low-order faults.Applying this method in the central Shunbei area,we successfully detected multiple low-order faults.Furthermore,by integrating structural dynamics background and fault development characteristics,we conducted secondary interpretation of the detection results,achieving comprehensive description,evaluation,and target optimization of low-order fault zones.
作者
李弘艳
李海英
靖剑坤
龚伟
韩俊
刘成芳
Li Hongyan;Li Haiying;Jing Jiankun;Gong Wei;Han Jun;Liu Chengfang(SINOPEC Northwest Oil Company,Urumqi,Xinjiang,830011,China;China University of Geosciences(Wuhan),Wuhan,Hubei,430074,China)
出处
《新疆地质》
2025年第2期287-293,共7页
Xinjiang Geology
基金
中石化科技研发超深层碳酸盐岩缝洞体人工智能识别研究(P23137)项目资助。
关键词
顺北地区
低序级断裂
人工智能
样本集
深度学习
断裂识别
Shunbei region
Low-order faults
Artificial intelligence
Sample set
Deep learning
Fracture recognition
