基于神经网络的水力压裂监测数据校正方法研究  

The study of correcting the data of monitoring fracture azimuth based on neural network

作　　者：王佳[1] 李亭亭[1] 司晓波[1] 朱凯光[1] 

机构地区：[1]吉林大学仪器科学与电气工程学院,长春130061

出　　处：《地球物理学进展》2013年第4期2181-2185,共5页Progress in Geophysics

基　　金：吉林省科技支持计划工业高新技术重点项目(20110317)资助

摘　　要：油气田井况地表复杂,为准确推断水力压裂裂缝方位,提出一种基于BP神经网络的压裂裂缝监测数据校正方法.本文将复杂地表条件下测点布置不均匀时的电位梯度响应,分解为正常场响应和畸变场响应.利用神经网络估算因测点布置不均匀引起的畸变场响应,通过剔除电位梯度响应中的畸变场量,获得测点布置均匀条件下的电位梯度响应,即正常场响应.采用井地电阻率法三维正演模拟裂缝推断过程,结果表明神经网络可有效估计畸变场响应,提高了利用电位梯度响应推断裂缝方位的准确度,具有良好的应用前景.The surface of oil and gas field is complex. To find out the position of hydraulic fracture azimuth of oil and gas field, we released solution of correcting the data of monitoring fracture based on BP neural network by dividing the potential gradient response into normal field and abnormal field on the condition that the electrodes are arranged asymmetric on complex surface, estimate the abnormal field brought about by the asymmetric electrodes by using neural network and remove it from the potential gradient response, the results are the potential gradient response of symmetric electrodes and it is normal field response. Three dimensional well-to-ground resistivity method forward modeling simulates the fracture monitoring process whose results indicate that neural network can effectively estimate the abnormal field, which greatly improves the accuracy of potential gradient response to fracture azimuth. And there will be much more application prospect.

关 键 词：裂缝方位 电阻率法 电位梯度响应 BP神经网络 畸变场 

分 类 号：P631[天文地球—地质矿产勘探]