摘要
电成像测井广泛应用于碳酸盐岩、砂砾岩和火成岩等复杂储层的测井评价,对于计算孔隙度、识别裂缝和划分岩性具有重要作用.但在大井眼的情况下,电成像图像无法做到全井眼覆盖,需要对电成像图像上的空白条带进行充填,保证后期处理和解释的精度.结合深度学习框架,提出一种基于卷积神经网络模型的空白条带充填方法.在没有大量学习样本的情况下,通过优化卷积神经网络模型结构,捕获图像上的大量底层先验统计特征,实现整幅图像的结构和纹理特征信息的推理.通过与Filtersim主流充填方法的充填效果比较,发现该方法对于砂泥岩剖面和砂砾岩体的电成像测井图像空白条带充填,都具有较好的效果.
Electrical imaging logging has been widely used for evaluating carbonate,glutenite and igneous rock,and plays a very important role in porosity calculation,fracture recognition and lithology classification.However in a large hole,the image cannot cover the whole borehole,and the gaps on the image should be filled to ensure the accuracy of subsequent data processing and interpretation.A gap filling method is proposed based on a convolutional neural network(CNN)after introducing a deep learning framework.It first captures a great deal of low-level image prior statistic information by optimizing the CNN structure without a large amount of training samples,and then infers the structure and texture characteristic on the whole image.Compared with the popular Filtersim method,this method provides much better results for sandy shale and glutenite rock.
作者
王哲峰
高娜
曾蕊
杜雪菲
杜欣睿
陈思宇
WANG Zhefeng;GAO Na;ZENG Rui;DU Xuefei;DU Xinrui;CHEN Siyu(Changqing Branch,China Petroleum Logging CO.LTD.,Xi’an,Shaanxi 710201,China)
出处
《测井技术》
CAS
2019年第6期578-582,共5页
Well Logging Technology
关键词
测井评价
电成像测井
深度学习
深度神经网络
空白条带充填
卷积神经网络
大井眼
log evaluation
electrical imaging logging
deep learning
deep neural network
gap filling
convolution neural network
large hole
