摘要
蜀南地区须家河组地层岩性复杂,属于低孔、低渗致密气藏,致密砂岩测井识别是该地层天然气勘探中所面临的关键问题和难点之一。针对常规测井岩性识别准确率不高的状况,提出利用BP神经网络进行复杂岩性测井识别。神经网络识别输入样本采用选取靠近其所属岩类的平均值的样本以提高神经网络的精度,并对输入数据进行标准化以消除测井曲线间量纲的影响。运用BP神经网络模型对研究区域复杂岩性进行识别,识别结果与岩心岩性和录井岩性较为相符,对该区域的储层识别和沉积相的研究具有一定的参考价值。
The lithology of Xujiahe formation is complicated in the south of Sichuan basin, and the gas reservoirs have low porosity and low permeability. The lithology of tight sandstone identified by logging is the key in gas prospecting. Considering the low accuracy of conventional lithology identification methods, this paper proposed a method that the different types of tight sandstone are distinguished by BP neural net on well logging data. To im- prove identification accuracy and standardize the input data to remove the effect between the log dimensions, the BP neural network will select the samples of the average value which are similar to its petrographie category from the input data samples. The complex lithology are identified by the BP neural network model in the study area, the result show that the identified result relatively tally with the lithology of the core and lithology of the well logging, it has a certain reference value for the reservoir identification and study of deposition phase in the area.
出处
《油气地球物理》
2013年第1期39-42,共4页
Petroleum Geophysics
基金
西南油气田分公司外协项目
项目编号:XNS19JS2011-046
关键词
BP神经网络
须家河组
致密砂岩气藏
岩性识别
BP neural network, tight sandstone gas reservoir, lithology identification and Xujiahe Formation
