基于BP神经网络的储集层物性参数预测与评价——以准噶尔盆地东部北三台地区为例  

Prediction and Evaluation about Physical Parameter of Reservoir Stratum Based on Neural Network ——Taking Beisantai Area of Junggar Basin as an Example

作　　者：李斌斌[1] 郭大立[1] 毛新军[2] 杜国峰[1] 陈超峰[2] 

机构地区：[1]西南石油大学理学院,成都610500 [2]新疆油田公司勘探公司,新疆克拉玛依834000

出　　处：《重庆科技学院学报（自然科学版）》2014年第2期39-42,共4页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

基　　金：国家科技重大专项重点项目示范工程19(2011ZX05062)

摘　　要：利用测井资料建立沉积岩岩性识别的测井交会图版,并对研究区储集层岩性进行识别。用所建立的遗传BP神经网络模型预测储集层孔渗参数,将识别的岩性数字化参数及常规测井参数作为输入项,预测精度大幅提高。用平面克里金插值法分析井层的孔隙度,预测储集层参数的平面分布,预测结果与沉积相分布的吻合度较高。Using the well - logging materials, this paper establishes the well - logging crossplot of the sedimentary rocks in the study area, and does the recognition for the area＇s reservoir. The study shows the reservoir＇s lithology is a key factor affecting physical property. When using the Genetic - BP neural network model predicts the reservoir holes and permeability, applying the parameters about the recognized lithology digitized and the formal well - log- ging as the input item, which improves the prediction precision sharply. The distribution of reservoir parameters is predicted under domination of even porosity and permeability of each segment of the key wells.

关 键 词：准噶尔盆地 岩性识别 遗传BP神经网络 储集层 

分 类 号：P618[天文地球—矿床学]