基于随钻数据的岩性智能识别方法  

An intelligent identification method of lithology based on data while drilling

作　　者：邴磊 倪朋勃 张文颖 杨毅 BING Lei;NI Pengbo;ZHANG Wenying;YANG Yi(China France Bohai Geoservices Co.,Ltd.,Tianjin 300450,China)

机构地区：[1]中法渤海地质服务有限公司,天津市300450

出　　处：《录井工程》2024年第4期26-31,共6页Mud Logging Engineering

基　　金：中法渤海地质服务有限公司“基于多源信息融合的随钻地层岩性智能判识系统开发项目”(编号:CFB_TG_TI_2020_014)。

摘　　要：现场岩性准确识别在油气勘探开发过程中起着至关重要的作用,随着勘探难度增加,现场岩性识别难度也相应加大。针对现场随钻地层岩性识别的难题,提出一种基于随钻数据的岩性智能识别方法,通过对随钻地层元素和钻井参数进行分析,提取特征向量,建立不同岩性的多特征样本库,并用最大最小蚂蚁系统优化算法的广义回归神经网络模型实现了基于随钻数据的岩性智能识别。将该方法应用于实际随钻岩性识别中,对非特征提取数据与特征提取融合数据所构建的MMAS-GRNN模型进行比较,结果表明:基于多源信息特征提取融合数据建立的MMAS-GRNN模型地层岩性识别准确率达到了90.71%,比基于非特征提取数据的模型准确率高出了4.76%,展现出特征提取融合后的多源数据在地层岩性识别效果方面的优越性。Accurate identification of site lithology plays an important role in oil and gas exploration and development.As the difficulty of exploration increases,the difficulty of lithology identification on site also increases accordingly.An intelligent identification method of lithology based on data while drilling is proposed to solve the problem of on-site stratigraphic lithology identification while drilling.This method analyzes the formation elements and parameters while drilling,extracts the eigenvectors,establishes the multiple-feature collection library of different lithology,and realizes the lithology intelligent identification based on data while drilling by using the generalized regression neural network model of the Max-Min Ant System optimization algorithm.This method is applied to actual lithology identification while drilling,and the MMAS-GRNN model constructed from non-feature extraction data and feature extraction fusion data is compared.The experimental results show that the accuracy rate of stratigraphic lithology identification of MMAS-GRNN model based on multi-source information feature extraction fusion data reaches 90.71%,which is 4.76%higher than that of non-feature extraction data model,showing the superiority of multi-source data after feature extraction fusion in stratigraphic lithology identification effect.

关 键 词：随钻数据 岩性识别 特征提取融合 最大最小蚂蚁系统 

分 类 号：TE132.1[石油与天然气工程—油气勘探]