基于Kohonen神经网络的页岩油岩相测井识别方法——以吉木萨尔凹陷二叠系芦草沟组为例

Shale Oil Lithofacies Identification by Kohonen Neural Network Method:The case of the Permian Lucaogou Formation in Jimusaer Sag

准噶尔盆地吉木萨尔凹陷二叠系芦草沟组岩性复杂,岩相多样,亟需建立配套的岩相测井识别方法。综合利用岩心薄片、常规测井与成像测井等资料对芦草沟组的岩性、沉积构造等特征进行研究,在此基础上进行岩相类型划分,并分析不同岩相的测井响应特征,建立岩相测井评价模型。结果表明,研究区主要由云质泥岩、泥质粉砂岩、粉砂质泥岩等岩性组成;根据岩心的矿物组成,考虑在岩心尺度上的可区分性、测井中的可识别性,划分出块状白云质泥岩相、平行/波状层理泥质粉砂岩相、块状泥岩相、块状泥晶白云岩相、平行层理粉砂岩相和波状/水平层理粉砂质泥岩相6种主要岩相类型。在此基础上通过岩心刻度测井资料建立岩相判别准则,并利用Kohonen神经网络方法实现单井岩相测井自动判别,划分结果与薄片匹配较好。研究成果可提高未取心井的岩相识别效率与精度,为该地区页岩油勘探提供一定的理论依据与方法支撑。

The Permian Lucaogou Formation in the Jimusaer Sag,Junggar Basin,has a complex lithology and various lithofacies,and urgently requires a lithofacies-matching logging identification method to be established.In this study,core slice,conventional logging and image logging data,among other methods,were comprehensively used to examine the lithology,sedimentary structure and other characteristics of the Lucaogou Formation in order to classify lithofacies types and analyze their logging response,and to establish a lithofacies logging evaluation model.The mineral composition of the core indicated that the study area comprises mainly dolomitic mudstone,argillaceous siltstone,silty mudstone and other lithologies.This took into consideration the distinguishability of the lithofacies at the core scale.Lithofacies identification determined for logs was divided into six main types:massive dolomitic mudstone,silty mudstone with parallel/wavy bedding,massive mudstone,massive micrite dolomite,siltstone with parallel bedding,and argillaceous siltstone with wavy/horizontal bedding.Adopting these criteria to establish the scale,the Kohonen neural network method of single well logging was used to automatically determine the lithofacies and to match these divisions as closely as possible.The research results indicated improvement in the efficiency and accuracy of lithofacies identification for coreless wells.The study also provides a theoretical basis and a method supporting shale oil exploration in this area.