核主成分分析法在测井浊积岩岩性识别中的应用

在复杂岩性油气藏储层评价中，如何合理利用测井曲线信息判别岩性是一个难题，东营凹陷董集洼陷浊积岩岩性复杂，采用常规交会图法以及主成分分析法都难以有效地识别岩性。为了解决这一问题，基于粒子群算法以及核函数理论，结合该区的测井响应特征，采用核主成分分析法，建立新的主成分计算方法，选取该区实测的自然伽马测井（GR）、声波时差测井（AC）、中子孔隙度测井（CNL）、密度测井（DEN）、原状地层电阻率（RT）构建出五个主成分变量，其中前两个主成分变量累积贡献率达到了93.83%，可以有效地代替原始多维测井信息。实例分析结果表明，利用前两个主成分变量主成分进行交会分析，可以有效地识别浊积岩的岩性，并将该方法在研究区进行了试验，岩性识别准确率达到了90%，较传统方法具有更高的岩性识别精度，取得了良好的应用效果。

Application of kernel principal component analysis in well logging turbidite lithology identification

It is difficult to identify the lithology on well-logging data in the evaluation of complex lithologic reservoirs.Due to complexity of turbidite reservoirs in Dongji Sag, Dongying Depression, conventional cross-plot and principal component analysis methods fail to identify their lithology.In order to solve this problem, based on particle swarm optimization and kernel function theory and combining with log response characteristics of the area, an improved principal component analysis method is used to establish a new principal component calculation.Five principal component variables are constructed with measured natural gamma-ray logging (GR), acoustic logging (AC), compensated neutron porosity logging (CNL), density logging (DEN), and virgin zone resistivity (RT) of reservoirs.The accumulate contribution rate of the first two principal component variables reached 93.83%, which can effectively replace the original multi-dimensional logging information.The proposed method is tested in the study area.Based on our application results, this proposed method can effectively identify the lithology of turbidite reservoirs, and its identification rate reaches up to 90%.