滩相碳酸盐岩储层测井物性参数智能机器解释——以JN区块飞三段为例

Intelligent Machine Interpretation of Logging Physical Parameters of Beach Facies Carbonate Reservoir —Taking Fei-3 Member of JN Block as an Example

摘要: 滩相碳酸盐岩储层因强烈的非均质性和复杂的孔隙结构，其测井物性参数解释一直是研究的难点和重点。目前的研究主要集中在通过岩心、成像测井、核磁共振等特殊资料，结合机器学习等新技术，建立针对不同岩石物理相和孔隙结构的解释模型，以实现储层参数精细计算和有效性评价。传统AC-POR预测方法难以准确解释储层的孔隙度，为准确了解JN区块飞三段储层的孔隙度和渗透率，更精确地评估储层的含油性和产能，本文基于录井、测试等资料，利用AC、GR、DEN等测井数据，通过机器学习回归法(LightGBM模型回归和梯度提升决策树(GBDT))，对JN区块飞三段储层孔隙度和渗透率进行解释模拟，显著提高了测井参数预测的准确性。结果表明，与传统方法相比，机器学习模型显著提高了测井参数的预测精度，为JN区块飞三段储层评价与开发提供了可靠依据。

Abstract: Due to the strong heterogeneity and complex pore structure of beach carbonate reservoirs, the interpretation of logging physical parameters has always been a difficulty and focus of research. The current research mainly focuses on the establishment of interpretation models for different rock physical phases and pore structures through special data such as core, imaging logging, and nuclear magnetic resonance, combined with new technologies such as machine learning to achieve fine calculation and effectiveness evaluation of reservoir parameters. The traditional AC-POR prediction method is difficult to accurately explain the porosity of the reservoir. In order to accurately understand the porosity and permeability of the reservoir in the Fei-3 Member of the JN Block and more accurately evaluate the oil content and productivity of the reservoir, this paper is based on data from mud logging and testing. Using AC, GR, DEN and other logging data, through machine learning regression method (LightGBM model regression and gradient boosting decision tree (GBDT)), the porosity and permeability of the Fei-3 Member of the JN Block are explained and simulated, which significantly improves the accuracy of logging parameter prediction. The results demonstrate that the machine learning models significantly improve the prediction accuracy of logging parameters compared to conventional methods, providing a reliable foundation for reservoir evaluation and development in the Fei-3 Member of the JN Block.