Shale Porosity Prediction and Analysis of Main Controlling Factors in Area A of Sichuan Basin

孔隙度是深层页岩储层含气性评价及有利区优选的关键参数。为评估LightGBM算法在深层页岩储层孔隙度预测评价中的适用性，选取四川盆地A地区五峰组–龙马溪组深层页岩储层实测孔隙度为标签以及GR等7个测井参数为特征，建立基于LightGBM算法的孔隙度预测评价模型。研究结果表明：LightGBM算法模型的孔隙度评价精度及泛化能力(R2 = 0.979, RMSE = 0.622, MAPE = 31.5%)优于多元回归算法模型，能够精确地评价纵向上孔隙度变化规律；建立的LightGBM算法模型对深层页岩储层中孔隙度评价具有良好的适用性，为深层页岩含气性预测提供了新的思路，并明确研究区孔隙度主要受到TOC及黏土含量的共同控制。Porosity is a key parameter for evaluating gas bearing properties of deep shale reservoirs and selecting favorable areas. In order to evaluate the applicability of LightGBM algorithm in the prediction and evaluation of deep shale reservoir porosity, the measured porosity of deep shale reservoir in Wufeng Formation and Longmaxi Formation in area A of Sichuan Basin was selected as the label and 7 logging parameters such as GR were featured, and a porosity prediction and evaluation model based on LightGBM algorithm was established. The results show that the porosity evaluation accuracy and generalization ability of LightGBM algorithm model (R2 = 0.979, RMSE = 0.622, MAPE = 31.5%) are superior to the multiple regression algorithm model, and can accurately evaluate the longitudinal porosity variation law. The established LightGBM algorithm model has good applicability to the evaluation of porosity in deep shale reservoirs, providing a new idea for the prediction of gas content in deep shale reservoirs, and it is clear that the porosity of the study area is mainly controlled by TOC and clay content.