中国石油大学学报（自然科学版）2025，Vol.49Issue(3)：84-96,13.DOI:10.3969/j.issn.1673-5005.2025.03.009
基于测录数据的潜山地层压力评价新方法
A new method of formation pressure evaluation of buried hill based on logging and drilling data
摘要
Abstract
Accurate prediction of formation pressure is crucial for the safe and efficient exploration and development of bed-rock buried-hill reservoirs.Taking the buried-hill formation in the Qiongdongnan Basin as a case study,this paper analyzes pressure distribution and logging response characteristics using a comprehensive dataset including logging and related geolog-ical data.A log parameter cross-plot for overpressure amplitude identification is constructed,and a new machine learning-based evaluation method is developed using coupled binary and multivariate logging data.The results reveal that there is no clear correlation between overpressure in buried-hill zone and overpressure in overlying sedimentary strata.Instead,the mag-nitude and distribution of pressure are primarily controlled by pressure-generation and pressure-retention conditions.Logging features such as low acoustic velocity and low density indicate well-developed fractures in the buried-hill formation,providing favorable conditions for fluid charging and overpressure accumulation.Low values in parameters such as drilling cuttings in-dex(Dc)and mechanical specific energy(MSE)suggest that,under similar structural frameworks and fracture densities,the rock requires less energy to break,has lower bottom hole pressure holding effect and triaxial compression strength,and exhibits higher overpressure amplitudes.Traditional methods based on single log parameters or simple binary linear/nonlinear fitting are inadequate for evaluating formation pressure in buried-hill reservoirs.The newly developed machine learning meth-od,based on random forest modeling of coupled binary/multivariate logging data,demonstrates superior performance and ap-plicability.These findings provide valuable technical support for the safe and efficient development of buried-hill oil and gas resources.关键词
机器学习/测井及录井参数/地层压力评价/基岩潜山Key words
machine learning/logging and drilling parameters/formation pressure evaluation/bedrock buried hill分类
能源科技引用本文复制引用
陈现军,徐长敏,郭书生,陈玉鑫,孙挺,师淑怡..基于测录数据的潜山地层压力评价新方法[J].中国石油大学学报（自然科学版）,2025,49(3):84-96,13.基金项目
中海石油(中国)有限公司重大科技项目(KJGG2022-0405) （中国）
