摘要
岩石的断裂韧性是影响裂缝扩展及延伸的重要因素,同时也是储层可压性评价的关键参数。但目前断裂韧性直接测试较为复杂,且现有的断裂韧性预测方法多基于断裂韧性与其他物理参数之间的拟合关系,难以形成整个井段的连续剖面。通过室内断裂韧性实验,分析了页岩断裂韧性与其他物理力学参数之间的关系,建立了断裂韧性拟合公式,同时采用XGBoost模型,利用地球物理测井数据,通过改进的Q-learning算法优化XGBoost模型超参数,实现了岩石断裂韧性的预测。研究结果表明,Ⅰ型断裂韧性与抗拉强度、声波速度相关性较高,与密度相关性较低,与纵波速度、横波速度、抗拉强度、岩石密度均成正相关。基于改进的Q-learning优化断裂韧性智能预测的XGBoost模型预测准确性较高,预测断裂韧性与拟合断裂韧性相关度高达0.981,所提出的岩石断裂韧性预测模型是可靠的,可为压裂工程设计提供参考。
Fracture toughness of rock is an important factor affecting fracture propagation and extension,and it is also a key parameter of reservoir compressibility evaluation.However,the direct measurement of fracture toughness is complicated at present,and the existing fracture toughness prediction methods are mostly based on the fitting relationship between fracture toughness and other physical parameters,which is difficult to form a continuous section of the whole well section.In this paper,the relationship between shale fracture toughness and other physical and mechanical parameters is analyzed through indoor fracture toughness experiments,and the fracture toughness fitting formula is established.At the same time,XGBoost model is adopted,geophysical logging data is utilized,and the hyperparameters of XGBoost model are optimized by improved Q-learning algorithm,so as to realize the prediction of rock fracture toughness.The results show that typeⅠfracture toughness is highly correlated with tensile strength and acoustic velocity,but poorly correlated with density,and positively correlated with P-wave velocity,S-wave velocity,tensile strength and rock density.The XGBoost model based on improved Q-learning optimized intelligent prediction of fracture toughness has a high prediction accuracy,and the correlation between predicted fracture toughness and fitted fracture toughness is as high as 0.981.It is considered that the rock fracture toughness prediction model proposed in this paper is reliable and can provide reference for fracturing engineering design.
作者
张艳
王宗勇
张豪
吴建成
祝春波
吴高平
ZHANG Yan;WANG Zongyong;ZHANG Hao;WU Jiancheng;ZHU Chunbo;WU Gaoping(School of Petroleum Engineering,Yangtze University,Wuhan 430100,Hubei;National Engineering Research Centre of Oil&Gas Drilling and Completion Technology,Wuhan 430100,Hubei;State Key Laboratory of Petroleum Resources and Engineering(China University of Petroleum(Beijing)),Beijing 102249;NO.9 Oil Production Plant,Daqing Oilfiled Company Limited,CNPC,Daqing 163000,Heilongjiang)
出处
《长江大学学报(自然科学版)》
2025年第5期58-65,共8页
Journal of Yangtze University(Natural Science Edition)
基金
油气资源与工程全国重点实验室开放课题“页岩油储层井壁失稳机理及对策技术研究”(PRE/open-2307)。
