摘要
横波时差资料对开展致密砂岩储层水平井井壁稳定性与压裂效果研究有着关键作用。受开发成本的制约,横波时差测井资料极少,对研究致密砂岩力学性质造成很大困难。以井径、自然伽马和纵波时差等常规测井资料为基础,提出了基于GA-BP神经网络的致密砂岩横波时差预测方法。利用定边油田L区D166井长7、长8段数据,分别进行了GA-BP模型和BP模型的训练和检验,并对比分析了2种模型的预测效果。结果表明,GA-BP模型不受井眼环境、岩性和沉积环境等因素的影响,平均绝对百分比误差较BP模型小3.109个百分点,精准性更高、泛化性更强、可靠性更好。该方法对提高横波时差预测精度具有实际应用价值,为后续研究奠定了基础。
Conducting research on the stability of wellbore and fracturing effects in horizontal wells in tight sandstone reservoirs plays a crucial role in the utilization of shear wave time difference data.Due to the constraints of development costs,there are very few shear wave time difference logging data,which poses great difficulties in studying the mechanical properties of tight sandstone.This study proposed a shear wave travel time prediction method for tight sandstone based on the GA-BP neural network using conventional logging data such as well diameter,natural gamma ray,and compressional wave travel time.The GA-BP model and BP model were trained and tested using data from the Chang 7 and Chang 8 sections of well D166 in the Dingbian Oilfield L area,and a comparative analysis of the prediction performance of the two models was conducted.The results showed that the GA-BP model was not affected by factors such as wellbore environment,lithology,and sedimentary environment.The average absolute percentage error was 3.109 percentage points smaller than the BP model,that had higher accuracy,stronger generalization ability,and better reliability.This method has practical application value in improving the accuracy of shear wave time difference prediction and lays the foundation for subsequent research.
作者
任宇飞
强璐
程妮
白耀文
张军东
王瑞生
Ren Yufei;Qiang Lu;Cheng Ni;Bai Yaowen;Zhang Jundong;Wang Ruisheng(Exploration and Development Research Center,Yanchang Oil Field Co.,Ltd.,Yan′an 716000,China;Dingbian Oil Production Plant,Yanchang Oil Field Co.,Ltd.,Yulin 718600,China;Jingbian Oil Production Plant,Yanchang Oil Field Co.,Ltd.,Yulin 718600,China)
出处
《能源与环保》
2024年第4期124-129,共6页
CHINA ENERGY AND ENVIRONMENTAL PROTECTION
基金
延长油田股份有限公司科技项目(ycsy2021ky-B-15-6)。
关键词
致密砂岩
横波时差
BP神经网络
遗传算法
预测方法
tight sandstone
shear wave time difference
BP neural network
genetic algorithm
predication method
