大庆石油地质与开发2025，Vol.44Issue(3)：70-76,7.DOI:10.19597/J.ISSN.1000-3754.202312020
应用人工智能方法计算致密气藏可采储量
Applying artificial intelligence methods to calculate recoverable reserves of tight gas reservoirs:Taking BP neural network as an example
摘要
Abstract
In response to harsh conditions of traditional calculation methods for recoverable reserves,especially for tight gas reservoirs with problems of much workload,large calculation errors,and incomplete testing data that can-not be effectively calculated for gas wells.The process of using artificial intelligence method to calculate recoverable reserves can be regarded as providing products,services and applications for recoverable reserves calculation by us-ing models,algorithms and computility on the basis of gas field big data,so as to apply the advantages of artificial intelligence in elimating data ambiguity,efficient coordination capability,strong learning capability and nonlinear capability to recoverable reserves calculation.Taking calculated recoverable reserves of gas wells with complete and accurate data as learning samples,calculation parameters are preliminarily selected by geological and dynamic study results of gas reservoirs,and final parameters are selected by grey correlation.The relationship between final parameters and recoverable reserves is established by artificial intelligence training,and the established relation-ship is applied to calculate recoverable reserves of other gas wells.Application in Y50 well block of Yan'an Gas Field shows single-well verification error of-1.88%～4.80%and cumulative multiple-well error of 1.13%.The prac-tice shows that recoverable reserves calculation by artificial intelligence method can meet the needs of engineering calculation,significantly improve calculation efficiency,save labor cost and reduce test cost,and also can calcu-late recoverable reserves of gas wells without test data or without incomplete data.关键词
致密气藏/可采储量/人工智能/BP神经网络Key words
tight gas reservoir/recoverable reserves/artificial intelligence/BP neural network分类
能源科技引用本文复制引用
米乃哲,乔向阳,李旭芬,吕远,许伟,谢小飞..应用人工智能方法计算致密气藏可采储量[J].大庆石油地质与开发,2025,44(3):70-76,7.基金项目
陕西省重点研发计划项目"致密气藏产水气井产量优化技术研究"(2021GY-167) （2021GY-167）
陕西延长石油(集团)有限责任公司"揭榜挂帅"项目"延气2井区提高采收率关键开发技术研究"(csy2021jbgs-A-05). （集团）
