油气井多相流虚拟计量研究现状与展望
Research status and prospect of virtual metering of multiphase flow in oil and gas wells
矫欣雨 1檀朝东 2朱永乐 3魏方方 4蔡岳 3檀庭庄 4张汇5
作者信息
- 1. 中国石油大学(北京)石油工程学院,北京 102200
- 2. 中国石油大学(北京)自动化系,北京 102200
- 3. 中国石油集团长城钻探工程有限公司,辽宁盘锦 124000
- 4. 安徽中控仪表有限公司,安徽池州 247200
- 5. 新疆油田公司数据公司,新疆克拉玛依 834000
- 折叠
摘要
文章阐述了时频分析、机理仿真、数据驱动、机理与数据驱动融合等技术在油气井多相流虚拟计量领域的研究现状与应用进展,讨论了油气井多相流虚拟计量的四种模型的优点和局限性,指出基于物理约束的油气井虚拟计量深度学习模型,能用较少的样本学习获得较好泛化能力的虚拟计量模型,计量精度高、计算速度快,成为油气井多相流虚拟计量技术的研究热点和未来发展趋势.
Abstract
This paper describes the research status and application progress of time-frequency analysis,mechanism simulation,data-driven,mechanism and data-driven fusion in the field of virtual metering of multiphase flow in oil and gas wells.this paper discusses the advantages and limitations of four models of virtual metering of multiphase flow in oil and gas wells,and points out that the depth learning model of virtual metering of oil and gas wells based on physical constraints can obtain a virtual metering model with better generalization ability with fewer samples.With high metering accuracy and fast calculation speed,it has become the research focus and future development trend of virtual metering technology for multi-phase flow in oil and gas wells.
关键词
虚拟计量/时频分析/物理约束/深度学习/灰盒模型/发展趋势Key words
virtual metering/time-frequency analysis/physical constraints/deep learning/grey-box model/development trend引用本文复制引用
出版年
2024
NETL
NSTL
万方数据