机理仿真与数据驱动融合的电泵举升故障诊断预警理论研究进展

Research progress on fault diagnosis and early warning theory of electric submersible pump lifting based on mechanism simulation and data driven fusion

檀朝东 黄新春 王松 张光一 付军 杜广浩

TAN Chaodong HUANG Xinchun WANG Song ZHANG Guangyi FU Jun DU Guanghao

中国石油大学(北京)油气资源与探测国家重点实验室 中国石油大学(北京)石油工程学院 中国石油大学(北京)人工智能学院 中海油能源发展股份有限公司工程技术分公司

State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum(Beijing), Beijing 102249, China College of Petroleum Engineering, China University of Petroleum(Beijing), Beijing 102249, China College of AI, China University of Petroleum(Beijing), Beijing 102249, China CNOOC EnerTech-Drilling & Production Co., Tianjin 300452, China

在复杂、不确定、非结构化的油藏环境中，潜油电泵举升井的工况诊断预警优化决策成为智能油井研究的热点与难点问题之一。阐述了机理仿真、数据驱动、专家系统、机理仿真与数据驱动融合等技术在电泵举升故障诊断预警领域的研究与应用进展，探讨并展望未来电泵举升故障诊断预警技术的发展方向与重点。综述研究表明：建立客观准确的反映电泵运行动态工况特性的机理模型和快速精确求解极具挑战性，基于数据驱动诊断模型未必能完整描述电泵生产系统工况的真实映射关系，电泵生产故障诊断推理机理及专家系统的自学习能力有待加强。提出基于机理仿真与数据分析方法相融合的数字孪生技术必然成为研究复杂电泵抽油生产系统故障诊断预警的核心技术。

In the complex, uncertain and unstructured reservoir environment, the working condition diagnosis, alert and optimization decision of electric submersible pump (ESP) well has become one of the hot and difficult problems in the research of intelligent oil well. This paper describes the research progress and application of mechanism simulation, data-driven, expert system, mechanism simulation and data-driven fusion technology in the field of ESP fault diagnosis and alert, discusses and prospects the development direction and focus of ESP fault diagnosis and alert technology in the future. Literature research shows that, it is a challenging work to establish a mechanism model and fast and accurate solution of the model which can accurately reflect the dynamic characteristics of ESP; the diagnosis model based on big data may not be able to completely describe the real mapping relationship of the working conditions of ESP production system; the reasoning mechanism of ESP production fault diagnosis and the self-learning ability of the expert system need to be strengthened. It is pointed out that the digital twin technology based on mechanism simulation and data analysis method must become the core technology of fault diagnosis and early warning for complex ESP production system.

10.13639/j.odpt.2021.04.011