实时导向钻井任务型对话机器人的研发与应用  

Development and application of task-oriented dialogue robot for real-time geosteering drilling

作　　者：林昕 苑仁国 罗鹏 李文元 杨柳河 李琰 尹涛 王海涛 徐昱 李鹏飞 LIN Xin;YUAN Renguo;LUO Peng;LI Wenyuan;YANG Liuhe;LI Yan;YIN Tao;WANG Haitao;XU Yu;LI Pengfei(CNOOC EnerTech-Drilling&Production Co.,Tianjin 300452,China;Tianjin Branch of CNOOCCo.,Ltd.,Tianjin 300452,China)

机构地区：[1]中海油能源发展股份有限公司工程技术分公司 [2]中海石油(中国)有限公司天津分公司

出　　处：《天然气工业》2025年第3期132-139,共8页Natural Gas Industry

基　　金：中海油能源发展股份有限公司重大科研专项(课题)“钻井地质风险评价及实时导向技术”(编号:HFZDZX-GJ2020-01-05);中海油能源发展股份有限公司工程技术分公司科研项目“深湖相元素地层划分及岩性识别研究——以黄河口地区为例”(编号:GCJSXMHT-2420)。

摘　　要：实时导向钻井涉及专业面广且复杂,需要频繁的跨专业信息交互,目前仍处于严重依赖人员参与的低效模式。为助力油气勘探开发上游产能建设的提速增效,在已建成的中海油数字化生态平台基础上,以地质导向钻井智能算法为基础,基于Rasa开源框架,采用微服务API调用和模块化开发,研发了面向实时导向钻井的任务型对话机器人系统,实现了正钻井的钻井、录井、测井数据到导向信息的实时分析转化。研究结果表明:(1)对话机器人利用自动分析用户画像、快速读取海量知识库、多轮对话给出个性化答案等优势,在导向过程中针对实时工况频繁更新、目标半明确且需要引导等交互场景具有较好的应用效果;(2)对话机器人可以支持非关键井段自动跟井、关键井段导向决策支持等实时场景使用,已在渤海油田应用超过100井次;(3)对话机器人实现了人类语言和导向系统后台算法成果的自然交互,计算时长小于0.5 min,满足了导向钻井的米级响应要求。结论认为,实时导向钻井任务型对话机器人提高了随钻作业监测和质量控制过程中的决策响应时效,精简了作业流程,挖掘了正钻井的实时数据价值,有效提高了导向过程中各专业间一体化协同效率,对于探索适用于油气勘探开发的新型随钻跟踪管理模式具有重要参考价值。The professional fields involved in real-time geosteering drilling are extensive and complex,which makes frequent cross-disciplinary information interactions necessary.At present,however,it is still in a low-efficiency mode that relies heavily on human participation.To help speed up and improve the upstream productivity construction of oil and gas exploration and development,a task-oriented dialogue robot system for real-time geosteering drilling is researched and developed on the basis of CNOOC's existing digital ecology platform,intelligent algorithm of geosteering drilling and Rasa open-source framework,by means of microservice API call and modular development.Thus,the real-time analysis and transformation of drilling,mud logging,and wireline logging data to geosteering information of being drilled wells is realized.The following results are obtained.First,making use of the advantages such as automatically analyzing user profiles,quickly reading massive relevant knowledge bases and giving personalized answers through multi-round dialogues,the dialogue robot has a good application effect in interactive scenarios in the process of geosteering,such as frequent update of real-time working conditions,and semi-clear goal that needs guidance.Second,the dialogue robot can support real-time scenarios such as the automatic following in non-critical well sections and the geosteering decision making in critical well sections,and it has been applied more than 100 well times in Bohai Oilfield.Third,the dialogue robot realizes the natural interaction between human languages and backend calculation results of geosteering system,with the calculation time of less than 0.5 min,which meets the requirement of meter level response in geosteering drilling.In conclusion,the task-oriented dialogue robot for real-time geosteering drilling improves the decision response efficiency in the process of while drilling operation monitoring and quality control,simplifies the work flow,explores the value of real-time data of being drilled wells

关 键 词：实时导向钻井 对话机器人 算法 渤海油田 随钻跟踪 Rasa 

分 类 号：TE243[石油与天然气工程—油气井工程]