基于工业大数据的FPSO生产流程自主诊断系统

打开文本图片集

中图分类号：TP277 文献标识码：A 文章编号：2096-4706（2025）16-0070-07

FPSO Production Process Autonomous Diagnosis System Based on Industrial Big Data

LIUDonghui，LIUXuesong，SUNHaifang，TONGYingli，DENGXin （CNOOC Energy Technologyand Services OilProduction Services Company，Tianjin 30o452，China）

Abstract： Aiming at the problems of dynamic anomaly of control loop and abnormal detection of process parameter fluctuation in the operation ofFloating Production Storage and Offloading （FPSO）unit，this paper studies and proposes an autonomous diagnosis system forFPSO production process.Firstly，the anomaly diagnosis module ofthecontrol loop is constructedbycorelationanalysis，K-meansclusteringand SupportVectorDataDescription（SVDD）.Secondly，theaboal diagnosis moduleofprocess parameters isconstructedbyrulesandrolling PrincipalComponentAnalysis （PCA）method.Fialy combinedwithhistoricaldataandexpertknowledgebase，afaulttreemodelisestablishedtoinferthecauseoftheabnormality online.Experiments show that the system can identifycomplex anomaliesand provide maintenance decision support througha visual interface，which effectively improves the accuracy ofFPSO autonomous diagnosis.

Keywords： data driven; online diagnosis; Machine Learning; control loop; process parameters

0 引言

海洋油气开发已成为国家战略，随着海洋业务扩大，开发有效的生产流程自主诊断系统成为提高生产安全性和效率的迫切需求。（剩余8933字）