基于大数据挖掘的热油管道闭环安全生产体系  

Closed-Loop Safety Production System of Hot Oil Pipeline Based on Big Data Mining

作　　者：于涛[1,2] 李传宪 刘丽君[3] 郭悠悠 YU Tao;LI Chuanxian;LIU Lijun;GUO Youyou(College of Pipeline and Civil Engineering,China University of Petroleum (East China),Qingdao,Shandong 266580,China;PipeChina Oil & Gas Control Center,Beijing 100007,China;Kunlun Shuzhi Technology Co.,Ltd.,Beijing 100007,China)

机构地区：[1]中国石油大学(华东)储运与建筑工程学院,山东青岛266580 [2]国家管网生产经营本部(油气调控中心),北京100007 [3]昆仑数智科技有限责任公司,北京100007

出　　处：《西安石油大学学报（自然科学版）》2021年第1期85-91,共7页Journal of Xi’an Shiyou University（Natural Science Edition）

基　　金：国家自然科学基金(51774311);中国博士后科学基金(2019TQ0354);青岛博士后研究人员应用研究项目(229319)。

摘　　要：长输加热原油管道运行工艺复杂,安全生产与优化相互矛盾,且存在管道沿线油温理论计算误差大,管壁结蜡评估难度大等问题。本研究利用管道实际生产数据,基于数据挖掘算法从热力和水力两个方面,构建了热油管道闭环安全生产体系。其中热力系统是利用BP神经网络,ARMA和Seq2Seq算法模型,建立稳态和非稳态油温预测模型,预测误差小于0.5℃。水力系统则是通过分析管道清管后一定时间内管道摩阻,利用MEA-BP神经网络建立管道标准摩阻预测模型,可对高含蜡热油管道管壁结蜡情况进行有效评估。建立摩阻数据库,利用高斯公式,对历史摩阻数据进行分析,设置历史摩阻数据的90%、95%为预警值和报警值,有效监控由于管道结蜡等引起的长周期摩阻变化。将研究成果应用于指导HY热油管道工艺调整,实现节能达92.4%。基于生产数据建立的油温预测模型,构建的热油管道闭环安全生产体系,具有很好的适用性,为未来管道智能化控制奠定了基础。The operation process of long-distance heating crude oil pipeline is complex,and the safety production and process optimization are contradictory.Moreover,the theoretical calculation error of oil temperature along the pipeline is large,and the evaluation of wax deposition on the pipe wall is difficult.Based on the actual data of the pipeline,a closed-loop safety production system of hot oil pipeline is constructed from thermal and hydraulic aspects using data mining algorithm.In thermal aspect,the steady and unsteady oil temperature prediction models are established using BP neural network,ARMA and Seq2Seq algorithm,and their oil temperature prediction error is less than 0.5℃.In hydraulic aspect,through the analysis of pipeline friction within a certain time after pigging,the standard prediction model of pipeline friction is established by using MEA-BP neural network,and the wax deposition on the wall of high waxy hot oil pipeline can be effectively evaluated using it.A pipeline friction database is established,the historical friction data of pipelines are analyzed using Gauss formula.90%and 95%of historical friction data are set as early warning value and alarm value separately to effectively monitor the long-term change of frictional resistance.The research result was applied to guide the process adjustment of HY hot oil pipeline,achieving energy saving of 92.4%.The oil temperature prediction model based on production data and the closed-loop safety production system of hot oil pipeline have good applicability,which lays the foundation for the intelligent control of pipeline in the future.

关 键 词：热油管道 数据挖掘 神经网络 摩阻 安全优化 

分 类 号：TE832[石油与天然气工程—油气储运工程]