基于两轮内检测数据挖掘的管道完整性评价技术  

Pipeline integrity evaluation technology based on two-round internal detection data mining

作　　者：刘国荣 LIU Guorong(No.4 Oil Production Plant of PetroChina Huabei Oilfield Company,Langfang,Hebei 065000,China)

机构地区：[1]中国石油华北油田分公司第四采油厂,河北廊坊065000

出　　处：《世界石油工业》2023年第5期62-68,共7页World Petroleum Industry

摘　　要：随着智慧油田和智慧管道的发展,深入整合和挖掘管道基础数据已成为大势所趋。针对两轮内检测数据对齐和剩余寿命计算中存在的问题,利用Kullback-Leibler(KL)散度实现分段管节各项分布特征的对齐,基于误差来源和正态分布,利用非活性缺陷点的特征,对内检测中存在的系统误差进行消除,选择90%分位数的腐蚀速率作为预测剩余寿命的基础,得到差异化的全线腐蚀速率,通过实例分析制定更加合理的维护响应计划,形成一套基于数据挖掘的管道完整性评价技术。结果表明,通过KL散度的数据对齐后,阀门、三通和绝缘接头的对齐比例为100%,其余特征的对齐比例均在90%以上;误差消除后,两轮内检测的非活性缺陷深度差值服从均值为0,标准差一定的正态分布;通过开挖验证证实方法的维修点大部分实际腐蚀深度大于30%,属于计划响应的范围,其余方法增加了不必要的维修点。研究结果可为管道完整性管理水平的提升提供实际参考。With the development of smart oil fields and smart pipelines,it has become a general trend to integrate and dig the basic data of pipelines.In view of the problems existing in the alignment of the data and the calculation of the remaining life of the two rounds of internal detection,the Kullback-Leibler(KL)divergence was used to realize the alignment of the distribution features of the subsection pipe section.Based on the source of error and the normal distribution,the characteristics of inactive defect points were used to eliminate the systematic errors existing in the internal detection.The corrosion rate of 90%quantile was selected as the basis for predicting the remaining life.Through case analysis,a more reasonable maintenance response plan was developed,and a set of pipeline integrity evaluation technology based on data mining was formed.The results show that the alignment ratio of valve,tee and insulation joint is 100%after the alignment of KL divergence data,and the alignment ratio of other features is above 90%.After the error is eliminated,the difference of the depth of inactive defects in the two rounds follows a normal distribution with a mean value of zero and a certain standard deviation.It is verified by excavation that the actual corro-sion depth of most maintenance points of the proposed method is greater than 30%,which belongs to the range of planned re-sponse.The other methods add unnecessary maintenance points.The research results can provide practical reference for the im-provement of pipeline integrity management level.

关 键 词：管道完整性 内检测 数据挖掘 数据对齐 正态分布 管道检测 

分 类 号：TE973.6[石油与天然气工程—石油机械设备]