基于循环神经网络的抽油杆柱寿命预测新方法  

Prediction Model of Rod String Anomaly in Pumping Well

作　　者：邴绍强 王振[2] 段鸿杰 张旭跃 邢雷 BIN Shao-qiang;WANG Zhen;DUAN Hong-jie;ZHANG Xu-yue;XING Lei(Shengli Oilfield Branch Information Management Center,Dongying 257000,China;Shengli Oilfield Public Service Center Marketing Development Department,Dongying 257000,China;School of Petroleum Engineering,Changzhou University,Changzhou 213016,China;Shengli Oilfield Luming Oil and Gas Exploration and Development Co.,Ltd.,Dongying 257000,China)

机构地区：[1]胜利油田分公司信息化管理中心,山东东营257000 [2]胜利油田鲁明油气勘探开发有限公司,山东东营257000 [3]常州大学石油工程学院,江苏常州213016 [4]胜利油田公共事业服务中心市场开发部,山东东营257000

出　　处：《电脑知识与技术》2019年第12Z期178-182,187,共6页Computer Knowledge and Technology

摘　　要：当前抽油杆柱易发生异常甚至发生抽油杆脱断,因此预测抽油杆柱的剩余寿命变得越来越重要。而当前传统的抽油杆柱剩余寿命预测方法效率低,准确度差,计算模型较为复杂。基于此,本文借助循环神经网络方法,训练建立一套油井杆柱寿命预测的神经网络新方法。结果表明:对于抽油杆柱可以通过循环神经网络(RNN)准确预测抽油杆柱剩余寿命。At present,the sucker rod column is prone to abnormality or even the sucker rod is broken,so it is more and more important to predict the remaining life of the sucker rod column.However,the current traditional method for predicting the remaining life of the sucker rod column is low in efficiency and poor in accuracy,and the calculation model is complicated.Based on this,this paper uses the cyclic neural network method to train a new neural network method for life prediction of oil well rods.The results show that for the suck⁃er rod column,the residual life of the sucker rod column can be accurately predicted by the circulating neural network(RNN).

关 键 词：抽油杆柱 寿命预测 循环神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]