基于均值滤波和BP神经网络的功图量油技术
|
摘要
功图量油技术是通过对油井示功图数据的实时采集和处理分析来计算油井产液量,从而实现对产液量的远程自动计量。针对目前功图量油技术存在的不足,提出了一些改进算法。其中,运用均值滤波法对功图数据进行处理,降低了曲率变化量最大点的识别难度,提高了有效冲程求取的准确性;利用Matlab神经网络工具箱,对BP神经网络模型进行训练,建立油井漏失量BP神经网络模型,从而实现油井漏失量的准确求取。应用结果表明,改进的功图量油算法有效提高了油井示功图产液量计量的准确性和可靠性,最大相对误差6.5%,最小相对误差0.7%,平均相对误差小于5%,能够满足油田的生产计量要求,对功图量油技术的推广和油田信息化建设具有重要意义。
Through real-time acquisition of oil well indicator diagram data and process, analysis and calculation of the oil well produced liquid rate,the indicator diagram oil metering technology can realize the remote automatic metering of oil well produced liquid rate. In view of the shortcomings of the existing technology, some improved algorithms are put forward. Among them, the average filtering method is applied to process the data of indicator diagrams,which can reduce the recognition difficulty of the maximum curvature variation point and improve the accuracy of the effective stroke calculation;the Matlab neural network toolbox is used to conduct simulated training of the BP neural network mode, to establishes the BP neural network mode for oil well leakage, which can realize the accurate calculation of oil well leakage. Application results show that the improved indicator diagram oil metering algorithm can effectively improve the accuracy and reliability of oil well produced liquid rate. Its maximum relative error is 6.5%,its minimum relative error is 0.7%,and its average relative error is less than5%. The improved algorithm meets the requirements of oilfield production metering, and is of great significance for the promotion of the indicator diagram oil metering technology and oilfield informatization construction.
Through real-time acquisition of oil well indicator diagram data and process, analysis and calculation of the oil well produced liquid rate,the indicator diagram oil metering technology can realize the remote automatic metering of oil well produced liquid rate. In view of the shortcomings of the existing technology, some improved algorithms are put forward. Among them, the average filtering method is applied to process the data of indicator diagrams,which can reduce the recognition difficulty of the maximum curvature variation point and improve the accuracy of the effective stroke calculation;the Matlab neural network toolbox is used to conduct simulated training of the BP neural network mode, to establishes the BP neural network mode for oil well leakage, which can realize the accurate calculation of oil well leakage. Application results show that the improved indicator diagram oil metering algorithm can effectively improve the accuracy and reliability of oil well produced liquid rate. Its maximum relative error is 6.5%,its minimum relative error is 0.7%,and its average relative error is less than5%. The improved algorithm meets the requirements of oilfield production metering, and is of great significance for the promotion of the indicator diagram oil metering technology and oilfield informatization construction.
引文
[1]高银中,雷小强,杨江天.基于泵示功图的抽油机井井口产量计算[J].石油矿场机械,2008,37(5):88-92.
[2]钟张起,侯读杰,马尚俊,等.载荷位移法在功图量油技术中的应用[J].复杂油气藏,2014,7(4):69-72.
[3]金钟辉,李明江,毕振飞,等.功图量液技术在数字化条件下的应用研究[J].自动化博览,2017(10):92-97.
[4]张乃禄,赵岐,张钰哲,等.示功图法计算油井产液量的影响因素[J].西安石油大学学报(自然科学版),2011,26(4):53-55.
[5]XU P,XU S,YIN H.Application of self-organizing competitive neural network in fault diagnosis of suck rod pumping system[J].Journal of Petroleum Science&Engineering,2007,58(1):43-48.
[6]杨瑞,黄伟,辛宏,等.功图法油井计量技术在长庆油田的应用[J].油气田地面工程,2010,29(2):55-57.
[7]史明义,项勇,邹晓燕,等.功图法计量技术在大港油田的应用[J].石油规划设计,2006,17(3):9-12.
[8]唐庆,王新红,孙福山,等.抽油机井功图法量油技术先导试验[J].油气井测试,2006,15(6):64-65.
[9]郝世彦,樊平天,张新涛.油井远程监控与计量技术在延长油田的应用[J].石油钻探技术,2010,38(2):84-87.
[10]檀朝东,史明义,易南华,等.油井远程在线计量优化分析系统技术在大港油田的应用[J].中国石油和化工,2007(20):43-48.
[11]张龙飞.基于示功图的有杆泵抽油机井产液量计算方法的研究与实现[D].沈阳:东北大学,2014:27-28.
[12]李小兵,岑康,廖开贵,等.油井功图法自动计量与监测技术研究[J].石油化工自动化,2008,44(2):65-67.
[13]岳广韬.抽油机井地面示功图量油技术研究[D].东营:中国石油大学(华东),2011:34-37.
[14]管桐.功图法计算抽油机井产液量的研究[D].荆州:长江大学,2014:39-40.
[15]王振军.深井泵油井漏失示功图计算法的研究与应用[J].特种油气藏,2007,14(增刊1):166-168.
[16]郑娟.功图法在油田单井计量系统中的应用与研究[D].西安:西安石油大学,2015:28.
[17]黄兴鸿,齐玉梅,黄艳萍.抽油机井功图法计量中计量结果误差大的解决方法[J].中国化工贸易,2017,9(17):82.
[18]王新华,黄建林,敖科,等.BP神经网络在井组配注中的应用[J].钻采工艺,2006,29(2):112-113.
[19]李虎.基于改进BP神经网络的油井流入动态研究[J].复杂油气藏,2011,4(3):71-75.
[20]陈晓天,张顺颐,田婷婷.基于BP神经网络的IP网络流量预测[J].南京邮电大学学报(自然科学版),2010,30(2):16-21.
[21]李欢,曹立恒,杨娟,等.基于matlab的改进BP神经网络在油井产液量预测中的应用[J].低渗透油气田,2015(1):103-107.
[2]钟张起,侯读杰,马尚俊,等.载荷位移法在功图量油技术中的应用[J].复杂油气藏,2014,7(4):69-72.
[3]金钟辉,李明江,毕振飞,等.功图量液技术在数字化条件下的应用研究[J].自动化博览,2017(10):92-97.
[4]张乃禄,赵岐,张钰哲,等.示功图法计算油井产液量的影响因素[J].西安石油大学学报(自然科学版),2011,26(4):53-55.
[5]XU P,XU S,YIN H.Application of self-organizing competitive neural network in fault diagnosis of suck rod pumping system[J].Journal of Petroleum Science&Engineering,2007,58(1):43-48.
[6]杨瑞,黄伟,辛宏,等.功图法油井计量技术在长庆油田的应用[J].油气田地面工程,2010,29(2):55-57.
[7]史明义,项勇,邹晓燕,等.功图法计量技术在大港油田的应用[J].石油规划设计,2006,17(3):9-12.
[8]唐庆,王新红,孙福山,等.抽油机井功图法量油技术先导试验[J].油气井测试,2006,15(6):64-65.
[9]郝世彦,樊平天,张新涛.油井远程监控与计量技术在延长油田的应用[J].石油钻探技术,2010,38(2):84-87.
[10]檀朝东,史明义,易南华,等.油井远程在线计量优化分析系统技术在大港油田的应用[J].中国石油和化工,2007(20):43-48.
[11]张龙飞.基于示功图的有杆泵抽油机井产液量计算方法的研究与实现[D].沈阳:东北大学,2014:27-28.
[12]李小兵,岑康,廖开贵,等.油井功图法自动计量与监测技术研究[J].石油化工自动化,2008,44(2):65-67.
[13]岳广韬.抽油机井地面示功图量油技术研究[D].东营:中国石油大学(华东),2011:34-37.
[14]管桐.功图法计算抽油机井产液量的研究[D].荆州:长江大学,2014:39-40.
[15]王振军.深井泵油井漏失示功图计算法的研究与应用[J].特种油气藏,2007,14(增刊1):166-168.
[16]郑娟.功图法在油田单井计量系统中的应用与研究[D].西安:西安石油大学,2015:28.
[17]黄兴鸿,齐玉梅,黄艳萍.抽油机井功图法计量中计量结果误差大的解决方法[J].中国化工贸易,2017,9(17):82.
[18]王新华,黄建林,敖科,等.BP神经网络在井组配注中的应用[J].钻采工艺,2006,29(2):112-113.
[19]李虎.基于改进BP神经网络的油井流入动态研究[J].复杂油气藏,2011,4(3):71-75.
[20]陈晓天,张顺颐,田婷婷.基于BP神经网络的IP网络流量预测[J].南京邮电大学学报(自然科学版),2010,30(2):16-21.
[21]李欢,曹立恒,杨娟,等.基于matlab的改进BP神经网络在油井产液量预测中的应用[J].低渗透油气田,2015(1):103-107.