基于Hessian正则化支持向量机的多视角协同识别抽油机井工况方法

doi:10.7623/syxb201812011

石油学报 ›› 2018, Vol. 39 ›› Issue (12): 1429-1436.DOI: 10.7623/syxb201812011

基于Hessian正则化支持向量机的多视角协同识别抽油机井工况方法

周斌, 王延江, 刘伟锋, 刘宝弟

中国石油大学信息与控制工程学院山东青岛 266580

收稿日期:2018-02-01 修回日期:2018-09-28 出版日期:2018-12-25 发布日期:2018-12-29
通讯作者: 王延江,男,1966年9月生,1986年获华东石油学院学士学位,2001年获北方交通大学信号与信息处理专业博士学位,现为中国石油大学(华东)信息与控制工程学院教授、博士生导师,主要从事模式识别与智能信息处理及其在石油勘探、开发中的应用研究。Email:yjwang@upc.edu.cn
作者简介:周斌,女,1970年1月生,1998年获石油大学(华东)学士学位,现为中国石油大学(华东)信息与控制工程学院博士生,主要从事故障诊断与智能信息处理及其在石油开采中的应用研究。Email:freetzb@163.com
基金资助:
国家自然科学基金项目（No.61671480）资助。

A working condition recognition method of sucker-rod pumping wells based on Hessian-regularized SVM and multi-view co-training algorithm

Zhou Bin, Wang Yanjiang, Liu Weifeng, Liu Baodi

College of Information and Control Engineering, China University of Petroleum, Shandong Qingdao 266580, China

Received:2018-02-01 Revised:2018-09-28 Online:2018-12-25 Published:2018-12-29

摘要/Abstract

摘要：

为了更好地了解抽油机井工作状况以提高抽油机井生产工况识别率，目前在抽油机井采油生产过程中采集了大量数据。传统的工况识别方法，主要是利用示功图或电参数单独进行训练建立学习模型，参数之间缺乏关联，影响了识别效果；此外，常用的传统识别方法，如基于支持向量机（SVM）学习的方法，需要对所有的采集样本进行类别标注，耗费大量的人力和物力，影响了工程应用。针对大数据下抽油机井生产特点，为实现在仅有少量已标注工况数据下能同时利用大量未知工况数据信息，且有效利用示功图和电参数两种测量参数，进一步提高抽油机井工况识别的精准率和实用性，提出一种基于Hessian正则化支持向量机（Hessian正则化SVM）的多视角协同识别抽油机井工况方法。通过分析目前工况识别研究中存在的局限性，结合先验知识和专家经验，选择实测地面示功图和实测电功率信号作为特征视角并进行特征提取，然后利用Hessian正则化SVM多视角协同训练算法建立抽油机井工况识别模型并进行分类识别。应用该方法对胜利油田X区块60口抽油机井的11种典型工况进行识别。以SVM方法为基准，该方法识别效果比基于实测地面示功图、实测电功率及传统特征连接多源识别方法分别提高了约3.2%、4.3%和7.4%，而在少量工况样本下该方法识别效果更优，从而验证了该方法的有效性。

关键词: 抽油机井, 工况识别, 多视角学习, 协同训练, 支持向量机, Hessian正则化

Abstract:

At p resent, massive data has been collected during oil production from sucker-rod pumping wells, aiming to better understand the working conditions and improve the recognition rate of working conditions of sucker-rod pumping wells. However, the traditional recognition methods mainly use dynamometer cards or electrical parameters to perform training, respectively, and then create the learning models. There is no correlation between parameters, thus influencing the recognition effect. In addition, traditional recognition methods, such as support vector machine (SVM)-based learning methods, require to label all samples according to their categories, thus taking a lot of manpower and material resources, and exerting an impact on the engineering application. Aiming at the production characteristics of the sucker-rod pumping system in the context of big data, this paper proposes a novel working condition recognition method based on Hessian-regularized SVM and multi-view co-training algorithm; depending on a small amount of labeled data for working conditions, this can further improve the recognition accuracy and practicality of sucker-rod pumping wells by effectively utilizing massive unknown working condition data and two kinds of measured parameters above mentioned. Through analyzing the limitations in the existing recognition research, based on the prior information and empirical knowledge, this study has completed feature extraction from perspective of the measured ground dynamometer cards and electrical power signals, respectively. Then a working condition recognition model of sucker-rod pumping wells is established using multi-view co-training algorithm based on Hessian-regularized SVM. This method is applied to recognize eleven typical working conditions from sixty sucker-rod pumping wells in a certain block of Shengli Oilfield. Based on SVM method, the recognition effect of this method is higher than that of those methods based on the measured ground dynamometer card, measured electrical power and traditional multi-source feature connection method by 3.2%, 4.3% and 7.4%, respectively. The performance is better even in the cases of fewer working condition samples, thus validates the effectiveness of this method.

Key words: sucker-rod pumping wells, working condition recognition, multi-view learning, co-training, support vector machine, Hessian regularization

中图分类号:

TE933.1

周斌, 王延江, 刘伟锋, 刘宝弟. 基于Hessian正则化支持向量机的多视角协同识别抽油机井工况方法[J]. 石油学报, 2018, 39(12): 1429-1436.

Zhou Bin, Wang Yanjiang, Liu Weifeng, Liu Baodi. A working condition recognition method of sucker-rod pumping wells based on Hessian-regularized SVM and multi-view co-training algorithm[J]. Acta Petrolei Sinica, 2018, 39(12): 1429-1436.

参考文献

[1] 刘卓,罗明良,陆雪皎,等. 有杆抽油系统工况诊断的数学模型研究[J].数学的实践与认识,2013,43(15):41-48.
LIU Zhuo,LUO Mingliang,LU Xuejiao,et al.Study on the working condition diagnosis method for suck rod pumping system based on wavelet moment features[J].Mathematics in Practice and Theory,2013,43(15):41-48.
[2] ZHANG Shirong,TANG Yuling.Indirect measurement of dynamometer card of pumping unit[C]//Proceedings of the 7th World Congress on Intelligent Control and Automation.Chongqing,China:IEEE,2008:4952-4955.
[3] WU Wei,SUN Wenli,WEI Hangxin.A fault diagnosis of suck rod pumping system based on wavelet packet and RBF network[J].Advanced Materials Research,2011,189-193:2665-2669.
[4] LI Kun,GAO Xianwen,TIAN Zhongda,et al.Using the curve moment and the PSO-SVM method to diagnose downhole conditions of a sucker rod pumping unit[J].Petroleum Science,2013,10(1):73-80.
[5] 李金诺,龚仁彬,李群,等.一种抽油机示功图数据无损压缩存储方法[J].石油学报,2016,37(2):266-272.
LI Jinnuo,GONG Renbin,LI Qun,et al.A lossless data compression and storage method of pump dynamometer[J].Acta Petrolei Sinica,2016,37(2):266-272.
[6] 梁华.有杆抽油系统故障递阶诊断的故障识别研究[J].西南石油大学学报:自然科学版,2015,37(1):165-171.
LIANG Hua.Hierarchical fault diagnosis of rod pumping system based on fault distinguishing[J].Journal of Southwest Petroleum University:Science & Technology Edition, 2015,37(1):165-171.
[7] LI Kun,GAO Xianwen,ZHOU Haibo,et al.Fault diagnosis for down-hole conditions of sucker rod pumping systems based on the FBH-SC method[J].Petroleum Science,2015,12(1):135-147.
[8] 陈黎兴.基于电功率数据的抽油机故障诊断方法研究[D].青岛:中国石油大学,2016.
CHEN Lixing.Electrical power data based fault diagnosis of oil pumping units[D].Qingdao:China University of Petroleum,2016.
[9] 孙振华.游梁式抽油机采油系统实时评价方法研究[D].青岛:中国石油大学,2011.
SUN Zhenhua.The real-time evaluation method for the beam-pump recovery system[D].Qingdao:China University of Petroleum,2011.
[10] LIU Shuping,RAGHAVENDRA C S,LIU Yintao,et al.Automatic early fault detection for rod pump systems[R].SPE 146038,2011.
[11] 王凯.基于产生式规则系统的抽油泵故障诊断[J].石油勘探与开发,2010,37(1):116-120.
WANG Kai.Fault diagnosis of rod-pumping unit based on production rules system[J].Petroleum Exploration and Development,2010,37(1):116-120.
[12] 胡广杰,易斌,田宝库.抽油机井实测示功图泵况诊断分析[M].北京:石油工业出版社,2008:1-45.
HU Guangjie,YI Bin,TIAN Baoku.Indicator diagram pump condition diagnosis analysis of pumping unit well[M].Beijing:Petroleum Industry Press,2008:1-45.
[13] 王凯,刘宏昭,熊俊,等.基于改进的超球支持向量机的有杆抽油泵故障诊断研究[J].机械科学与技术,2011,30(1):133-137.
WANG Kai,LIU Hongzhao,XIONG Jun,et al.Fault diagnosis of a rod-pumping unit based on improved hyper sphere vector machines[J].Mechanical Science and Technology for Aerospace Engineering,2011,30(1):133-137.
[14] 李子丰.油气井杆管柱力学研究进展与争论[J].石油学报,2016,37(4):531-556.
LI Zifeng.Research advances and debates on tubular mechanics in oil and gas wells[J]. Acta Petrolei Sinica,2016,37(4):531-556.
[15] 宋玉杰,彭惠芬,夏晔.再制造抽油杆纵向共振模糊可靠性分析[J].石油学报,2017,38(4):468-474.
SONG Yujie,PENG Huifen,XIA Ye.A fuzzy reliability analysis on the longitudinal vibration of remanufacturing sucker rod[J].Acta Petrolei Sinica,2017,38(4):468-474.
[16] 李杰训,贾贺坤,宋扬,等.油井产量计量技术现状与发展趋势[J].石油学报,2017,38(12):1434-1440.
LI Jiexun,JIA Hekun,SONG Yang,et al.Current technical status and development trend of oil well production measurement[J].Acta Petrolei Sinica,2017,38(12):1434-1440.
[17] LIU Weifeng,LI Yang,TAO Dapeng,et al.A general framework for co-training and its applications[J].Neurocomputing,2015,167:112-121.
[18] IOSIFIDIS A,TEFAS A,PITAS I.Multi-view action recognition based on action volumes,fuzzy distances and cluster discriminant analysis[J].Signal Processing,2013,93(6):1445-1457.
[19] LIU Weifeng,LIU Hongli,TAO Dapeng,et al.Multiview Hessian regularized logistic regression for action recognition[J].Signal Processing,2015,110:101-107.
[20] LIU Weifeng,LI Yang,LIN Xu,et al.Hessian-regularized co-training for social activity recognition[J].PLoS One,2014,9(9):e108474.
[21] LIU Weifeng,TAO Dacheng.Multiview hessian regularization for image annotation[J].IEEE Transactions on Image Processing,2013,22(7):2676-2687.
[22] TAO Dapeng,JIN Lianwen,LIU Weifeng,et al.Hessian regularized support vector machines for mobile image annotation on the cloud[J].IEEE Transactions on Multimedia, 2013,15(4):833-844.
[23] BELKIN M,NIYOGI P,SINDHWANI V.Manifold regularization:a geometric framework for learning from labeled and unlabeled examples[J].The Journal of Machine Learning Research,2006,7:2399-2434.
[24] ZHANG Minling,ZHOU Zhihua.CoTrade:confident co-training with data editing[J].IEEE Transactions on Systems,Man,and Cybernetics,Part B (Cybernetics),2011,41(6):1612-1626.
[25] 周志华.机器学习[M].北京:清华大学出版社,2016:29-35.
ZHOU Zhihua.Machine learning[M].Beijing:Tsinghua University Press,2016:29-35.

基于Hessian正则化支持向量机的多视角协同识别抽油机井工况方法

A working condition recognition method of sucker-rod pumping wells based on Hessian-regularized SVM and multi-view co-training algorithm

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	王宏博, 董世民, 甘庆明, 辛红, 朱葛. 低产抽油机井动态参数仿真模型与提高系统效率的途径[J]. 石油学报, 2018, 39(11): 1299-1307.
[2]	邢海燕, 党永斌, 王犇, 葛桦, 王尊策. 基于K-近邻隶属度模糊支持向量机的再造抽油杆损伤等级磁记忆定量识别[J]. 石油学报, 2015, 36(11): 1427-1432,1456.
[3]	朱怡翔石广仁. 火山岩岩性的支持向量机识别[J]. 石油学报, 2013, 34(2): 312-322.
[4]	于德亮齐维贵邓盛川张永明张凤武王新民. 基于支持向量机的潜油柱塞泵沉没度预测[J]. 石油学报, 2011, 32(3): 534-538.
[5]	易方李著信吕宏庆马维平. 基于模糊核支持向量机的管道磁记忆检测缺陷识别[J]. 石油学报, 2010, 31(5): 863-866.
[6]	王明达张来斌梁伟陈志刚. 基于独立分量分析和支持向量机的管道泄漏识别方法[J]. 石油学报, 2010, 31(4): 659-663.
[7]	董世民张喜顺吴长杰郭吉民. 抽油机井区块整体节能参数仿真优化方法[J]. 石油学报, 2010, 31(3): 475-479.
[8]	余东亮, 张来斌, 梁伟, 叶迎春, 王朝晖. 长输管道信号降噪及工况识别方法研究[J]. 石油学报, 2009, 30(6): 937-941.
[9]	李洪奇, 郭海峰, 郭海敏, 孟照旭, 谭锋奇, 张军. 复杂储层测井评价数据挖掘方法研究[J]. 石油学报, 2009, 30(4): 542-549.
[10]	马海, 王延江, 魏茂安, 胡睿. 地层可钻性级值预测新方法[J]. 石油学报, 2008, 29(5): 761-765.
[11]	石广仁. 支持向量机在多地质因素分析中的应用[J]. 石油学报, 2008, 29(2): 195-198.
[12]	韩修廷, 王德喜, 王研, 周录方, 王秀玲. 利用低摩阻柱塞抽油泵提高泵效及防偏磨技术的应用[J]. 石油学报, 2007, 28(4): 138-141.
[13]	闫铁, 毕雪亮, 王长江. 基于支持向量机和聚类分析理论的钻具失效分析方法[J]. 石油学报, 2007, 28(3): 135-140.
[14]	张辉, 高德利. 钻头下部未钻开地层的可钻性预测新方法[J]. 石油学报, 2006, 27(1): 97-100.
[15]	姚春东. 提高抽油机井系统效率的计算机仿真分析[J]. 石油学报, 2005, 26(4): 106-110114.