油水界面测量过程中自适应阈值聚类优化算法
|
摘要
精确检测原油储罐内油水界面及液位高度是保证净油外输含水率控制精度及联合站盘库系统计量精度的前提条件,是石油化工过程系统工程中的重要环节。鉴于油水界面测量过程中传统分类统计算法和经典K-means聚类算法存在依赖人工选取典型值和初始聚类质心、计算结果不确定性以及精度难于保证等问题,本文提出了一种改进的K-means自适应阈值聚类优化算法。该算法能自动获取最优初始阈值,并改进了油水界面测量传统分类统计算法和经典K-means聚类算法的思想,可实现最优数据分类。首先采用自适应阈值查找算法自动查找油水界面最优初始阈值,其次采用改进K-means聚类优化算法对油水界面数据进行最优划分,最后根据最优化聚类结果计算油水界面及液位高度。实验结果表明:相对于油水界面测量的传统分类统计算法和经典K-means聚类算法,该算法无需人工选值、能够完全保证计算结果的准确性,且比经典K-means聚类算法和其他改进K-means聚类算法所需的迭代次数更少、运行时间更短。
Accurately detecting the oil-water interface and the liquid level in a crude oil tank is very important for petrochemical process system engineering related to the accuracy of the net oil moisture control and that of combination station check system metering.In order to solve the problems that the conventional classification algorithm needs to select typical values manually and the traditional K-means clustering algorithm also needs to select initial clustering centers manually,and the results might not be accurate and stable,this paper proposed an adaptive threshold algorithm for oil-water interface detection,which is based on improved K-means clustering optimization algorithm.The proposed algorithm can calculate the best typical values automatically,and improves the conventional classification algorithm and the traditional K-means clustering algorithm.The proposed algorithm works as follows:first,find the typical values automatically by using an adaptive threshold searching algorithm,then search the optimal points by using the improved K-means clustering optimization algorithm,and finally calculate the oil water interface and the liquid level according to the optimal partition.Experimental results showed that,compared with the conventional classification algorithm and the traditional K-means clustering calculation algorithm,the proposed adaptive threshold clustering algorithm does not need artificial selection,and can guarantee the results to be correct.Moreover,the number of iterations and running time needed by our proposed algorithm are smaller than those by the traditional and other improved K-means clustering calculation algorithms.
Accurately detecting the oil-water interface and the liquid level in a crude oil tank is very important for petrochemical process system engineering related to the accuracy of the net oil moisture control and that of combination station check system metering.In order to solve the problems that the conventional classification algorithm needs to select typical values manually and the traditional K-means clustering algorithm also needs to select initial clustering centers manually,and the results might not be accurate and stable,this paper proposed an adaptive threshold algorithm for oil-water interface detection,which is based on improved K-means clustering optimization algorithm.The proposed algorithm can calculate the best typical values automatically,and improves the conventional classification algorithm and the traditional K-means clustering algorithm.The proposed algorithm works as follows:first,find the typical values automatically by using an adaptive threshold searching algorithm,then search the optimal points by using the improved K-means clustering optimization algorithm,and finally calculate the oil water interface and the liquid level according to the optimal partition.Experimental results showed that,compared with the conventional classification algorithm and the traditional K-means clustering calculation algorithm,the proposed adaptive threshold clustering algorithm does not need artificial selection,and can guarantee the results to be correct.Moreover,the number of iterations and running time needed by our proposed algorithm are smaller than those by the traditional and other improved K-means clustering calculation algorithms.
引文
[1]吴应湘,许晶禹.油水分离技术[J].力学进展, 2015, 45(6):179-216.WU Yingxiang, XU Jingyu.Oil and water separation technology[J].Advances in Mechanics, 2015, 45(6):179-216.
[2]陈祥光,薛锦诚,傅若农.储罐原油含水率在线测量及动态油量计量技术的研究[J].仪器仪表学报, 2000, 21(6):569-573.CHEN Xiangguang, XUE Jincheng, FU Ruonong. Study oncomputation technique in real time to dynamic oil quantity and measurement in online to water content ratio of crude oil tanks[J].Chinese Journal of Scientific Instrument, 2000, 21(6):569-573.
[3]曲艺,陈祥光.原油储罐油量动态计量技术研究[J].仪器仪表学报, 2005, 26(10):998-1001, 1010.QU Yi, WU Yingxiang. A research on the measurement ofdynamic quantity of crude oil in storage tank[J]. Chinese Journalof Scientific Instrument, 2005, 26(10):998-1001, 1010.
[4]刘春涛,陈祥光,刘志成.多传感器信息融合技术在动态油量计量中的应用[J].仪器仪表学报, 2005, 26(8):43-46.LIU Chuntao, CHEN Xiangguang, LIU Zhicheng.Application ofmulti-sensor information fusion in calculating dynamic oilquantity[J]. Chinese Journal of Scientific Instrument, 2005, 26(8):43-46.
[5]任磊,陈祥光,刘春涛.原油储罐油水界面测量方法的研究及应用[J].仪器仪表学报, 2008, 29(10):2186-2191.REN Lei, CHEN Xiangguang, LIU Chuntao. Research andapplication of measurement method for oil-water interface incrude oil tank[J]. Chinese Journal of Scientific Instrument, 2008,29(10):2186-2191.
[6]王云松,付玉红,宋艳慧.储油罐多相界面液位测控系统的研究[J].仪器仪表学报, 2003, 24(s1):285-286.WANG Yunsong, FU Yuhong, SONG Yanhui. Research onmonitor and control system of diversified liquid levels for storagetank[J]. Chinese Journal of Scientific Instrument, 2003, 24(s1):285-286.
[7]李一博,靳世久,孙墨杰,等.基于磁致伸缩效应的液体密度在线测量方法[J].天津大学学报, 2004, 37(4):368-372.LI Yibo, JIN Shijiu, SUN Mojie, et al. Liquid density on-linemeasuring based on magnetostriction effect[J]. Journal of TianjinUniversity, 2004, 37(4):368-372.
[8]彭勇,张映辉.光纤油水界面监控仪研制与应用[J].仪器仪表学报, 2005, 26(8):857-859.PENG Yong, ZHANG Yinghui. Applying and designing thecontroller of optical fiber oil interface[J]. Chinese Journal ofScientific Instrument, 2005, 26(8):857-859.
[9]马世海,吴珂,李晓薇.油-水两相界面测量技术综述[J].化工自动化及仪表, 2016, 43(12):1233-1238.MA Shihai, WU Ke, LI Xiaowei. Reviews of technologies for oil-water interface measurement[J]. Control and Instruments inChemical Industry, 2016, 43(12):1233-1238.
[10]任喜伟,任工昌,杨帆.电磁场式油水界面测量分析及数据优化方法[J].化工自动化及仪表, 2012, 39(7):858-861.REN Xiwei, REN Gongchang, YANG Fan. Oil-water interfaceanalysis and data optimization methods of electromagnetic field-typed measuring device[J]. Control and Instruments in ChemicalIndustry, 2012, 39(7):858-861.
[11]苏鑫,吴迎亚,裴华健,等.大数据技术在过程工业中的应用研究进展[J].化工进展, 2016, 35(6):1652-1659.SU Xin, WU Yingya, PEI Huajian, et al. Recent development ofthe application of big data technology in process industries[J].Chemical Industry and Engineering Progress, 2016, 35(6):1652-1659.
[12]刘远红,习玉阳,吴华远,等. K-means聚类算法在油水界面检测中的应用[J].化工自动化及仪表, 2016, 41(1):19-21.LIU Yanhong, XI Yuyang, WU Huayuan, et al. Application of K-means clustering algorithm in oil-water interface detection[J].Control and Instruments in Chemical Industry, 2016, 41(1):19-21.
[13] BAGIROY A M, UGON J, Webb D. Fast modified global K-means algorithm for incremental cluster construction[J]. Pattern Recognition, 2011, 44(4):866-876.
[14] TZORTZIS G, LIKAS A, Tzortzis G. The minmax K-meansclustering algorithm[J]. Pattern Recognition, 2014, 47(7):2505-2516.
[15] AMORIM R C D, MAKARENKOV V. Applying subclustering andLpdistance in weighted K-means with distributed centroids[J].Neurocomputing, 2016, 173(3):700-707.
[16] KHAN F. An initial seed selection algorithm for K-meansclustering of georeferenced data to improve replicability of clusterassignments for mapping application[J]. Applied Soft Computing,2012, 12(11):3698-3700.
[17] REDMOND S J, HENEGHAN C. A method for initialising the K-means clustering algorithm using kd-trees[J]. Pattern RecognitionLetters, 2007, 28(8):965-973.
[18]李武,赵娇燕,严太山.基于平均差异度优选初始聚类中心的改进K-均值聚类算法[J].控制与决策, 2017, 32(4):759-762.LI Wu, ZHAO Jiaoyan, YAN Taishan. Improved K-meansclustering algorithm optimizing initial clustering centers based onaverage difference degree[J]. Control and Decision, 2017, 32(4):759-762.
[19]陈兴蜀,吴小松,王文贤,等.基于特征关联度的K-means初始聚类中心优化算法[J].四川大学学报(工程科学版), 2015, 47(1):13-19.CHEN Xingshu, WU Xiaosong, WANG Wenxian, et al. Animproved initial cluster centers selection algorithm for K-meansbased on features correlative degree[J]. Journal of SichuanUniversity(Engineering Science Edition), 2015, 47(1):13-19.
[20]李春生,王耀南.聚类中心初始化的新方法[J].控制理论与应用, 2010, 27(10):1435-1440.LI Chunsheng, WANG Yaonan. New initialization method forcluster center[J]. Control Theory&Applications, 2010, 27(10):1435-1440.
[21]原玥,王宏,原培新,等.一种改进的Hu不变矩算法在存储介质图像识别中的应用[J].仪器仪表学报, 2016, 37(5):1043-1048.YUAN Yue, WANG Hong, YUAN Peixin, et al. An improved Huinvariant moment algorithm for storage medium image recognition[J]. Chinese Journal of Scientific Instrument, 2016, 37(5):1043-1048.
[22]董明利,马闪闪,张帆,等.基于核熵成分分析的流式数据自动分群方法[J].仪器仪表学报, 2017, 38(1):206-211.DONG Mingli, MA Shanshan, ZHANG Fan, et al. Autoclassification method of flow cytometry data based on kernelentropy component analysis[J]. Chinese Journal of ScientificInstrument, 2017, 38(1):206-211.
[23]曹鸿吉,盛斌,吴雯,等.基于改进K-means的腹内脂肪自动定量检测算法[J].计算机辅助设计与图形学学报, 2017, 29(4):576-583.CAO Hongji, SHENG Bin, WU Wen, et al. Automatic quantitativedetection algorithm of abdominal fat based on improved K-meansalgorithm[J]. Journal of Computer-Aided Design&ComputerGraphics, 2017, 29(4):576-583.
[24]杨帆,王箭,柳亚男,等.缫丝排序算法[J].计算机学报, 2012,35(4):802-810.YANG Fan, WANG Jian, LIU Yanan, et al. Reeling sort:a newalgorithms in cluster-sorting[J]. Chinese Journal of Computers,2012, 35(4):802-810.
[25] UCI Machine Learning Repository. 438 Data sets[EB/OL].[2018-06-22]. http://archive.ics.uci.edu/ml/datasets.html.
[2]陈祥光,薛锦诚,傅若农.储罐原油含水率在线测量及动态油量计量技术的研究[J].仪器仪表学报, 2000, 21(6):569-573.CHEN Xiangguang, XUE Jincheng, FU Ruonong. Study oncomputation technique in real time to dynamic oil quantity and measurement in online to water content ratio of crude oil tanks[J].Chinese Journal of Scientific Instrument, 2000, 21(6):569-573.
[3]曲艺,陈祥光.原油储罐油量动态计量技术研究[J].仪器仪表学报, 2005, 26(10):998-1001, 1010.QU Yi, WU Yingxiang. A research on the measurement ofdynamic quantity of crude oil in storage tank[J]. Chinese Journalof Scientific Instrument, 2005, 26(10):998-1001, 1010.
[4]刘春涛,陈祥光,刘志成.多传感器信息融合技术在动态油量计量中的应用[J].仪器仪表学报, 2005, 26(8):43-46.LIU Chuntao, CHEN Xiangguang, LIU Zhicheng.Application ofmulti-sensor information fusion in calculating dynamic oilquantity[J]. Chinese Journal of Scientific Instrument, 2005, 26(8):43-46.
[5]任磊,陈祥光,刘春涛.原油储罐油水界面测量方法的研究及应用[J].仪器仪表学报, 2008, 29(10):2186-2191.REN Lei, CHEN Xiangguang, LIU Chuntao. Research andapplication of measurement method for oil-water interface incrude oil tank[J]. Chinese Journal of Scientific Instrument, 2008,29(10):2186-2191.
[6]王云松,付玉红,宋艳慧.储油罐多相界面液位测控系统的研究[J].仪器仪表学报, 2003, 24(s1):285-286.WANG Yunsong, FU Yuhong, SONG Yanhui. Research onmonitor and control system of diversified liquid levels for storagetank[J]. Chinese Journal of Scientific Instrument, 2003, 24(s1):285-286.
[7]李一博,靳世久,孙墨杰,等.基于磁致伸缩效应的液体密度在线测量方法[J].天津大学学报, 2004, 37(4):368-372.LI Yibo, JIN Shijiu, SUN Mojie, et al. Liquid density on-linemeasuring based on magnetostriction effect[J]. Journal of TianjinUniversity, 2004, 37(4):368-372.
[8]彭勇,张映辉.光纤油水界面监控仪研制与应用[J].仪器仪表学报, 2005, 26(8):857-859.PENG Yong, ZHANG Yinghui. Applying and designing thecontroller of optical fiber oil interface[J]. Chinese Journal ofScientific Instrument, 2005, 26(8):857-859.
[9]马世海,吴珂,李晓薇.油-水两相界面测量技术综述[J].化工自动化及仪表, 2016, 43(12):1233-1238.MA Shihai, WU Ke, LI Xiaowei. Reviews of technologies for oil-water interface measurement[J]. Control and Instruments inChemical Industry, 2016, 43(12):1233-1238.
[10]任喜伟,任工昌,杨帆.电磁场式油水界面测量分析及数据优化方法[J].化工自动化及仪表, 2012, 39(7):858-861.REN Xiwei, REN Gongchang, YANG Fan. Oil-water interfaceanalysis and data optimization methods of electromagnetic field-typed measuring device[J]. Control and Instruments in ChemicalIndustry, 2012, 39(7):858-861.
[11]苏鑫,吴迎亚,裴华健,等.大数据技术在过程工业中的应用研究进展[J].化工进展, 2016, 35(6):1652-1659.SU Xin, WU Yingya, PEI Huajian, et al. Recent development ofthe application of big data technology in process industries[J].Chemical Industry and Engineering Progress, 2016, 35(6):1652-1659.
[12]刘远红,习玉阳,吴华远,等. K-means聚类算法在油水界面检测中的应用[J].化工自动化及仪表, 2016, 41(1):19-21.LIU Yanhong, XI Yuyang, WU Huayuan, et al. Application of K-means clustering algorithm in oil-water interface detection[J].Control and Instruments in Chemical Industry, 2016, 41(1):19-21.
[13] BAGIROY A M, UGON J, Webb D. Fast modified global K-means algorithm for incremental cluster construction[J]. Pattern Recognition, 2011, 44(4):866-876.
[14] TZORTZIS G, LIKAS A, Tzortzis G. The minmax K-meansclustering algorithm[J]. Pattern Recognition, 2014, 47(7):2505-2516.
[15] AMORIM R C D, MAKARENKOV V. Applying subclustering andLpdistance in weighted K-means with distributed centroids[J].Neurocomputing, 2016, 173(3):700-707.
[16] KHAN F. An initial seed selection algorithm for K-meansclustering of georeferenced data to improve replicability of clusterassignments for mapping application[J]. Applied Soft Computing,2012, 12(11):3698-3700.
[17] REDMOND S J, HENEGHAN C. A method for initialising the K-means clustering algorithm using kd-trees[J]. Pattern RecognitionLetters, 2007, 28(8):965-973.
[18]李武,赵娇燕,严太山.基于平均差异度优选初始聚类中心的改进K-均值聚类算法[J].控制与决策, 2017, 32(4):759-762.LI Wu, ZHAO Jiaoyan, YAN Taishan. Improved K-meansclustering algorithm optimizing initial clustering centers based onaverage difference degree[J]. Control and Decision, 2017, 32(4):759-762.
[19]陈兴蜀,吴小松,王文贤,等.基于特征关联度的K-means初始聚类中心优化算法[J].四川大学学报(工程科学版), 2015, 47(1):13-19.CHEN Xingshu, WU Xiaosong, WANG Wenxian, et al. Animproved initial cluster centers selection algorithm for K-meansbased on features correlative degree[J]. Journal of SichuanUniversity(Engineering Science Edition), 2015, 47(1):13-19.
[20]李春生,王耀南.聚类中心初始化的新方法[J].控制理论与应用, 2010, 27(10):1435-1440.LI Chunsheng, WANG Yaonan. New initialization method forcluster center[J]. Control Theory&Applications, 2010, 27(10):1435-1440.
[21]原玥,王宏,原培新,等.一种改进的Hu不变矩算法在存储介质图像识别中的应用[J].仪器仪表学报, 2016, 37(5):1043-1048.YUAN Yue, WANG Hong, YUAN Peixin, et al. An improved Huinvariant moment algorithm for storage medium image recognition[J]. Chinese Journal of Scientific Instrument, 2016, 37(5):1043-1048.
[22]董明利,马闪闪,张帆,等.基于核熵成分分析的流式数据自动分群方法[J].仪器仪表学报, 2017, 38(1):206-211.DONG Mingli, MA Shanshan, ZHANG Fan, et al. Autoclassification method of flow cytometry data based on kernelentropy component analysis[J]. Chinese Journal of ScientificInstrument, 2017, 38(1):206-211.
[23]曹鸿吉,盛斌,吴雯,等.基于改进K-means的腹内脂肪自动定量检测算法[J].计算机辅助设计与图形学学报, 2017, 29(4):576-583.CAO Hongji, SHENG Bin, WU Wen, et al. Automatic quantitativedetection algorithm of abdominal fat based on improved K-meansalgorithm[J]. Journal of Computer-Aided Design&ComputerGraphics, 2017, 29(4):576-583.
[24]杨帆,王箭,柳亚男,等.缫丝排序算法[J].计算机学报, 2012,35(4):802-810.YANG Fan, WANG Jian, LIU Yanan, et al. Reeling sort:a newalgorithms in cluster-sorting[J]. Chinese Journal of Computers,2012, 35(4):802-810.
[25] UCI Machine Learning Repository. 438 Data sets[EB/OL].[2018-06-22]. http://archive.ics.uci.edu/ml/datasets.html.