一种改进高斯核度量的HEC算法在变压器故障诊断中的应用
|
摘要
针对传统超球型聚类算法难以解决变压器故障诊断问题的特性,使用一种改进的高斯核的超椭球聚类(hyper-ellipsoidal clustering,HEC)算法,并将其解释为寻找体积和密度都紧凑的椭球分簇,该算法能够有效地处理形状为椭球、大小不同和密度不同的分簇。在模拟数据集上的仿真实验表明所提算法在聚类结果和性能上优于K-Means算法、模糊C-Means算法和混合高斯模型期望最大化算法,从而验证了该提算法在处理椭球形或复杂形状数据集聚类时的可行性和有效性;同时将该算法应用在基于变压器油中溶解气体(dissolved gas-in-oil analysis,DGA)的变压器故障诊断中,验证了该方法更高的故障诊断准确度。
In allusion to the problem of traditional hyper sphere clustering algorithm being unable to solve the problem of transformer fault diagnosis,a kind of hyper-ellipsoidal clustering(HEC)algorithm based on improved Gaussian kernel metric is used.This HEC algorithm is illustrated as searching for ellipsoid clusters with compact volume and density,which is proved to be effectively handle with clusters of ellipsoid shape with different sizes and densities.Experiment on simulating dataset indicates the proposed HEC algorithm is prior to K-Means algorithm,fuzzy C-Means algorithm and GMM-EM algorithm,which verifies feasibility and validity of HEC algorithm in processing problems of ellipsoid dataset or complex-shaped dataset.Application of HEC algorithm in transformer fault diagnosis based on dissolved gas-in-oil also proves higher fault diagnosis accuracy of this method.
In allusion to the problem of traditional hyper sphere clustering algorithm being unable to solve the problem of transformer fault diagnosis,a kind of hyper-ellipsoidal clustering(HEC)algorithm based on improved Gaussian kernel metric is used.This HEC algorithm is illustrated as searching for ellipsoid clusters with compact volume and density,which is proved to be effectively handle with clusters of ellipsoid shape with different sizes and densities.Experiment on simulating dataset indicates the proposed HEC algorithm is prior to K-Means algorithm,fuzzy C-Means algorithm and GMM-EM algorithm,which verifies feasibility and validity of HEC algorithm in processing problems of ellipsoid dataset or complex-shaped dataset.Application of HEC algorithm in transformer fault diagnosis based on dissolved gas-in-oil also proves higher fault diagnosis accuracy of this method.
引文
[1]金建国.聚类方法综述[J].计算机科学,2014,41(增刊2):288–293.JIN Jianguo.Suvey on Clustering Methods[J].Computer Science,2014,41(S2):288-293.
[2]刘勇,赵斌,夏绍玮.模糊超椭球分类算法及其在无约束手写体数字识别中的应用[J].清华大学学报(自然科学版),2000,40(9):120–124.LIU Yong,ZHAO Bin,XIA Shaowei.Self-organizing Network with Fuzzy Hyperellipsoidal Classifying and Its Application in Unconstrained Handwritten Numeral Recognition[J].Journal of Tsinghua University(Science&Technology),2000,40(9):120-124.
[3]MAO J C,JAIN A K.A Self-organizing Network for Hyperellipsoidal Clustering(HEC)[J].IEEE Transactions on Neural Networks,1996,7(1):16-29.
[4]秦玉平,王祎,伦淑娴,等.基于超椭球支持向量机的兼类文本分类算法[J].计算机科学,2013,40(增刊2):98–100.QIN Yuping,WANG Yi,LUN Shuxian,et al.Multi-label Text Classification Algorithm Based on Hyper Ellipsoidal SVM[J].Computer Science,2013,40(S2):98-100.
[5]梁夷龙,王松,夏绍玮,等.基于超椭球模糊聚类的人脑磁共振图象分割[J].软件学报,1998,9(9):683–689.LIANG Yilong,WANG Song,XIA Shaowei,et al.Human Brain Magnetic Resonance Image Segmentation Based on Hyperellipsoidal Fuzzy Clustering Algorithm[J].Journal of Software,1998,9(9):683-689.
[6]MOSHTAGHI M,RAJASEGARAR S,LECKIE C,et al.An Effient Hyperellipsoidal Clustering Algorithm forResourceconstrained Environments[J].Pattern Recognition,2011,44(9):2197-2209.
[7]朱峰,宋余庆,陈健美.基于椭球等高分布混合模型的聚类方法[J].江苏大学学报(自然科学版),2011,32(6):701–705.ZHU Feng,SONG Yuqing,CHEN Jianmei.Clustering Method Based on Elliptical Contoured Mixture Model[J].Journal of Jiangsu University(Natural Science Edition),2011,32(6):701-705.
[8]LEE H,PARK J,PARK D.Hyper-ellipsoidal ClusteringAlgorithm Using Linear Matrix Inequality[J].Journal of Korea Institute Intelligent Systems,2002,12(4):300-305.
[9]MAHESH K O,JAMES B.Scale-invariant Clustering with Minimum Volume Ellipsoids[J].Computer&Operations Research,2008,35(4):1017-1029.
[10]SHIODA R,TUNCEL L.Clustering via Minimum Volume Ellipsoids[J].Computational Optimization and Applications,2007,37(3):247-295.
[11]STEPHEN B,LIEVEN V.Convex Optimization[M].Cambridge,UK:Cambridge University Press,2004.
[12]TODD M J,YILDIRIM E A.On Khachiyan's Algorithm for the Computation of Minimum-volume Enclosing Ellipsoids[J].Discrete Applied Mathematics,2007,155(13):1731-1744.
[13]CAO J Z,CHEN P,ZHENG Y,et al.A Max-flow-based Similarity Measure for Spectral[J].Etri Journal,2013,35(2):311-320.
[14]JOHN S T,NELLO C.Kernel Methods for Pattern Analysis[M].Cambridge,UK:Cambridge University Press,2004.
[15]张冠军,严璋,张仕君.电力变压器故障诊断中新方法的应用[J].高压电器,1998(4):32–34.ZHANG Guanjun,YAN Zhang,ZHANG Shijun.Application of New Fault Diagnosis Method on Electric Power Transformers[J].High Voltage Apparatus,1998(4):32-34.
[16]郑含博,王伟,李晓纲,等.基于多分类最小二乘支持向量机和改进粒子群优化算法的电力变压器故障诊断方法[J].高电压技术,2014,40(11):3424–3429.ZHENG Hanbo,WANG Wei,LI Xiaogang,et al.Fault Diagnosis Method of Power Transformers Using Multi-class LS-SVM and Improved PSO[J].High Voltage Engineering,2014,40(11):3424-3429.
[17]朱永利,尹金良.组合核相关向量机在电力变压器故障诊断中的应用研究[J].中国电机工程学报,2013,33(22):68–75.ZHU Yongli,YIN Jinliang.Study on Application of Multikernel Learning Relevance Vector Machines in Fault Diagnosis of Power Transformers[J].Proceedings of the CSEE,2013,33(22):68-75.
[2]刘勇,赵斌,夏绍玮.模糊超椭球分类算法及其在无约束手写体数字识别中的应用[J].清华大学学报(自然科学版),2000,40(9):120–124.LIU Yong,ZHAO Bin,XIA Shaowei.Self-organizing Network with Fuzzy Hyperellipsoidal Classifying and Its Application in Unconstrained Handwritten Numeral Recognition[J].Journal of Tsinghua University(Science&Technology),2000,40(9):120-124.
[3]MAO J C,JAIN A K.A Self-organizing Network for Hyperellipsoidal Clustering(HEC)[J].IEEE Transactions on Neural Networks,1996,7(1):16-29.
[4]秦玉平,王祎,伦淑娴,等.基于超椭球支持向量机的兼类文本分类算法[J].计算机科学,2013,40(增刊2):98–100.QIN Yuping,WANG Yi,LUN Shuxian,et al.Multi-label Text Classification Algorithm Based on Hyper Ellipsoidal SVM[J].Computer Science,2013,40(S2):98-100.
[5]梁夷龙,王松,夏绍玮,等.基于超椭球模糊聚类的人脑磁共振图象分割[J].软件学报,1998,9(9):683–689.LIANG Yilong,WANG Song,XIA Shaowei,et al.Human Brain Magnetic Resonance Image Segmentation Based on Hyperellipsoidal Fuzzy Clustering Algorithm[J].Journal of Software,1998,9(9):683-689.
[6]MOSHTAGHI M,RAJASEGARAR S,LECKIE C,et al.An Effient Hyperellipsoidal Clustering Algorithm forResourceconstrained Environments[J].Pattern Recognition,2011,44(9):2197-2209.
[7]朱峰,宋余庆,陈健美.基于椭球等高分布混合模型的聚类方法[J].江苏大学学报(自然科学版),2011,32(6):701–705.ZHU Feng,SONG Yuqing,CHEN Jianmei.Clustering Method Based on Elliptical Contoured Mixture Model[J].Journal of Jiangsu University(Natural Science Edition),2011,32(6):701-705.
[8]LEE H,PARK J,PARK D.Hyper-ellipsoidal ClusteringAlgorithm Using Linear Matrix Inequality[J].Journal of Korea Institute Intelligent Systems,2002,12(4):300-305.
[9]MAHESH K O,JAMES B.Scale-invariant Clustering with Minimum Volume Ellipsoids[J].Computer&Operations Research,2008,35(4):1017-1029.
[10]SHIODA R,TUNCEL L.Clustering via Minimum Volume Ellipsoids[J].Computational Optimization and Applications,2007,37(3):247-295.
[11]STEPHEN B,LIEVEN V.Convex Optimization[M].Cambridge,UK:Cambridge University Press,2004.
[12]TODD M J,YILDIRIM E A.On Khachiyan's Algorithm for the Computation of Minimum-volume Enclosing Ellipsoids[J].Discrete Applied Mathematics,2007,155(13):1731-1744.
[13]CAO J Z,CHEN P,ZHENG Y,et al.A Max-flow-based Similarity Measure for Spectral[J].Etri Journal,2013,35(2):311-320.
[14]JOHN S T,NELLO C.Kernel Methods for Pattern Analysis[M].Cambridge,UK:Cambridge University Press,2004.
[15]张冠军,严璋,张仕君.电力变压器故障诊断中新方法的应用[J].高压电器,1998(4):32–34.ZHANG Guanjun,YAN Zhang,ZHANG Shijun.Application of New Fault Diagnosis Method on Electric Power Transformers[J].High Voltage Apparatus,1998(4):32-34.
[16]郑含博,王伟,李晓纲,等.基于多分类最小二乘支持向量机和改进粒子群优化算法的电力变压器故障诊断方法[J].高电压技术,2014,40(11):3424–3429.ZHENG Hanbo,WANG Wei,LI Xiaogang,et al.Fault Diagnosis Method of Power Transformers Using Multi-class LS-SVM and Improved PSO[J].High Voltage Engineering,2014,40(11):3424-3429.
[17]朱永利,尹金良.组合核相关向量机在电力变压器故障诊断中的应用研究[J].中国电机工程学报,2013,33(22):68–75.ZHU Yongli,YIN Jinliang.Study on Application of Multikernel Learning Relevance Vector Machines in Fault Diagnosis of Power Transformers[J].Proceedings of the CSEE,2013,33(22):68-75.