基于特征气体关联特征的变压器故障诊断方法
|
摘要
现有的基于油中溶解气体分析(dissolved gases analysis, DGA)的变压器故障诊断方法未能充分挖掘不同故障下特征气体间的关联特征。基于此,论文提出一种基于故障特征气体间关联特征的变压器故障诊断方法。利用最大信息系数(maximal information coefficient, MIC)方法定量表征不同故障类型下特征气体间的关联程度,并使用受试者工作特征(receiveroperatingcharacteristic,ROC)曲线获得不同故障类型下特征气体间的关联特征量及其分布范围,进而建立变压器故障诊断模糊推理系统;针对实际使用中数据采集周期较长的问题,对比选取了牛顿插值方法扩大待识别样本数据,保证了提取特征的有效性。针对选取的故障时序数据,论文所提方法的故障诊断准确率达到了100%,远高于三比值法和大卫三角法,说明论文所提方法充分地利用了故障特征气体间的关联特征,有效地提升了基于DGA的变压器故障诊断方法的性能,为基于DGA的变压器故障诊断提供了一种新的特征。
Existing transformer fault diagnosis methods based on dissolved gases analysis(DGA) fail to make full use of association relationship between characteristic gases in different kinds of faults. We proposed a transformer fault diagnosis method based on association characteristics between different gases. The maximal information coefficient(MIC) method was used for the quantitative characterization of the association relationship between characteristic gases, the receiver operating characteristic(ROC) curve was used to extract the association characteristics between characteristic gases and their distribution ranges, and the fuzzy inference system was established. Furthermore, in order to avoid long data-acquisition cycle in actual use, the Newton interpolation method was used to expand the size of data samples for diagnosis, which can help improve the effectiveness of characteristics. The accuracy of the method proposed on the time series data for test is 100%, which is much higher than that of the three-ratio method and David triangle method. The results illustrate that the proposed method makes full use of association relationship between characteristic gases, and can improve the performance of DGA based methods. Thus, a new kind of characteristics can be provided for transformer fault diagnosis method based on DGA.
Existing transformer fault diagnosis methods based on dissolved gases analysis(DGA) fail to make full use of association relationship between characteristic gases in different kinds of faults. We proposed a transformer fault diagnosis method based on association characteristics between different gases. The maximal information coefficient(MIC) method was used for the quantitative characterization of the association relationship between characteristic gases, the receiver operating characteristic(ROC) curve was used to extract the association characteristics between characteristic gases and their distribution ranges, and the fuzzy inference system was established. Furthermore, in order to avoid long data-acquisition cycle in actual use, the Newton interpolation method was used to expand the size of data samples for diagnosis, which can help improve the effectiveness of characteristics. The accuracy of the method proposed on the time series data for test is 100%, which is much higher than that of the three-ratio method and David triangle method. The results illustrate that the proposed method makes full use of association relationship between characteristic gases, and can improve the performance of DGA based methods. Thus, a new kind of characteristics can be provided for transformer fault diagnosis method based on DGA.
引文
[1]吴广宁,袁海满,高波,等.基于特征评估与核主元分析的电力变压器故障诊断[J].高电压技术,2017,43(8):2533-2540.WU Guangning,YUAN Haiman,GAO Bo,et al.Fault diagnosis of power transformer based on feature evaluation and kernel principal component analysis[J].High Voltage Engineering,2017,43(8):2533-2540.
[2]黄新波,李文君子,宋桐,等.采用遗传算法优化装袋分类回归树组合算法的变压器故障诊断[J].高电压技术,2016,42(5):1617-1623.HUANG Xinbo,LI Wenjunzi,SONG Tong,et al.Application of bagging-CART algorithm optimized by genetic algorithm in transformer fault diagnosis[J].High Voltage Engineering,2016,42(5):1617-1623.
[3]周建华,胡敏强.自构形神经网络在变压器故障诊断中的应用[J].电工技术学报,2004,19(9):77-81.ZHOU Jianhua,HU Minqiang.Application of auto-structural neural network in diagnosing transformer faults[J].Transactions of China Electrotechnical Society,2004,19(9):77-81.
[4]高文胜,钱政,严璋.基于决策树神经网络模型的电力变压器故障诊断方法[J].西安交通大学学报,1999,33(6):11-16.GAO Wensheng,QIAN Zheng,YAN Zhang.Fault diagnosis of power transformer using neural network of decision tree[J].Journal of Xi’an Jiaotong University,1999,33(6):11-16.
[5]吕干云,程浩忠,董立新,等.基于多级支持向量机分类器的电力变压器故障识别[J].电力系统及其自动化学报,2005,17(1):19-22.LüGanyun,CHENG Haozhong,DONG Lixin,et al.Fault diagnosis of power transformer based on multi-layer SVM classifier[J].Proceedings of Electric Power System&Automation,2005,17(1):19-22.
[6]郑蕊蕊,赵继印,赵婷婷,等.基于遗传支持向量机和灰色人工免疫算法的电力变压器故障诊断[J].中国电机工程学报,2011,31(7):56-63.ZHENG Ruirui,ZHAO Jiyin,ZHAO Tingting,et al.Power transformer fault diagnosis based on genetic support vector machine and gray artificial immune algorithm[J].Proceedings of the CSEE,2011,31(7):56-63.
[7]梁永亮,李可军,牛林,等.一种优化特征选择-快速相关向量机变压器故障诊断方法[J].电网技术,2013,37(11):3262-3267.LIANG Yongliang,LI Kejun,NIU Lin,et al.A Transformer diagnosis method based on optimized feature selection methods and fast relevance vector machine[J].Power System Technology,2013,37(11):3262-3267.
[8]王永强,律方成,李和明.基于贝叶斯网络和油中溶解气体分析的变压器故障诊断方法[J].电工技术学报,2004,19(12):74-77.WANG Yongqiang,LüFangcheng,LI Heming.Fault diagnosis for power transformer based on BN and DGA[J].Transactions of China Electrotechnical Society,2004,19(12):74-77.
[9]王学磊.变压器复合故障智能识别与热动力学焓变诊断技术研究[D].济南:山东大学,2015.WANG Xuelei.Research on the intelligent identification and thermodynamics-based reaction enthalpy diagosis methodology for transformer[D].Jinan,China:Shandong University,2015.
[10]孙莹,高贺,李可军,等.基于多时段信息融合的配电变压器运行状态评估模型[J].高电压技术,2016,42(7):2054-2062.SUN Ying,GAO He,LI Kejun,et al.Condition assessment model of distribution transformer based on multi-period information fusion[J].High Voltage Engineering,2016,42(7):2054-2062.
[11]郑元兵,孙才新,李剑,等.变压器故障特征量可信度的关联规则分析[J].高电压技术,2012,38(1):82-88.ZHENG Yuanbing,SUN Caixin,LI Jian,et al.Association rule analysis on confidence of features for transformer faults[J].High Voltage Engineering,2012,38(1):82-88.
[12]谢龙君,李黎,程勇,等.融合集对分析和关联规则的变压器故障诊断方法[J].中国电机工程学报,2015,35(2):277-286.XIE Longjun,LI Li,CHENG Yong,et al.A fault diagnosis method of power transformers by integrated set pair analysis and association rules[J].Proceedings of the CSEE,2015,35(2):277-286.
[13]孙广路,宋智超,刘金来,等.基于最大信息系数和近似马尔科夫毯的特征选择方法[J].自动化学报,2017,43(5):795-805.SUN Guanglu,SONG Zhichao,LIU Jinlai,et al.Feature selection method based on maximum information coefficient and approximate Markov blanket[J].Acta Automatica Sinica,2017,43(5):795-805.
[14]RESHEF D N,RESHEF Y A,FINUCANE H K,et al.Detecting novel associations in large data sets[J].Science,2011,334(6062):1518.
[15]FAN Y,LIU S,QIN L,et al.A novel online estimation scheme for static voltage stability margin based on relationships exploration in a large data set[J].IEEE Transactions on Power Systems,2015,30(3):1380-1393.
[16]韦修喜,周永权.基于ROC曲线的两类分类问题性能评估方法[J].计算机技术与发展,2010,20(11):47-50.WEI Xiuxi,ZHOU Yongquan.A new performance categories evaluation method based on ROC curve[J].Computer Technology&Development,2010,20(11):47-50.
[17]孙长亮,何峻,肖怀铁.基于ROC曲线的目标识别性能评估方法[J].雷达科学与技术,2007,5(1):17-21.SUN Changliang,HE Jun,XIAO Huaitie.A new performance evaluation method based on ROC curve[J].Radar Science&Technology,2007,5(1):17-21.
[18]NOORI M,EFFATNEJAD R,HAJIHOSSEINI P.Using dissolved gas analysis results to detect and isolate the internal faults of power transformers by applying a fuzzy logic method[J].IET Generation Transmission&Distribution,2017,11(10):2721-2729.
[19]杨纶标,高英仪.模糊数学原理及应用[M].广州:华南理工大学出版社,2006.YANG Lunbiao,GAO Yingyi.Theory and application of fuzzy mathematics[M].Guangzhou,China:South China University of Technology Press,2006.
[20]张琦,朱合华,黄贤斌,等.基于Mamdani模糊推理的山岭隧道围岩RMR14分级[J].岩土工程学报,2017,39(11):2116-2124.ZHANG Qi,ZHU Hehua,HUANG Xianbin,et al.A new rock mass rating method based on Mamdani fuzzy inference for rock tunnels[J].Chinese Journal of Geotechnical Engineering,2017,39(11):2116-2124.
[21]CHEN S J,CHEN S M.Fuzzy risk analysis based on similarity measures of generalized fuzzy numbers[J].IEEE Transactions on Fuzzy Systems,2005,13(6):860-860.
[22]杜洋.基于牛顿插值和神经网络的时间序列预测研究[J].石油化工高等学校学报,2007(3):20-23.DU Yang.Research of time series forecasting based on Newton interpolation and neural network[J].Journal of Petrochemical Universities,2007(3):20-23.
[2]黄新波,李文君子,宋桐,等.采用遗传算法优化装袋分类回归树组合算法的变压器故障诊断[J].高电压技术,2016,42(5):1617-1623.HUANG Xinbo,LI Wenjunzi,SONG Tong,et al.Application of bagging-CART algorithm optimized by genetic algorithm in transformer fault diagnosis[J].High Voltage Engineering,2016,42(5):1617-1623.
[3]周建华,胡敏强.自构形神经网络在变压器故障诊断中的应用[J].电工技术学报,2004,19(9):77-81.ZHOU Jianhua,HU Minqiang.Application of auto-structural neural network in diagnosing transformer faults[J].Transactions of China Electrotechnical Society,2004,19(9):77-81.
[4]高文胜,钱政,严璋.基于决策树神经网络模型的电力变压器故障诊断方法[J].西安交通大学学报,1999,33(6):11-16.GAO Wensheng,QIAN Zheng,YAN Zhang.Fault diagnosis of power transformer using neural network of decision tree[J].Journal of Xi’an Jiaotong University,1999,33(6):11-16.
[5]吕干云,程浩忠,董立新,等.基于多级支持向量机分类器的电力变压器故障识别[J].电力系统及其自动化学报,2005,17(1):19-22.LüGanyun,CHENG Haozhong,DONG Lixin,et al.Fault diagnosis of power transformer based on multi-layer SVM classifier[J].Proceedings of Electric Power System&Automation,2005,17(1):19-22.
[6]郑蕊蕊,赵继印,赵婷婷,等.基于遗传支持向量机和灰色人工免疫算法的电力变压器故障诊断[J].中国电机工程学报,2011,31(7):56-63.ZHENG Ruirui,ZHAO Jiyin,ZHAO Tingting,et al.Power transformer fault diagnosis based on genetic support vector machine and gray artificial immune algorithm[J].Proceedings of the CSEE,2011,31(7):56-63.
[7]梁永亮,李可军,牛林,等.一种优化特征选择-快速相关向量机变压器故障诊断方法[J].电网技术,2013,37(11):3262-3267.LIANG Yongliang,LI Kejun,NIU Lin,et al.A Transformer diagnosis method based on optimized feature selection methods and fast relevance vector machine[J].Power System Technology,2013,37(11):3262-3267.
[8]王永强,律方成,李和明.基于贝叶斯网络和油中溶解气体分析的变压器故障诊断方法[J].电工技术学报,2004,19(12):74-77.WANG Yongqiang,LüFangcheng,LI Heming.Fault diagnosis for power transformer based on BN and DGA[J].Transactions of China Electrotechnical Society,2004,19(12):74-77.
[9]王学磊.变压器复合故障智能识别与热动力学焓变诊断技术研究[D].济南:山东大学,2015.WANG Xuelei.Research on the intelligent identification and thermodynamics-based reaction enthalpy diagosis methodology for transformer[D].Jinan,China:Shandong University,2015.
[10]孙莹,高贺,李可军,等.基于多时段信息融合的配电变压器运行状态评估模型[J].高电压技术,2016,42(7):2054-2062.SUN Ying,GAO He,LI Kejun,et al.Condition assessment model of distribution transformer based on multi-period information fusion[J].High Voltage Engineering,2016,42(7):2054-2062.
[11]郑元兵,孙才新,李剑,等.变压器故障特征量可信度的关联规则分析[J].高电压技术,2012,38(1):82-88.ZHENG Yuanbing,SUN Caixin,LI Jian,et al.Association rule analysis on confidence of features for transformer faults[J].High Voltage Engineering,2012,38(1):82-88.
[12]谢龙君,李黎,程勇,等.融合集对分析和关联规则的变压器故障诊断方法[J].中国电机工程学报,2015,35(2):277-286.XIE Longjun,LI Li,CHENG Yong,et al.A fault diagnosis method of power transformers by integrated set pair analysis and association rules[J].Proceedings of the CSEE,2015,35(2):277-286.
[13]孙广路,宋智超,刘金来,等.基于最大信息系数和近似马尔科夫毯的特征选择方法[J].自动化学报,2017,43(5):795-805.SUN Guanglu,SONG Zhichao,LIU Jinlai,et al.Feature selection method based on maximum information coefficient and approximate Markov blanket[J].Acta Automatica Sinica,2017,43(5):795-805.
[14]RESHEF D N,RESHEF Y A,FINUCANE H K,et al.Detecting novel associations in large data sets[J].Science,2011,334(6062):1518.
[15]FAN Y,LIU S,QIN L,et al.A novel online estimation scheme for static voltage stability margin based on relationships exploration in a large data set[J].IEEE Transactions on Power Systems,2015,30(3):1380-1393.
[16]韦修喜,周永权.基于ROC曲线的两类分类问题性能评估方法[J].计算机技术与发展,2010,20(11):47-50.WEI Xiuxi,ZHOU Yongquan.A new performance categories evaluation method based on ROC curve[J].Computer Technology&Development,2010,20(11):47-50.
[17]孙长亮,何峻,肖怀铁.基于ROC曲线的目标识别性能评估方法[J].雷达科学与技术,2007,5(1):17-21.SUN Changliang,HE Jun,XIAO Huaitie.A new performance evaluation method based on ROC curve[J].Radar Science&Technology,2007,5(1):17-21.
[18]NOORI M,EFFATNEJAD R,HAJIHOSSEINI P.Using dissolved gas analysis results to detect and isolate the internal faults of power transformers by applying a fuzzy logic method[J].IET Generation Transmission&Distribution,2017,11(10):2721-2729.
[19]杨纶标,高英仪.模糊数学原理及应用[M].广州:华南理工大学出版社,2006.YANG Lunbiao,GAO Yingyi.Theory and application of fuzzy mathematics[M].Guangzhou,China:South China University of Technology Press,2006.
[20]张琦,朱合华,黄贤斌,等.基于Mamdani模糊推理的山岭隧道围岩RMR14分级[J].岩土工程学报,2017,39(11):2116-2124.ZHANG Qi,ZHU Hehua,HUANG Xianbin,et al.A new rock mass rating method based on Mamdani fuzzy inference for rock tunnels[J].Chinese Journal of Geotechnical Engineering,2017,39(11):2116-2124.
[21]CHEN S J,CHEN S M.Fuzzy risk analysis based on similarity measures of generalized fuzzy numbers[J].IEEE Transactions on Fuzzy Systems,2005,13(6):860-860.
[22]杜洋.基于牛顿插值和神经网络的时间序列预测研究[J].石油化工高等学校学报,2007(3):20-23.DU Yang.Research of time series forecasting based on Newton interpolation and neural network[J].Journal of Petrochemical Universities,2007(3):20-23.