改进多分类支持向量机的配电网故障识别方法
|
摘要
准确识别故障是配电网故障定位与治理研究的实现前提。提出以故障分量均方根及欧氏距离为特征量,结合改进多分类支持向量机(SVM)的配电网短路故障识别方法。首先,对馈线三相电流及母线零序电压故障后一周波的故障分量进行小波分解,并重构第2层的近似分量;其次,求取重构信号的均方根及欧氏距离作为特征向量;最终输入至改进多分类支持向量机完成配电网故障类型识别。10kV典型配电网软件仿真模型的与配电网物理仿真实验系统测试结果表明,所提出方法不但可高准确率地识别典型中压配电网常见故障,且能适应于中性点运行方式调整、分布式电源并网后等情况,验证了方法的准确性与适应性。
Accurate fault identification is the premise of study work of fault location and management in power distribution network. An identification method of short-circuit fault types used in power distribution network is proposed,it takes root-mean-square and Euclidean distance as characteristic variables and combines these with the improved multi-classification support vector machine( SVM),then a new multi-classifier is built. First,the three-phase currents of low-voltage inlet line of the main transformer and zero-sequence voltage of grid bus are obtained. Next,the one-week fault components of the waveforms after fault moment are decomposed by wavelet transform,and approximation component is reconstructed with the second layer decomposition results. Then,root-mean-square and Euclidean distance of the reconstructed signals are obtained as the feature vectors. Finally,by taking these vectors as the input of the improved multiclassification support vector machine,fault types of the distribution network can be identified. Test results of the typical 10 k V software simulation model and the physical simulation experiment system of distribution network show that the proposed method not only identifies ten types of common faults in a typical distribution network with high accuracy but also adapts well to the situations such as neutral ground operation mode adjustment,distributed generation and so on,which verified the accuracy and adaptability of the proposed method.
Accurate fault identification is the premise of study work of fault location and management in power distribution network. An identification method of short-circuit fault types used in power distribution network is proposed,it takes root-mean-square and Euclidean distance as characteristic variables and combines these with the improved multi-classification support vector machine( SVM),then a new multi-classifier is built. First,the three-phase currents of low-voltage inlet line of the main transformer and zero-sequence voltage of grid bus are obtained. Next,the one-week fault components of the waveforms after fault moment are decomposed by wavelet transform,and approximation component is reconstructed with the second layer decomposition results. Then,root-mean-square and Euclidean distance of the reconstructed signals are obtained as the feature vectors. Finally,by taking these vectors as the input of the improved multiclassification support vector machine,fault types of the distribution network can be identified. Test results of the typical 10 k V software simulation model and the physical simulation experiment system of distribution network show that the proposed method not only identifies ten types of common faults in a typical distribution network with high accuracy but also adapts well to the situations such as neutral ground operation mode adjustment,distributed generation and so on,which verified the accuracy and adaptability of the proposed method.
引文
[1] SALIM R H,DE OLIVEIRA K R C,FILOMENA A D,et al. Hybrid fault diagnosis scheme implementation forpower distribution systems automation[J]. IEEETransactions on Power Delivery, 2008, 23(4):1846-1856.
[2] DAS B. Fuzzy logic-based fault-type identification inunbalanced radial power distribution system[J]. IEEETransactions on Power Delivery,2006,21(1):278-285.
[3] EYADA A A,MAHMOUD A Y,AZRUL M A. Detectionof faulted phase type in distribution systems based on oneend voltage measurement[J]. Electrical Power andEnergy Systems,2014,54(1):288-292.
[4]郭谋发,游林旭,洪翠,等.基于LCD-Hilbert谱奇异值和多级支持向量机的配电网故障识别方法[J].高电压技术,2017,43(4):1239-1247.GUO M F,YOU L X,HONG C,et al. Identificationmethod of distribution network faults based on singularvalue of LCD-Hilbert spectrums and multilevel SVM[J].High Voltage Engineering,2017,43(4):1239-1247.
[5] GOPAKUMAR P,REDDY M J B,MOHANTA D K.Adaptive fault identification and classificationmethodology for smart power grids using synchronousphasor angle measurements[J]. Generation,Transmission&Distribution, IET, 2015, 9(2):133-145.
[6] MISHRA M,ROUT P K. Detection and classification ofmicro-grid faults based on HHT and machine learningtechniques[J]. IET Generation Transmission&Distribution,2018,12(2):388-397.
[7]高伟,陈伟凡,杨耿杰,等.基于奇异值分解和多级支持向量机的配电网故障类型识别[J].电子测量与仪器学报,2018,32(2):62-71.GAO W,CHEN W F,YANG G J,et al. Fault typeidentification for distribution network based on singularvalue decomposition and multi-level support vectormachine[J]. Journal of Electronic Measurement andInstrumentation,2018,32(2):62-71.
[8] ZHANG J,HE Z Y,LIN S,et al. An ANFIS-basedfault classification approach in power distribution system[J]. International Journal of Electrical Power&EnergySystems,2013,49(1):243-252.
[9]孙鹏,曹雨晨,刘洋,等.采用二进制蚁群模糊神经网络的配电网故障分类方法[J].高电压技术,2016,42(7):2063-2072.SUN P,CAO Y CH,LIU Y,et al. Fault classificationtechnique for power distribution network using binary antcolony algorithm and fuzzy neural network[J]. HighVoltage Engineering,2016,42(7):2063-2072.
[10]王思明,童安蓉.混合模糊H网在配电网故障识别中的应用[J].电力系统及其自动化学报,2016,28(10):74-79.WANG S M,TONG AN R. Application of hybrid fuzzy Hnet to recognition of fault types for distribution network[J].Proceedings of the Chinese Society of Universities forElectric Power System and Automation,2016,28(10):74-79.
[11] NGUI W K,LEONG M S,HEE L M,et al. Waveletanalysis:Mother wavelet selection methods[J]. AppliedMechanics&Materials,2013(393):953-958.
[12]洪涛,黄志奇,杨畅.涡轮泵实时故障检测的快速支持向量机算法[J].仪器仪表学报,2012,33(8):1786-1792.HONG T, HUANG ZH Q, YANG CH. Fast supportvector machine algorithm for turbopump real-time faultdetection[J]. Chinese Journal of Scientific Instrument,2012,33(8):1786-1792.
[13]王星华,陈卓优,彭显刚.一种基于双层聚类分析的负荷形态组合识别方法[J].电网技术,2016,40(5):1495-1501.WANG X H, CHEN ZH Y, PENG X G. A newcombinational electrical load analysis method based onbilayer clustering analysis[J]. Power System Technology,2016,40(5):1495-1501.
[14]王道明,鲁昌华,蒋薇薇,等.基于粒子群算法的决策树SVM多分类方法研究[J].电子测量与仪器学报,2015,29(4):611-615.WANG D M,LU CH H,JIANG W W,et al. Study onPSO-based decision-tree SVM multi-class classificationmethod[J]. Journal of Electronic Measurement andInstrumentation,2015,29(4):611-615.
[15]王振树,李林川,牛丽.基于贝叶斯证据框架的支持向量机负荷建模[J].电工技术学报,2009,24(8):127-134,140.WANG ZH SH,LI L CH,NIU L. Load modeling basedon support vector machine based on Bayesian evidenceframework[J]. Transactions of China ElectrotechnicalSociety,2009,24(8):127-134,140.
[16]宋国明,王厚军,姜书艳,等.一种聚类分层决策的SVM模拟电路故障诊断方法[J].仪器仪表学报,2010,31(5):998-1004.SONG G M,WANG H J,JIANG SH Y,et al. Analogcircuit fault diagnosis approach using clustering andhierarchical decision SVM[J]. Chinese Journal ofScientific Instrument,2010,31(5):998-1004.
[17]傅望,周林,郭珂,等.光伏电池工程用数学模型研究[J].电工技术学报,2011,26(10):211-216.FU W, ZHOU L, GUO K, et al. Research onengineering analytical model of solar cells[J].Transactions of China Electrotechnical Society, 2011,26(10):211-216.
[18]肖文波,胡方雨,戴锦,等.太阳电池工程数学模型精确度对比分析研究[J].太阳能学报,2016,37(3):703-708.XIAO W B,HU F Y,DAI J,et al. Comparsion andinvestigation on the accuracy of the photovoltaic cellsengerineering mathematical model[J]. Acta EnergiaeSolaris Sinica,2016,37(3):703-708.
[19] WANG L,LEE D. Load-tracking performance of anautonomous SOFC-based hybrid power generation/energystorage system[J]. IEEE Transactions on EnergyConversion,2010,25(1):128-139.
[2] DAS B. Fuzzy logic-based fault-type identification inunbalanced radial power distribution system[J]. IEEETransactions on Power Delivery,2006,21(1):278-285.
[3] EYADA A A,MAHMOUD A Y,AZRUL M A. Detectionof faulted phase type in distribution systems based on oneend voltage measurement[J]. Electrical Power andEnergy Systems,2014,54(1):288-292.
[4]郭谋发,游林旭,洪翠,等.基于LCD-Hilbert谱奇异值和多级支持向量机的配电网故障识别方法[J].高电压技术,2017,43(4):1239-1247.GUO M F,YOU L X,HONG C,et al. Identificationmethod of distribution network faults based on singularvalue of LCD-Hilbert spectrums and multilevel SVM[J].High Voltage Engineering,2017,43(4):1239-1247.
[5] GOPAKUMAR P,REDDY M J B,MOHANTA D K.Adaptive fault identification and classificationmethodology for smart power grids using synchronousphasor angle measurements[J]. Generation,Transmission&Distribution, IET, 2015, 9(2):133-145.
[6] MISHRA M,ROUT P K. Detection and classification ofmicro-grid faults based on HHT and machine learningtechniques[J]. IET Generation Transmission&Distribution,2018,12(2):388-397.
[7]高伟,陈伟凡,杨耿杰,等.基于奇异值分解和多级支持向量机的配电网故障类型识别[J].电子测量与仪器学报,2018,32(2):62-71.GAO W,CHEN W F,YANG G J,et al. Fault typeidentification for distribution network based on singularvalue decomposition and multi-level support vectormachine[J]. Journal of Electronic Measurement andInstrumentation,2018,32(2):62-71.
[8] ZHANG J,HE Z Y,LIN S,et al. An ANFIS-basedfault classification approach in power distribution system[J]. International Journal of Electrical Power&EnergySystems,2013,49(1):243-252.
[9]孙鹏,曹雨晨,刘洋,等.采用二进制蚁群模糊神经网络的配电网故障分类方法[J].高电压技术,2016,42(7):2063-2072.SUN P,CAO Y CH,LIU Y,et al. Fault classificationtechnique for power distribution network using binary antcolony algorithm and fuzzy neural network[J]. HighVoltage Engineering,2016,42(7):2063-2072.
[10]王思明,童安蓉.混合模糊H网在配电网故障识别中的应用[J].电力系统及其自动化学报,2016,28(10):74-79.WANG S M,TONG AN R. Application of hybrid fuzzy Hnet to recognition of fault types for distribution network[J].Proceedings of the Chinese Society of Universities forElectric Power System and Automation,2016,28(10):74-79.
[11] NGUI W K,LEONG M S,HEE L M,et al. Waveletanalysis:Mother wavelet selection methods[J]. AppliedMechanics&Materials,2013(393):953-958.
[12]洪涛,黄志奇,杨畅.涡轮泵实时故障检测的快速支持向量机算法[J].仪器仪表学报,2012,33(8):1786-1792.HONG T, HUANG ZH Q, YANG CH. Fast supportvector machine algorithm for turbopump real-time faultdetection[J]. Chinese Journal of Scientific Instrument,2012,33(8):1786-1792.
[13]王星华,陈卓优,彭显刚.一种基于双层聚类分析的负荷形态组合识别方法[J].电网技术,2016,40(5):1495-1501.WANG X H, CHEN ZH Y, PENG X G. A newcombinational electrical load analysis method based onbilayer clustering analysis[J]. Power System Technology,2016,40(5):1495-1501.
[14]王道明,鲁昌华,蒋薇薇,等.基于粒子群算法的决策树SVM多分类方法研究[J].电子测量与仪器学报,2015,29(4):611-615.WANG D M,LU CH H,JIANG W W,et al. Study onPSO-based decision-tree SVM multi-class classificationmethod[J]. Journal of Electronic Measurement andInstrumentation,2015,29(4):611-615.
[15]王振树,李林川,牛丽.基于贝叶斯证据框架的支持向量机负荷建模[J].电工技术学报,2009,24(8):127-134,140.WANG ZH SH,LI L CH,NIU L. Load modeling basedon support vector machine based on Bayesian evidenceframework[J]. Transactions of China ElectrotechnicalSociety,2009,24(8):127-134,140.
[16]宋国明,王厚军,姜书艳,等.一种聚类分层决策的SVM模拟电路故障诊断方法[J].仪器仪表学报,2010,31(5):998-1004.SONG G M,WANG H J,JIANG SH Y,et al. Analogcircuit fault diagnosis approach using clustering andhierarchical decision SVM[J]. Chinese Journal ofScientific Instrument,2010,31(5):998-1004.
[17]傅望,周林,郭珂,等.光伏电池工程用数学模型研究[J].电工技术学报,2011,26(10):211-216.FU W, ZHOU L, GUO K, et al. Research onengineering analytical model of solar cells[J].Transactions of China Electrotechnical Society, 2011,26(10):211-216.
[18]肖文波,胡方雨,戴锦,等.太阳电池工程数学模型精确度对比分析研究[J].太阳能学报,2016,37(3):703-708.XIAO W B,HU F Y,DAI J,et al. Comparsion andinvestigation on the accuracy of the photovoltaic cellsengerineering mathematical model[J]. Acta EnergiaeSolaris Sinica,2016,37(3):703-708.
[19] WANG L,LEE D. Load-tracking performance of anautonomous SOFC-based hybrid power generation/energystorage system[J]. IEEE Transactions on EnergyConversion,2010,25(1):128-139.