分享
分享赚钱 收藏 举报 版权申诉 / 4

类型Research of Typical Mechanical Fault Diagnosis for HV Circuit Breaker Based on Vibration Signals.pdf

  • 上传人:wrn
  • 文档编号:15718637
  • 上传时间:2023-01-14
  • 格式:PDF
  • 页数:4
  • 大小:2.03MB
  • 配套讲稿:

    如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。

    特殊限制:

    部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。

    关 键  词:
    Research of Typical Mechanical Fault Diagnosis for HV Circuit Breaker Based on Vibration Signals.pdf
    资源描述:

    1、 Abstract High voltage circuit breaker is an important switchgear of the power system.For opening and closing within a few mili-seconds,it is highly sensitive with speedy transmission and great reliability.According to statistics,63.8 percent of fault of high voltage circuit breakers is caused by me

    2、chanical faults.In this work,considering a high voltage circuit breaker as the subject,the vibration signal under typical mechanical faults are collected based on some testing,including spring failure,clearance error,cam eccentricity,over the dead point and so on.Then the intelligent diagnosis model

    3、 for HV circuit breaker based on vibration signals is established to realize the feature extraction and fault identification.The results indicate that the fault diagnosis rate is up to 94%under the current samples.Index Terms HV Circuit Breaker;Vibration signal;Fault diagnosis;Support vector machine

    4、I.INTRODUCTIONHigh voltage circuit breaker is one of the most important equipment for protecting and controlling high voltage transmission and distribution network in electric power system.As the use of electricity is increasing,higher stability and reliability of electric power operating mechanism

    5、are required.When the complex fault occurs it will lead power system accidents and result in large economic losses.According to statistics,the mechanical fault of operating mechanism accounted for 63.8%in all high-voltage circuit breaker faults1.Therefore,it is important to research of typical mecha

    6、nical fault diagnosis for high voltage circuit breaker.In recent years,the fault diagnosis for high voltage circuit breakers has become the hot spot of domestic and international research 2-9.In this investigation,considering a high voltage circuit breaker with spring actuator as object of study,thr

    7、ough online testing system,some typical mechanical fault vibration signals of high voltage circuit breaker are collected,then the wavelet packet and energy entropy are used to extract the signalfeatures.Finally,based on support vector machine(SVM)method,the intelligent fault diagnosis models(SVM,GS-

    8、SVM and PSO-SMV)for high voltage circuit breakers are presented and investigated.II.EXPERIMENTAL TEST SYSTEMThe mechanical fault of high voltage circuit breaker is the main forms of the faults.In this section,the experimental test system for mechanical characteristics of high voltage circuit breaker

    9、 is designed and constructed,as is shown in Fig.1.Moreover,the six kinds of typical mechanical faults are analyzed based the testing platform,which are spring failure,clearance error,lock catch failure,cam eccentricity,mechanical jam and over the dead point.(e)Installation of acceleration sensor(f)D

    10、AQ cardPCB-350C03ScratchFailing to closeDead point(a)Jam fault(b)Spring failure(c)Over dead pointFig.1.Experimental test system.MENG fangang1,2,LIU zhansheng1,Yang qiguo2,FENG yongzhi2,SUN liquan2(1.Harbin Institute of Technology Harbin 150001,China;2.Harbin Electric Company Limited Harbin 150040,Ch

    11、ina)2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201805180000004 Page1/4 2861 The acceleration vibration signals contain rich information in the opening and closing of high voltage circuit breaker,which has many characteristics such as low

    12、attenuation,easy collection and flexible measurement.Therefore,it is effective and feasible to diagnosis faults of high voltage circuit breaker using vibration acceleration signal.And the acceleration vibration signals based testing system is collected,as is shown in Fig.2.0 0.01 0.02 0.03 0.04-1001

    13、0Acceleration(m/s2)Time(s)III.WAVELET PACKET-ENERGY ENTROPY The vibration acceleration signal of high voltage circuit breaker due to the high power pulse has high dimensional and and nolinear characteristics.Therefore,the wavelet packet energy entropy method is used to realize the signal feature ext

    14、raction.In order to obtain more mechanical characteristic information,the three-layer wavelet packet is used to decompose the signal with db5 wavelet base.The wavelet packet-characteristic entropy is etracted,then vectors of signals are reconstructed.The original vibration signal S can be decomposed

    15、 by wavelet packet for three layers as:3 3 3 33 3 3 3S AAA DAA ADA DDAAA D DAD ADD DDD(1)Where A is low frequency component,D is high frequency component.Then the signal envelope is extracted based on Hilbert method,and the signal energy in different frequency phase by integrating is caculated as:02

    16、()ititE A t dt(2)Where t 0 is start time of signal,t i is end time,A(t)represents the signal envelope.Normalization processing of signal energy in different frequency phase is expressed as:1ii niiEE(3)Where n is the length of the original signal.Based on theory of information entropy,the wavelet pac

    17、ket characteristic entropy is expressed as:1logni iiH(4)IV.INTELLIGENT FAULT DIAGNOSIS MODELING The support vector machine(SVM)is a machine learning method based on the small sample statistics study theory,which originated from from the notion of optimal hyperplane 10.The classification principle of

    18、 SVM is shown in Fig.3.H1H2H2/MarginOptimalhyperplaneClass 1Class 2 Classification principle of SVM The classification principle of SVM can be expressed as:2,1min 02.()1ni iw biTi i ics t y x b(5)Where 1,ni iix y represents the data set of samples i is slack variable c is penalty factor b is basis f

    19、actor.The optimization object function based on lagrange multiplier method is calculated as:1 1 111max()()2.0,0l l li i j i j i ji i jli j iiQ y y x xs t y c(6)The kernel function K(.)is constructed as:(,)()()Ti j i jK x x x x(7)In this investigation,the RBF kernel function is used,it is expressed a

    20、s:2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201805180000004 Page2/4 2862 22()(,)exp iix xK x x(8)Therefore,the formula(6)is transformated as 1 1 111max()()2.0,0l l li i j i j i ji i jli j iiQ y y x xs t y c(9)The corresponding discrimina

    21、nt function is expressed as 1()sgn(,)ni i iif x y K x x b(10)Calculate trainingsamplesAffirm fitness function accuracy rate Initialize parametersCalculate fitnessfunction svmtrain Fitness scaling accuracy rate If the condition is metAffirm fitness function accuracy rate Update parametersOutput optim

    22、al result bestc bestg Get the final optimalfault diagnosis modelNYCalculate trainingsamplesAffirm fitness function accuracy rate Initialize population andspeedCalculate fitnessfunction svmtrain Calibrate fitness accuracy rate If the condition is metVerify optimal resultUpdate speed andparticleOutput

    23、 optimal result bestc bestg Get the final optimalfault diagnosis modelNY(a)GS-SVM algorithm flow chart(b)PSO-SVM algorithm flow chart The identification ability of SVM are directly determined by the key parameter values.Improper parameter will tend to occur overfitting and underfitting phenomenon.Th

    24、erefore,the particle swarm optimization(PSO)and grid search(GS)algorithms are used to realize parameters optimization in this work,the optimization algorithms flow chart is shown in Fig.4.V.SIMULATION RESULTS AND DISCUSSION In this study,the db5 wavelet base is selected to carry on the 3 layer wavel

    25、et packet transformation.Then the 8 spectrums are obtained and can be used as the elements to make up the energy entropy feature vector H 30,H 31,H 32,H 33,H 34,H 35,H 36,H 37.The wavelet packet energy entropy of some vibration signals based on experimental test is listed in table1.In this work,the

    26、six typical mechanical faults of high voltage circuit breaker are investigated.Therefore,the codes of the seven types of working conditions(normal condition,spring failure,clearance error,lock catch failure,cam eccentricity,mechanical jam and over the dead point)based on SVM are respectively.The 140

    27、 groups of the data from experimental test are used as training data,and 70 groups of the data are used as validation data.The initial parameters of SVM are set as:c 2-10,210 g 2-10,210.The GS algorithm result is shown in Fig.5.the optimal parameters are obtained as:bectc=0.574 bestg=0.268.And optim

    28、al parameters of PSO algorithms are calculated as:bectc=1.229 bestg=0.365.-10-50510-100104060801000510-6-4-20244060801005458 6262 62 626666 66666666707070707070 70 7074 7474747474 74 74 7474 747478787878 7878 78787878828282828282828282828282 82868686 86-2 0 2 4 6 8 10-6-4-2024505050 505054.554.554.5

    29、 54.559 595963.563.5 63.563.56868 68686868686868686868686868686868686872.5 72.572.572.572.5 72.572.5-10-5 0 5 10-10-50510Accuracy(%)log2glog2clog2c log2gAccuracy(%)log2c log2glog2glog2c(a)Rough searching(b)Fine searching The SVM,GS-SVM and PSO-SVM models are established and carried out typical mecha

    30、nical fault diagnosis for high voltage circuit breaker based on vibration acceleration signal.The result is shown in table 2.Through data analysis,the SVM model has a low accuracy of fault diagnosis,only 78.6%,but it possesses high computation efficiency.The GS-SVM and PSO-SVM models have the higher

    31、 accuracy of fault diagnosis,and the highest fault diagnosis rate of PSO-SVM is up to 94%.Therefore,considering the accuracy of fault diagnosis and computation efficiency,the result indicates that the PSO-SVM model for mechanical fault diagnosis of HV circuit breaker can be widely applied in actual

    32、engineering.2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201805180000004 Page3/4 2863CHARACTERISTIC VALUE OF WAVELET PACKET ENERGY ENTROPY Number H 30 H 31 H 32 H 33 H 34 H 35 H 36 H 37 1 0.0474 0.0520 0.131 0.133 0.3133 0.06739 0.1488 0.10

    33、56 2 0.0407 0.0588 0.1435 0.1143 0.3042 0.0791 0.1537 0.1054 3 0.0636 0.0523 0.1815 0.0817 0.1854 0.1247 0.1474 0.1630 4 0.1244 0.1110 0.1448 0.0997 0.1858 0.0751 0.1675 0.0913 5 0.0956 0.0957 0.1952 0.0830 0.2434 0.0739 0.1417 0.0711 TABLE 2 RERULT ANALYSIS Intelligent diagnosis model Nomal conditi

    34、on Cam eccentricity Clearance error Spring failure Lock catch failure Dead point Mechanical jam recognition rate Computation time SVM 70%80%70%70%80%90%90%78.6%12.7s GS-SVM 60%100%70%100%100%90%100%88.6%83.3s PSO-SVM 100%100%80%100%80%100%100%94.3%107.2s VI.CONCLUSION In this work,the experimental t

    35、est system for mechanical characteristics of high voltage circuit breaker is designed and constructed,and some typical mechanical fault vibration signals are collected.The intelligent diagnosis models(SVM,GS-SVM and PSO-SMV)based on faults sample database is presented and investigated.The result ind

    36、icates that the method has a high ability on mechanical fault identification for high voltage circuit breakers.Moreover,the accuracy and the efficiency of diagnosis fault method based on PSO-SVM is higher than the traditional SVM and GS-SVM models,the fault diagnosis rate is up to 94%under the curre

    37、nt samples.REFERENCES 1 Song hao,Cui Jingchun,Yuan Dalu,Operating analysis of high voltage circuit breaker in 1999-2003,Electrical Equipment 6(2)(2005)6-13.2 Huang J,Hu X,Geng X.An intelligent fault diagnosis method of high voltage circuit breaker based on improved EMD energy entropy and multi-class

    38、 support vector machine J.Electric Power Systems Research,2011,81(2):400-407.3 Huang N,Fang L,Cai G,et al.Mechanical Fault Diagnosis of High Voltage Circuit Breakers with Unknown Fault Type Using Hybrid Classifier Based on LMD and Time Segmentation Energy EntropyJ.Entropy,2016,18(9).4 Chen S,Niu W H

    39、,Li B S,et al.Based on EEMD and Multiclass Relevance Vector for High Voltage Circuit Breaker Mechanical Fault DiagnosisJ.Advanced Materials Research,2014,960-961:900-904.5 Zhong jianying,Liu yang,et al.Mechanical Fault Diagnosis Technology Research of High-voltage Circuit Breaker Based on the Vibrat

    40、ion Signal Characteristic.High Voltage Apparatus,2013,49(9):49-54.6 SHI Fei,LIN Xin,XU Jian-yuan.Mechanical characteristics and reliability of vacuum circuit breaker with spring operating mechanism J.Electrical Equipment,2005,6(6):79-81.7 CHANG Guang ZHANG Zhen-qian WANG Yi.Review on mechanical faul

    41、t diagnosis of high-voltage circuit breakers based on vibration diagnosis J.High Voltage Apparatus 2011,47(8):85-90.8 Cheng J,Yu D,Yu Y.The application of energy operator demodulation approach based on EMD in machinery fault diagnosis J.Mechanical Systems&Signal Processing 2007,21(2):668-677.9 Lopez

    42、-Roldan J,Pater R,Poirier S,et al.Development of non-intrusive monitoring for reactive switching of high voltage circuit breakerJ.International Journal of Electrical Power&Energy Systems,2014,61:219-228.10 NELLO C,TAYLOR J S.An introduction to support vector machines and other kernel-based learning

    43、methods M.New York:Cambridge University Press,2000.First A.Author MENG Fangang was born in Hei longjiang Province,November,1988.He was doctored at Wuhan University,major in mechanical engineering,researching on dynamics of mechanism.He published many articles:Modeling and Simulation of Flexible Tran

    44、smission Mechanism with Multiclearance Joints for Ultrahigh Voltage Circuit Breakers J.Shock and Vibration,2015.Numerical Modeling and Experimental Verification for High-Speed and Heavy-Load Planar Mechanism with Multiple Clearances J.Mathematical Problems in Engineering,2015.Efficiency optimization of the connecting rod drive mechanism of the ultra-high voltage circuit breaker Journal of Mechanical&Electrical Engineering,2014.2018 China International Conference on Electricity Distribution Tianjin,17-19 Sep.2018CICED2018 Paper No.201805180000004 Page4/4 2864

    展开阅读全文
    提示  道客多多所有资源均是用户自行上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作他用。
    关于本文
    本文标题:Research of Typical Mechanical Fault Diagnosis for HV Circuit Breaker Based on Vibration Signals.pdf
    链接地址:https://www.docduoduo.com/p-15718637.html
    关于我们 - 网站声明 - 网站地图 - 资源地图 - 友情链接 - 网站客服 - 联系我们

    道客多多用户QQ群:832276834  微博官方号:道客多多官方   知乎号:道客多多

    Copyright© 2025 道客多多 docduoduo.com 网站版权所有世界地图

    经营许可证编号:粤ICP备2021046453号    营业执照商标

    1.png 2.png 3.png 4.png 5.png 6.png 7.png 8.png 9.png 10.png



    收起
    展开