COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

Application of vibration analysis in the condition monitoring of

electrical equipment
Chao Fu*, Junying Sun
Hebei Normal University, Shi Jiazhuang, China
Received 01 October 2014, www.cmnt.lv

Abstract

Condition monitoring of power equipment has become an important research direction of fault diagnosis, then vibration signal analysis
technology plays an important role. This paper introduces the research status of vibration signal analysis technique, the emphasis is placed
on the main research method in the condition monitoring, fault diagnosis of vibration signal. Through the comparison of several vibration
signal de-noising method, focuses on the analysis of the operation state of electric equipment through vibration signal.

Keywords: vibration signal, power equipment, fault diagnosis

1 Introduction 2 Vibration detection of generators and induction

motors
Operation state of electric equipment of power grid on the
system safety is very important. The previous domestic use The generators in the power system and induction motor are
off-line routine test and operation for regular maintenance large rotating equipment. The common characteristics of
of electrical equipment, this planned maintenance of fre- this kind of equipment is precision rotor of high speed, auto-
quent operation, reliability and excessive disassembly main- mation and complexity of structure are very high, associated
tenance will reduce the operating electrical equipment. In with each part of the system is also very close [2]. Malignant
recent years, state detection technology has been widely accidents in the electric power system due to the generator
attention of power equipment manufacturers and the power and induction motor due to a mechanical failure occurred
sector. Condition based maintenance to the field staff to pro- frequently, therefore improve its state detection level has
vide the equipment running information in real time, fault become the common concern of the domestic and interna-
prediction possible, thus assist engineer to make optimal tional problems. All the time since, the vibration detection
maintenance scheme. Vibration analysis is an additional are the main means of state of generator and induction motor
means of detection, through the installation of vibration sig- detection, to obtain diagnostic information equipment
nal acquisition process equipment in operation in one or through the analysis of vibration signal of rotor. This is
more vibration sensor device surface, feature information mainly because the rotor is an important component of the
extraction time or frequency domain, and then using the devices, the manufacturing cost is very high, so in the state
method of fault diagnosis of some use to evaluate equipment detection should be the primary object; secondly because
working state. many equipment failures will be reflected in the running
We can extract the feature information of equipment state of the rotor changes. The actual rotor vibration signal
operation status more and more rich from vibration signals is very complex, mainly for the axial of the rotor, radial and
with the development of signal processing technology [1]. torsional vibration [3]. These kinds of vibration often exist,
In addition, the vibration sensor has small size, light weight, therefore need to be respectively measured by different vib-
convenient installation, reliable work and low price, very ration sensor. The former two kinds of movement belongs
suitable for online detection or outdoor temporary testing to linear vibration, for their detection using acceleration sen-
occasions. Analysis of fault diagnosis using vibration for sor in general can be. As for the torsional vibration, is usual-
large rotating equipment such as generators and induction ly measured by the rotor speed fluctuation, so it belongs to
motors, there have been a lot of research results and the vibration velocity measurement. In addition, to detect
application examples. In the past ten years, the state of the induction motor stator winding fault, non-uniform air
vibration analysis technology in power transformer and gap and voltage disturbance by vibration analysis, in recent
circuit breaker detection has been continuously study. At years also have the relevant reports.
present, application of vibration analysis in these two kinds Since proposed in fault diagnosis of rotating machinery,
of power equipment is still in the exploratory stage, but the vibration signal analysis method of corresponding constant-
domestic and foreign scholars generally believe that vibra- ly change rapidly [4]. At present a comprehensive report of
tion analysis is a kind of new thinking of research potential, literature, including to analysis techniques: vibration signal
the author mainly introduces some research hotspots in the filtering and noise reduction; transient analysis (waterfall,
two fields and methods of analysis of the existing. Potter diagram, Nyqist diagram, order etc.). Time domain
analysis (envelope analysis, correlation analysis, statistical
analysis, analysis of orbit and so on). Frequency domain

*
Corresponding author e-mail: sjzfuchao@126.com

85
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

analysis based on Fourier transform (amplitude spectrum, model is established, and this method is used for online de-
power spectrum, spectrum, the envelope spectrum, cepst- tection of problems which should be considered not related
rum etc.). Time-frequency analysis (short Fourier, Wigner, to the. The late 90's Chris in twentieth Century, K et al pro-
Cabo transform). AR model analysis of vibration signal. posed ideas through the detection of the transformer oil tank
The development of frequency analysis and wavelet surface vibration signal to analyze and judge the core and
theory to further improve the technical level of the fault winding condition. But so far, the vibration detecting trans-
diagnosis of rotating equipment [5]. Especially the powerful former winding and iron core in the world only Russia had
ability of wavelet transform in signal time scale characteris- just entered the practical stage, the developed system has
tic analysis, feature extraction, singularity detection and noi- been used in more than 60 sets of large transformer. Trans-
se filtering area, so its applications in the field of fault former field results also confirmed that this method is appli-
diagnosis of rotating machinery are developing rapidly. The cable to a variety of models, fault diagnosis accuracy rate up
literature on Laplace wavelet to identify modal parameters to 80%~90%. The deficiency is not physical characteristics
of large hydraulic turbines, has obtained the ideal effect; of winding and core vibration are fully studied, such as win-
through the singularity index calculation of the wavelet ding vibration and winding pressing situation. The relation-
transform, which is used as features identification index of ship between the core vibration cause and effect factors, no-
the signal envelope [6]. On the application of wavelet trans- load voltage on core effect of vibration, the surface of the oil
form in fault diagnosis of rotating machinery, gives a good tank is different along with the vibration signal of the simi-
summary. In addition, other new methods and new theories larities and differences between, so the analysis also need to
are constantly attract people, such as signal blind source se- run the more experienced judgment. In addition, vibration
paration, fractal, chaos. In the fault diagnosis and decision signal analysis is limited to the via Fourier transform to ob-
making, statistical analysis, cluster analysis, neural network, tain the spectrum, in the form is too simple.
information fusion, fuzzy theory and other methods have China started late in this area, the main work is still in
been applied and the continuous development. In recent the stage of lab research. The literature [9] through the estab-
years, some scholars proposed the support vector machine, lishment of mathematical model of winding vibration, on
Bayesian network based on the concept of intelligent diag- the relationship between the winding vibration and core
nosis, further increase profits detection system adaptability. pressing condition are studied. Aiming at the vibration cause
core, studied the magnetostriction and its influence factors.
3 Vibration testing transformer The experimental method is: vibration of core tested in the
no-load voltage is different, the corresponding relationship
A. Vibration test of the core and winding. between the study of no-load voltage and core vibration and
The core and winding fault is the common fault in trans- the similarities and differences between the surface of the
former. Power transformer in the stable operation, magne- oil tank under different conditions of each phase of the core
tostriction vibration caused by iron core, winding will pro- vibration signal. Experiment showed that: the normal run-
duce vibration in the load current under the action of electric ning of the transformer and the transformer winding or core
field force. Closely related to compaction condition, vibra- fault vibration signal, the surface is very different. But only
tion and transformer winding and iron core transformer sur- from the changes of vibration signal in time domain, it is
face displacement and deformation state, so it can pass difficult to form a certain criterion to evaluate the extent of
through the surface of the oil tank of power transformer vib- the size of transformer faults, so the wavelet packet time-
ration detection to monitor the working status of the core frequency decomposition is applied to the processing of vib-
and winding [7]. ration signals, so as to get the energy feature vector of
From the beginning of twentieth Century 80 time meta- vibration signal, the feature vector as an important basis for
phase, American, Russia and other countries in the labora- judging whether there is fault of iron core. In addition, the
tory to detect winding and iron core vibration signal state of surface vibration signal of transformer at different load cur-
the preliminary study. The main transformer is in offline rent were measured, the relationship curve by fitting the load
condition, sinusoidal signal respectively with the low vol- current and the fundamental frequency component of vibra-
tage excitation signal and power at high voltage side of the tion signal, get a load current method to obtain core funda-
transformer, and then measuring the vibration signals of the mental frequency vibration signal components without no-
surface [8]. The research work in foreign countries mainly load operation can. Laboratory simulation tests show the
in the following aspects: validity of this method.
1) The relationship between the vibration signal and the B. Vibration detection of On-load tap changer.
size of the load current; On load tap changer (OLTC) is the only moving part in
2) Modeling parameters of load current, temperature and the transformer. Statistical analysis showed that, OLTC ano-
vibration signal of transformer considering; maly is another major cause of operation fault of transfor-
3) During normal operation, considering only the load mer. The OLTC switching operation contains a series of me-
modeling of current and the vibration signal of the trans- chanical motion events. Typical events (such as contact
former; collision) vibration signals generated, not only can sign the
4) Parameters on vibration signal (effective value, fre- contact points, may also contain the three-phase contact pe-
quency spectrum amplitude) research trend riod, whether level off, whether the contact surface is in pla-
5) Best installation location of the vibration sensor. ce and operating components jammed switch status infor-
The basic work is performed in the off-line situation, fo- mation. Foreign related detection device and field operation
cus on how to consider the vibration signal of transformer condition have been reported in the literature. Some trans-
former on-line detection device in the country also gradually

86
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

joined the vibration sensor, but both at home and abroad, the TABLE 1 The Euclidean distance between the energy entropy
impact of the relevant OLTC working state of vibration sig- The fault type Euclidean distance
nal detail, are also studied [10]. The fault signal A1 0.8463
For comparison, this method will first be accurate align- The fault signal A2 0.7895
ment of the envelope signal under different conditions on The fault signal A3 0.8537
the time axis. In fact, because of large amount of random The fault signal B1 1.2569
noise is always present signal captured in the field, so that The fault signal B2 1.2967
the application of this method is restricted by the. The event The fault signal B3 1.3023
start time (Literature) and morphological analysis for OLTC The fault signal C1 1.4879
The fault signal C2 1.5033
state detection, to determine the OLTC state by comparing The fault signal C3 1.4932
the vibration event change time and amplitude, and achieved
very good results, but there is no reports of the application Kang P et al developed on the basis of a set of OLTC
in the field. based online detection system vibration analysis. Using the
The literature [11] envelope analysis of vibration signals analysis software on the PC machine, can accurately iden-
of the normal state of OLTC, and then extract the average tify and OLTC: fault on the sudden change. 11/33kV trans-
large test as the state testing standards. Through comparing former substation has been installed in a plurality of the
the difference between vibration signal envelope to judge current system, the results of application show that, the sys-
whether the OLTC normal in the fault state. For comparison, tem can correctly reflect the running state of OLTC, such as
this method will first be accurate alignment of the envelope aging, contact spring failure and loose parts etc.
signal under different conditions on the time axis. In fact, In recent years, the literature [13] and the continuous wa-
because of large amount of random noise is always present velet transform (CWT) applied in the OLTC state recog-
signal captured in the field, so that the application of this nition of. Because the wavelet transform has good characte-
method is restricted by the. The event start time and mor- ristics in time domain, multi scale refinement, so it can ex-
phological analysis for OLTC state detection, to determine tract the local features of more information from vibration
the OLTC state by comparing the vibration event change ti- signals. The result of field operation shows that, the CWT
me and amplitude, and achieved very good results, but there method not only can make the correct judgment on OLTC
is no reports of the application in the field. fault contingencies, make the right evaluation but also on
The literature [12] based on Bengtsson C and others in the OLTC deterioration degree and remaining life, thus for-
the intelligent recognition of the state of the OLTC have mulates the reasonable maintenance plan.
made a lot of research work. First of all, smoothing the mul-
ti-scale analysis of vibration signal envelope using wavelet, 4 Vibration detecting of circuit breaker
and then calculate the various states of the signal envelope
autocorrelation function as state characteristic quantity, with According to the practice of statistics at home and abroad
the state characteristic quantity to train self-organizing map- for high voltage circuit breaker, the breaker in the accident,
ping neural network (S0M) as the basis of fault diagnosis. which belongs to the mechanical, causes most (fault).
During operation, calculated the minimum quantization er- Among them, agencies refused classification, closing and
ror of the vibration signal, through the comparison and the maloperation accounted for the vast majority of. Therefore,
normal value to judge the running state of OLTC. to improve the detection method of mechanical fault circuit
The energy entropy is a method of positioning system in a breaker, have very important sense to identify potential risks
certain state, it is a measure of the degree sequence is unknown. in advance.
For the vibration of high voltage circuit breaker, a variety of The fault diagnosis for high voltage circuit breaker has a
normal state is pulsating a standard normal state, the fault state good application prospect by mechanical vibration signal in
can be thought of as different mutations in this standard normal the process of operation. But due to mechanical vibration
state. First of all, the IMF envelope signal along the time axis signal can be obtained during its operation, so the number
are divided into N segments, and the signals of each segment of signals is obviously insufficient, the method of signal pro-
by segment energy time integral calculation: cessing and accuracy are demanding. Analysis and treat-
ment of almost all the circuit breaker vibration signal of
ti
Q(i )  
2
A(t ) dt . (1) research concentrates on vibration signal. Twentieth Cen-
ti 1
tury since 90 time, foreign made a lot of research work in
In the Equation i  1,2,...,N , ti 1, ti is the i segment this area, so far has identified a number of vibration signal
starting and ending point of information. The envelope of processing method is more practical, mainly concentrated in
the signal energy of each segment were normalized: two aspects.
In the signal generation, propagation and characteristics
Q(i ) analysis, literature [14] of experimental research and analy-
q (i )  N
. (2) sis of large amount of field data. Study on some basic prob-
 Q(i)
i 1
lems of circuit breaker mechanical vibration detection, such
as acceleration sensor installation, the choice of measure-
According to the energy entropy theory, Empirical mode ment point and the vibration signal repeatability, and adopts
(EMD) decomposition of signal x(t): the dynamic time warping (DTW) technology to achieve
fault diagnosis of circuit breaker. On this basis, analysis of
N
vibration for high voltage circuit breaker is developed, the
H    q (i ) lg q (i ) . (3) site has been applied to the 145KV SF6 of the circuit breaker
i 1
status detection, can be on the operating mechanism of bad

87
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

lubrication, improper installation, crank position moving Fourier analysis. The disadvantage is the large amount of
contact insulating rod length inappropriate common fault calculation, in determining the characteristic quantity also
make correct judgment. Combined with the parameters of has certain difficulty.
contact travel and coil current, but also can realize the com- The original signal SNR= 3
prehensive evaluation of the mechanical state of the breaker. 10 10

In feature extraction volume, a research team at the State

0 0
University of New York American a method is presented to
identify the vibration signal characteristics. Through several -10
500 1000 1500 2000
-10
500 1000 1500 2000
groups of data measuring circuit breaker in normal state and De-noising by heursure De-noising by rigrsure
10 10
the tested condition, calculation of single parameter RR as a
characteristic quantity of state recognition [15]. The re- 0 0
search team includes the main signal processing technology:
1) short time spectrum analysis; 2) correction of spectral -10
500 1000 1500 2000
-10
500 1000 1500 2000
analysis; 3) modified structure; 4) the time of accident; 5) De-noising by sqtwolog De-noising by minimaxi
10 10
envelope analysis; 6) morphological analysis. On this basis,
developed a portable breaker vibration signal detection sys- 0 0

tem (PCPDS). The literature [16] show that: this system has
-10 -10
been applied in the 115kV and above the oil circuit breaker 500 1000 1500 2000 500 1000 1500 2000

and SF6 on the circuit breaker, can successfully identify the

crank arm is shortened, excessive wear of the contact spring FIGURE 1 The distribution of Wigner joint time-frequency analysis of
compression, and buffer failure of mechanical fault. vibration signal
In recent years, to explore the correspondence between 3) The vibration signal with a group delay of exponential
mechanical vibration signal detection circuit breaker and decay oscillating description, 4 characteristics of each fre-
circuit breaker mechanical state in theory began to some quency component values are determined by curve fitting
foreign research institutions. Some new signal processing (amplitude, attenuation index, angular frequency and star-
method has been widely used in this field. The literature [17] ting time), compare to normal signal parameters and the.
the wavelet analysis and neural network are applied to the The study found that, when there are more random noise
vibration analysis of the circuit breaker, by choosing sui- signal, the calculation accuracy of the method is restricted,
table wavelet function to extract the feature vector represen- and later on the basis of mathematical analysis, the above
ting different states of the circuit breaker, and then as the model method has been improved, put forward with the sin-
input of BP neural network, so as to realize the recognition gular value decomposition (SVD) method, to a certain ex-
of different fault types. The method is used in the 66kV oil tent, improve the ability of restraining noise.
circuit breaker status detection, result shows that the correct 4) Research on the propagation process of vibration sig-
recognition rate reached 100%, significantly higher than the nal of circuit breaker in operation process, so the method is
traditional Fourier analysis and single neural network iden- proposed to process the vibration wave propagation is des-
tification results. cribed by using signal phase frequency characteristic and
At present, the circuit breaker status detection and fault phase frequency characteristic, through the analytical signal
diagnosis of machinery vibration signal is still in the explo- and vibration signal corresponding to the analysis method of
ratory stage based on domestic, the main content of this stu- the vibration event, get an identification in time mixed with
dy focused on the development of vibration signal acquisi- each other.
tion, vibration signal processing method and diagnosis ex- Mechanical failure of high voltage breakers not only can
pert system. In twentieth Century 80's and 90's, the Tsinghua cause vibration event time shift, but also caused the changes
University electrical engineering and applied electronic of some crest shape in time domain. The literature [17] is
technology system of circuit breaker mechanical vibration proposed for fault diagnosis of square wavelet singularity
signal had preliminary research and achieved certain results. index based on the river, can well describe the circuit brea-
On the basis of experiment and Study on the circuit breaker ker vibration signal time-frequency characteristic, in the
of the vibration signal, the vibration analysis is applied to open and close to the singularity index: local feature for vib-
high voltage circuit breaker online status detection, and de- ration signal envelope to characterize the signal, determine
veloped the corresponding state detection device (literature). the type ten fault changes according to the degree of the in-
In addition, the analysis method of vibration signal of circuit dex. At the same time, the fuzzy wavelet singularity index,.
breaker hall had detailed research, put forward some prac- Multi scale feed forward neural network are combined to get
tical algorithms, including: a new diagnosis method - fuzzy Singular neural network.
1) The amplitude method and the cross-correlation me- Tests prove that, the fault diagnosis accu-racy of this method
thod to extract appear different vibration events back when, can reach 90%, is a good way to improve the fault diagnosis
to judge the mechanical status of the circuit breaker accor- process of the emergence of distinguishing the inaccuracy
ding to each of the vibration event time changes, also can and uncertainty.
calculate the amplitude of each of the vibration signal, and Xi'an Jiao Tong University electrical equipment and
a small amount of calculation, clear physical meaning. electrical insulation National Key Laboratory in recent years
2) Application of Wigner distribution are combined on also started the research work in this field [18]. According
the time-frequency analysis of vibration signal. The analysis to the characteristics of vibration signal of circuit breaker
can reasonably describe the time-frequency characteristic of
the circuit breaker vibration signals more than the traditional

88
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

operation, studied several vibration signal processing me- the load operation, influence of the mechanism will be bet-
thod is practical to the bow, the circuit breaker vibration ween the contacts of electric power. For short circuit brea-
signal characteristic extraction method is the member shall. king under the influence of electric power to the vibration
1) The Euclidean distance between the concept of vector signal, the current lack of relevant analysis experiment. In
in mathematics based on Euclidean distance. This method is order to better vibration analysis method is extended to the
simple, small computation online application, should be strengthened in the future
2) Integral parameter method can effectively be subtle research work in this area, the field data as much as possible
changes in the vibration signal reflected, at the same time, to obtain the circuit breaker to be analyzed and compared,
eliminate the dispersion of the same state of multiple measu- in order to constantly improve the practicability of these me-
rement, characteristics of vibration signals can express mul- thods. In addition, the correspondding relation between the
tiple parameter combination. mechanical state of the breaker and feature, has been a
3) Signal entropy method. The concept of entropy in in- common challenge the circuit breaker status detection, in the
formation theory based on entropy calculation through the future research should be paid more attention.
signal, vibration signal, are established to describe vibration
signal under different conditions. 5 Conclusions
4) The short-time energy method. Advantages of short-
time energy method is effective to eliminate the effect of Extracting the characteristic parameters of suitable is always
noise signal, can extract the starting time of vibration events the most important issue facing the vibration analysis of po-
from the field complex electromagnetic environment, im- wer equipment. Wavelet transform can decompose signal into
prove the event time resolution. Analysis of the vibration the time scale domain, having excellent properties of time-
signal of the closing test method is applied to the vacuum frequency localization analysis, very suitable for non-sta-
circuit breaker, load accurately obtained under the three- tionary signal analysis therefore has become a common me-
phase contact closing time, and thus calculate the state under thod to analysis of various power equipment vibration signal.
different mechanical closing simultaneity and closing time On load tap vibration switches and circuit breakers are
state parameter. non-stationary signal in the transient of transformer, so not
5) Based on the algorithm of wavelet packets, the circuit many methods of rotating equipment to apply, to a certain
breaker operation the vibration signal is decomposed into extent also limits the application of vibration analysis and
independent band, variations of the signal energy in diffe- development in these two areas, the focus of future research
rent bands reflect the breaker machinery operating state is still looking for a suitable signal processing method.
change. Extracting frequency band energy as the feature In the aspect of fault diagnosis, artificial neural network
vector of main circuit breaker status monitoring, and to de- because of its powerful self-learning ability and the ability of
termine the variation range of the energy eigenvector of the parallel processing has been widely applied. However, in the
different states according to the test results, thus achieving a field of the transformer and circuit breaker status detection
simple classification of different circuit breaker status. Re- equipment status, lack of a large number of samples is the
peated experiments show that the method has higher pre- main factor that restricts the neural network practical training
cision state detection. method, therefore it is necessary to find a better.
In a variety of signal characteristic quantity analysis me-
thod that as the basis, studied the application of artificial Acknowledgments
neural network in circuit breaker status recognition, we pro-
pose an improved RBF neural network, the network can not Thank the Hebei Normal University Career Technical
only recognition of known state, but also through the neural College Department of Electronics and Province-Ministry
network self-learning ability, to recognize the new state, Joint Key Laboratory of Electromagnetic field and Elect-
thereby greatly improving the detection system identifi- rical Apparatus Reliability.
cation ability. Thanks to the project of Hebei higher school science and
Method for analysis of the vibration signal and related technology research projects (NO.Z2014011 and
tests are performed in the lab, but mostly in the offline simu- NO.2011221) on myself.
lation of different state of circuit breaker. Circuit breaker in

References

[1] Shengchang J 2009 Transformer winding and iron core characteristics [8] Hu X, Dai J, Ji Y 2001 The fault diagnosis of high voltage breakers
and its application in fault monitoring Xi'an Jiao Tong University based on wavelet singularity monitoring Chinese Society for Electrical
[2] Han Y, Song Y H 2012 IEEE Transactions on Power delivery 18(1) 4-13 Engineering Proceedings of the CSEE 21(5) 67-70
[3] Siddique A, Yadava G S, Singh B 2005 IEEE Transactions on Energy [9] Kang P, Birtwhistle D 2003 IEEE Transactions on Power Delivery
Conversion 20(1) 106-14 18(1) 78-84
[4] He Z 2012 Present situation and development of the application of the [10] Nechval K N, Nechval N A, Purgailis M, Rozevskis U 2013
wavelet technology Journal of Vibration Engineering 13 16-22 Predictive inferences for future order statistics under parametric
[5] Wang S, Ji C, Li Y 201 Application Research on transformer online uncertainty Computer modelling and new technologies 17 1
monitoring with vibration method Transformer 39(s) 73-6 [11] Nechval K N, Nechval N A, Purgailis M, Rozevskis U,
[6] Lavalle G C Failure Detection in Transformer Using Vibration Strelchonok V F, Moldovan M 2012 constructing inspection
Analysis USA MIT 186 strategies under uncertainty Computer modelling and new
[7] Lee D S S, Lithgow B J, Morrison R E 2010 IEEE Transactions on technologies 16 2
Power Delivery 18(2) 454-9 [12] Hu X, Sun L 2006 SF6 study on the method of circuit breaker on-line

89
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12D) 85-90 Fu Chao, Sun Junying

insulation monitoring Electric power automation Equipment 26(4) [15] Ni Y, Chen H 2013 Detection of Underwater Carrier-Free Pulse based
253-8 on Time-Frequency Analysis Journal of Networks 8(1) 205-12
[13] Wang Y., Pu Z, Wang X, et al 2005 Several calculation methods of [16] Jing Z H, Hua J, Yang. X N 2007 A new method for detecting negative
circuit breaker life on-line monitoring Journal of Shenyang SNR UWB Signal Communication Countermeasures 4 17–21
Agricultural Uinversity 36(1) 78-83 [17] Li X, Yao H 2013 Improved Signal Processing Algorithm Based on
[14] Li Bin 2005 Value and its online monitoring technology conversion, Wavelet Transform Journal of Multimedia 8(3) 226-32
the electrical life of circuit breakers limit High Voltage Electrical [18] Hollot C V, Misra V, Towsley D, Gong W 2002 IEEE Trans Autom
Apparatus 41(6) 43-9 Control 47(6) 945-59

Authors
Chao Fu, 1983 August, Shijiazhuang City, China

Current position, grades: lecturer in Hebei Normal University, Shi Jiazhuang, China.
University studies: BSc in school of Electrical Engineering of Tianjin Hebei University of Technology in 2006, M.S. degree in 2010.
Scientific interest: signal processing and electrical technology.

Junying Sun, 1978 April, XingTai City, China

Current position, grades: associate professor in Hebei Normal University, Shi Jiazhuang, China.
University studies: BSc in school of North China Electric Power University in 2004.
Scientific interest: internet of things and embedded system.

90