Vous êtes sur la page 1sur 5

5D1-1 Joint International Conference on Electric Vehicular Technology and Industrial,

Mechanical, Electrical and Chemical Engineering (ICEVT & IMECE) 2015

Design, Implementation, and Testing of Partial


Discharge Signal Pattern Recognition and Judgment
System Application Using Statistical Method
Roro Roudhotul Jannah1, Umar Khayam2
Electrical Power Engineering Study Program
School of Electrical Engineering and Informatics
Institut Teknologi Bandung
Bandung 40132, Indonesia
1
roro.jannah@students.itb.ac.id, 2umar@hv.ee.itb.ac.id

Abstract – Partial discharge is a problem that often affects high- neural network, will determine type and level of partial
voltage equipments. Early diagnosis system for partial discharge discharge. This result can also describe condition of insulation
can minimize the risk that caused by partial discharge. One of the in high voltage equipment and age prediction of equipment.
steps of partial discharge diagnosis is partial discharge signal Along with the development of technology, partial discharge
pattern recognition and judgement system that play a role in detecting system, especially in pattern recognition and
determining the type and level of partial discharge, and one of the
methods that can be used in this step is statistical method. In
judgment system, can be done more easily. Existance of
order to make data processing easier, partial discharge pattern computer can ease human in count and measure some complex
recognition and signal judgement system with statistical methods equations with some spesific formula of code. But, this method
can be done with the help of applications created using MATLAB is still difficult, because it makes user must insert the formula
software. The objective of this research is designing, or code manually. Therefore, we need an application
implementating, and testing the application of partial discharge specifically for pattern recognition and judgment system which
signal pattern recognition and judgment system, so the can be used easily (user-friendly) and integrated to ease partial
application can determine type and level of partial discharge. discharge detection process.
Application is created by using MATLAB software, and to This research is focused in design, implementation, and
determine level and type of partial discharge, this application can
make and gather database, measure partial discharge and testing of partial discharge signal pattern recognition and
statistical parameter, and compare it with database using judgment system application using statistical method. This
Kolmogorov-Smirnov test. This application consist of two main research is expected to provide convenience in diagnosis of
part: database making and partial discharge diagnosis. The result high voltage equipment earlier as one of maintenance and
of this research is application can determine level and type of inspection effort for high voltage equipment optimally. So that,
partial discharge with 83% accuracy. in future, both insulation failure caused by partial discharge
and financial loss caused by the failure can be minimized.
Keyword: high voltage, partial discharge, partial discharge signal
pattern recognition and judgment system, statistical method.
II. BASIC THEORY
Partial discharge is phenomenon happened in insulation
I. INTRODUCTION
material or insulation surface of high voltage equipment.
Partial discharge is a problem that often happened in
Based on IEC 60270-2000 standard, partial discharge is partly
insulator of high voltage equipment, such as transformator,
generator, motor, and power cable. This phenomenon can (locally) electric breakdownn which only connected insulation
between conductors partially. It caused by electrical field stress
cause insulation failure and influence age of high voltage
concentrated at one point, either inside insulation material or at
equipment [1]. So that, one of effort to maintain high voltage
equipment is partial discharge diagnosis to equipment’s insulation surface, which is caused by heterogeneity in
insulation shape, void or contaminant in insulation material,
insulation.
imperfection of insulation shape (for example, a crack in
One of some mechanism in partial discharge diagnosis in
insulator), etc.
high voltage equipment is pattern recognition and judgment
system. In this system, value of phase angle, amount of electric
discharge, and amount of partial discharge pulse from
detection and measurement system will be processed by
statistical method [2]–[5]. Another method in partial discharge
signal pattern recognition and judgment system is artificial
neural network method.
Result of partial discharge signal pattern recognition and
judgment system, either with statistical method or artificial Fig.1. Mechanism of partial discharge caused by void in solid insulation [6]

978-1-4673-8504-6/15/$31.00 ©2015 IEEE 314


5D1-1 Joint International Conference on Electric Vehicular Technology and Industrial,
Mechanical, Electrical and Chemical Engineering (ICEVT & IMECE) 2015
In transformer, partial discharge can causes primary and csv files and application will process them to get partial
secondary winding’s damage, and winding damage is the 34% discharge and statistical parameter. To complete the database,
reason of transformer failure [7]. In high voltage motor, partial we have to insert type and level of partial discharge before we
discharge can causes stator winding’s damage, which causes save the database. We can do these steps repeatedly until we
37% motor failure [8]. have enough database. After we have enough database, we can
start partial discharge diagnosis with the application. This
application reads csv (comma separated value) files from
partial discharge measurement in laboratory, then process them
to get partial discharge parameters (such as peak voltage,
amount, and electric charge of partial discharge) and statistical
parameters. These partial discharge and statistical parameters
are used for testing partial discharge type and level with
Kolmogorov-Smirnov test. Type of partial discharge is divided
Fig. 1. Distribution of electric field in needle-plate electrode [9] into two types: corona and oil, meanwhile level of partial
discharge is divided into four levels: level 0, level 1, level 2,
Partial discharge measurement is the important insulation and level 3. This application is made based on flowchart in
diagnosis method and be applicated in high voltage equipment figure 3 below.
like GIS (Gas Insulated Switchgear), XLPE cable, power
transformer, etc. The main purpose of insulation diagnosis in
high voltage equipment is to give information to operator
about deterioration degree of equipment [8].
Partial discharge diagnosis system has some mechanism, such
as:
1. Partial discharge detection and measurement. There are two
methods can be used in this step: electric method (based on
IEC 60270-2000 standard) with HFCT (high frequency
current transform) and detecting impedance and non-electric
method with acoustic, optic, and UHF (ultra high frequency)
method.
2. Partial discharge signal processing. There are some method
in this step, such as envelope method, Fast Fourier
Transform (FFT)/Inverse Fast Fourier Transform (IFFT),
wavelet transform, and filter (band pass filter, low pass filter,
high pass filter).
3. Pattern recognition and judgment system. In this step, partial
discharge pattern and parameter (pulse quantity, phase angle,
and electric charge) is analyzed. There are two methods in
pattern recognition and judgment system: statistical method
and artificial neural network. From this step, we can
determine damage level and type of partial discharge.
Statistical method is used in partial discharge signal pattern
recognition and judgment system. In this method, statistical
parameters, such as mean, variance, standard deviation,
skewness, and kurtosis, is used to identify type of partial
discharge. This identification is done by using statistical test.
One of statistical test that can be used is Kolmogorov-Smirnov
test, which can be determined by equation 1 below. [5]
Fig.3. Flowchart of application
KS1 = maxk|H1(k) — H2(k)| (1)

III. METHOD AND INSTRUMENTATION


To determine pattern recognition and judgment system
faster, we can make application in computer. This application
is made by using MATLAB software, which contains statistic
toolbox for statistical measurement and GUIDE (Graphical
User Interface Design Environment) application to make GUI
(Graphical User Interface)-based application. First, we have to
check whether the database is available or not. If database is
not available, we have to add the database by importing some

315
5D1-1 Joint International Conference on Electric Vehicular Technology and Industrial,
Mechanical, Electrical and Chemical Engineering (ICEVT & IMECE) 2015
IV. RESULTS AND ANALYSIS voltage source curve can be seen when the ‘plot kurva
There are three applications made with specific function: sinyal+sumber’ button is pressed. The magnitude of positive
main menu application, database making application, and and negative background noise is obtained by measuring
diagnosis application. The layout of these applications is maximum and minimum value in background noise csv file,
shown in figure 4, 5, and 6 below. and measurement can be started when the ‘mulai pengukuran’
button is pressed. The results shown in this application is peak
voltage, amount of positive and negative partial discharge,
positive and negative partial discharge electric charge, mean,
variance, standard deviation, skewness, kurtosis, histogram
curve, and partial discharge type and level.
Testing of this application is done with using file from
partial discharge measurement in Laboratorium Teknik
Tegangan dan Arus Tinggi ITB. Partial discharge
measurement uses needle-plate electrode, three kind of
detector – RC detector, High Frequency Current Transformer,
and loop antenna – and two medium – air and oil. From the
testing, we get the results shown in figure 7 until figure 12
Fig.2. Layout of main menu application
below.

In figure 4 above, there are two main button: ‘Buat


Database’ (Make Database) button and ‘Mulai Diagnosis PD’
(Start PD Diagnosis) button. These buttons are connected to
database making application and diagnosis application, so the
applications will be openend when the buttons are clicked.

Fig.7. Application measurement result of partial discharge corona type at level


1 file with HFCT detector

Fig.5. Layout of database making application

Fig.8. Application measurement result of partial discharge corona type at level


1 file with loop antenna detector

Fig.3. Layout of diagnosis application

Showed in figure 5 and figure above, in these application,


three kind of csv file is used as input, such as csv file from
partial discharge signal, csv file from voltage source, and csv
file from background noise. These file is inserted to
application manually. Both of partial discharge signal and

316
5D1-1 Joint International Conference on Electric Vehicular Technology and Industrial,
Mechanical, Electrical and Chemical Engineering (ICEVT & IMECE) 2015

Fig.9. Application measurement result of partial discharge corona type at level


1 file with RC detector Fig.13. Diagram diagnosis result of partial discharge level and type
From these result, shown that magnitude of partial discharge
parameter such as peak voltage, electric discharge, and amount
of partial discharge pulse is proportional with partial discharge
level. It shows that these parameter is related and being main
aspects in partial discharge signal pattern recognition and
judgment system.
Meanwhile, according to diagram 1, in determining level and
type of partial discharge, Kolmogorov Smirnov test gives a
quite good result with 83% accuracy. It indicates that
Kolmogorov-Smirnov test is provide quite good result for
partial discharge signal pattern recognition and judgment
system.
Fig.10. Application measurement result of partial discharge oil type at level 1
file with HFCTdetector
V. CONCLUSION
Conclusions from this research is aplication is succesfully
made and can be used to make database, measure partial
discharge and statistical parameter, and determine partial
discharge type and level. Using Kolmogorov-Smirnov test to
compare measurement of partial discharge and statistical
parameter with database, the accuracy of this application is
reach 83%. However, this application is not completely
perfect, so it need some repairment (fix existing features) and
upgrading (adding new features) to minimize error and
increase performance of application.

REFERENCES
Fig.11. Application measurement result of partial discharge oil type file at [1] J. James, S. V. Kulkarni, and B. R. Parekh, “Partial discharge in high
level with loop antenna detector voltage equipments-HV cable,” Proc. IEEE Int. Conf. Prop. Appl.
Dielectr. Mater., pp. 445–448, 2009.
[2] G. Wang, Y. Zhu, and L. Liu, “Fractional Feature Extraction and Pattern
Recognition of Transformer Partial Discharge Sources,” Proc. 2nd Int.
Conf. Comput. Sci. Electron. Eng. (ICCSEE 2013), no. Iccsee, pp. 657–
661, 2013.
[3] M. Mansor, P. S. Ghosh, A. Basri, A. Ghani, and U. T. Nasional, “PD
PATTERNS RECOGNITION IN XLPE CABLE UNDER VARIOUS
THERMAL CONDITIONS USING STATISTICAL TECHNIQUE,” vol.
5, no. November, pp. 171–181, 2010.
[4] a. a. Maqrashi, “Pattern recognition of partial discharges using MATLAB
tools,” 39th Int. Univ. Power Eng. Conf. 2004. UPEC 2004., vol. 1, 2004.
[5] B. Yazici, “Statistical pattern analysis of partial discharge measurements
for quality assessment of insulation systems in high voltage electrical
machinery,” 4th IEEE Int. Symp. Diagnostics Electr. Mach. Power
Fig.12. Application measurement result of partial discharge oil type at level 1 Electron. Drives, 2003. SDEMPED 2003., pp. 58–61, 2003.
file with RC detector [6] “http://en.wikipedia.org.” .
[7] M. Koch and M. Kruger, “A New Method for On-Line Monitoring of
From partial discharge diagnosis using application, result of Bushings and Partial Discharges of Power Transformer,” pp. 1205–1208,
application diagnosis is processed and shown in diagram 1 2012.
below. [8] L. Renforth, P. S. Hamer, D. Clark, S. Goodfellow, and R. Tower,
“CONTINUOUS , REMOTE ON-LINE PARTIAL DISCHARGE (

317
5D1-1 Joint International Conference on Electric Vehicular Technology and Industrial,
Mechanical, Electrical and Chemical Engineering (ICEVT & IMECE) 2015
OLPD ) MONITORING OF HV EX / ATEX MOTORS IN THE OIL
AND GAS INDUSTRY,” 2013.
[9] Suwarno, “Phase-resolved Properties and Simulation of Partial Discharges
in High Voltage Liquid Insulation,” no. 1, pp. 2–5, 2012.
[10] H.-C. Chang, F.-C. Gu, H.-C. Chen, and C.-C. Kuo, “Partial discharge
measurement and pattern recognition in gas insulated switchgear,” Light.
(APL), 2011 7th Asia-Pacific Int. Conf., pp. 459–462, 2011.

318

Vous aimerez peut-être aussi