automated classification system for polymeric insulation
TRANSCRIPT
Automated Classification System for Polymeric
Insulation Surface Condition
A. N. Ramani and A. R. Abdullah Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka (UTeM), Malaysia
Email: {anisniza, abdulr}@utem.edu.my
N. Norddin
Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka (UTeM), Malaysia
Email: [email protected]
N. Q. Z. Abidin and A. Aman Universiti Teknikal Malaysia Melaka (UTeM), Malaysia
Abstract—Polymeric insulation is commonly used for high
voltage engineering as they are light, easy to fabricate, and
have good dielectric properties. Ageing factors affect the long
term performance of insulating material. Leakage current is
broadly accepted as tools in the determination of surface
condition and level of its severity. Hence, an automated
monitoring system is needed to reduce diagnostic time and
ensure quality of insulators performance. Due to the
limitation fast Fourier transforms (FFT), this research used
spectrogram as time frequency distribution technique that
represents the leakage current signals in the joint time
frequency domains which appropriate to analyze the leakage
current signals that consist of multi-frequency components
and magnitude variations. This technique extract relevant
information from leakage current (LC) signal and then
leakage current (LC) parameters are estimated to identify its
characteristics. Surface condition on High Density
Polyethylene (HDPE) and Polypropylene (PP) are
investigated. The classification of material surface state could
be determined instantaneously using the percentage of total
waveform distortion. Thus, the outcome of this study shows
that the system is very appropriate and reliable to be
implemented for leakage current online monitoring system.
I. INTRODUCTION
In high voltage engineering or its applications,
insulation is the most important part to prevent the flow of
current to undesired paths. Polymeric insulation is widely
accepted [1] compared to conventional ceramic insulation
due to its several advantages as mentioned by Hackam [2].
However, polymeric insulation still has certain limitations
such as difficulty in detecting defective insulation.
Hydrophobicity measurement techniques are mostly
studies on material surface characterization and chemical
Manuscript received December 15, 2014, revised April 30, 2015.
investigation. Besides that, one of the key indicators
widely accepted to determine performance of polymeric
insulation whether in service or accelerated ageing
laboratory test is by investigating the leakage current (LC)
signal. LC signal provides information of polymeric
insulation surface condition and the pollution severity [3].
LC measurement is popular technique because it can be
monitored either online or offline [4], [5]. Measurement of
the peak current, accumulated charge [6] and discharge
duration has been used by some researchers to provide
information on degradation.
However, onwards literatures show, LC harmonic
component analysis will give better information on
insulation surface [7]. As an example, A. H. El Hag, et al.
[8] examined low harmonic components of LC as a
diagnostic tool to study ageing and surface condition.
While N. Bashir, et al. [9] and Kordkheili [10] have
verified the ratios of lower harmonics components as the
major factor in determining the surface condition.
Nevertheless, FFT does not provide temporal
information and is not appropriate for non-stationary signal
[11]. Instead, the analysis using time frequency distribution
(TFD) such as is used. These TFD linear time frequency
distribution approach resolve the limitation of FFT. In this
recent work, TFD is employed to analyze LC components
to identify surface condition of polymer materials.
Fernando and Gubanski [7] in their work had been
separated the surface events into several stages. The stages
are early ageing period (EAP), transition period (TP) and
late ageing period (LAP). These stages determine the level
of surface condition as well as its contamination levels. At
early ageing period (EAP) only capacitive LC existed
which indicated the insulator in hydrophobic condition.
Then, during transition period (TP) the surface of insulator
becomes hydrophilic and LC significantly changes to
resistive condition. Finally, during late ageing period (LAP)
the LC level become higher and become completely
resistive. At this moment, the surface erosion took place as
well as produce non linear LC wave shape.
International Journal of Electrical Energy, Vol. 3, No. 2, June 2015
©2015 International Journal of Electrical Energy 105doi: 10.12720/ijoee.3.2.105-109
Index Terms—time frequency distribution, spectrogram,
insulation, tracking and erosion
Tracking and erosion test complying with BS EN
60587-2007 conducted on High Density Polyethylene
(HDPE) and Polypropylene (PP) as the polymeric
insulation material. A computer-based online LC
monitoring of the insulators surface is found to be more
practical due to the long period of time taken in conducting
the laboratory test. Online performance monitoring can
benefits engineering industries by confirming if the
materials is performing as desired by the user. Monitoring
system gives instantaneously information on LC
parameters such its harmonic components. Normal
performance analyses are usually conducted in offline
mode which the LC data is saved into other files to be
analyzed later.
This paper describes the online LC monitoring system
that classifies LC signals in time domain, frequency
spectrum, time-frequency representation as analysis
method, LC parameters such as root mean square (RMS)
value, fundamental value, total harmonic distortion (THD),
total non-harmonic distortion (TnHD), and total wave
distortion (TWD). These parameters instantaneously is
estimated in order to develop and monitoring state
condition of the materials insulating condition. The
percentage of total waveform distortion TWD% is applied
as rule based value for surface classification purpose. The
results of LC analysis the online monitoring system is
compared with other software that also used to monitor and
analyzed the LC performance. This comparison is certainty
that the online monitoring system is reliable and can be
used in the industry.
II. EXPERIMENTAL SETUP
A. Inclined-Plane Test
Inclined-Plane Test (IPT) is normally used to evaluate
the tracking and erosion resistance of insulating materials
and recommended by BS EN 60587 [12]. In the test
procedure, since it is done in the same place every day,
there are three parameters that are assumed constant;
humidity, pressure, and temperature. As stated in the
standard, the contaminant conductivity level is set to
2.533S/m measured by Hanna Dist 4 conductivity meter
and its flow rate at 0.6ml/min. The system for evaluating
the surface tracking and erosion test along with the online
LC monitoring system is shown in Fig. 1 and Fig. 2. The
whole system is placed inside a separated room to avoid
any outside noise or disturbance that could affect the test
measurement.
Figure 1. Schematic diagram of incline-plane tracking (IPT) test
Figure 2. Schematic diagram of incline-plane tracking (IPT) test
B. Material
The major factors that resolve to the failure of the
insulation are dielectric and electrical field which are the
key properties of insulating materials [6]. High Density
Polyethylene (HDPE) and Polypropylene (PP) are
non-ceramic or polymeric composite refer to hydrophobic
materials. These substances selected based on nature
dielectric strength and dielectric constant that is nearly
identical [7].
High Density Polyethylene (HDPE) is among polymer
that is good to be made as insulating material. Numerous
studies were done towards High Density Polyethylene
(HDPE) [8]. According to Lynn, high density polyethylene
has the capability as insulating material. This is as that
stated inside [9]. Fig. 3 shows discharge condition for
HDPE and during the discharge the spark and smoke exist.
After the test, there are small erosion exist on the surface of
the sample.
Figure 3. Discharge on high density polyethylene (HDPE)
(a) (b)
Figure 4. Discharge on polypropylene (PP)
Polypropylene (PP) has good balance between electrical
and mechanical properties [13]. It also has high dielectric
International Journal of Electrical Energy, Vol. 3, No. 2, June 2015
©2015 International Journal of Electrical Energy 106
strength [14]. Polypropylene has resistance to heat
distortion, have excellent electrical properties and fatigue
strength, chemically inert, relatively inexpensive, but it
have poor resistance to UV light [15]. Their melting
temperature generally range from 170-230°C and due to its
low moisture absorption, it is not necessary to dry its
compound before moulding [16]-[18]. In Fig. 4 the
discharge condition for Polypropylene and the condition
after the test on sample surface are shown.
C. Leakage Current Monitoring System
An LC data acquisition system is developed to analyze
the LC flowing on the sample surface under laboratory
conditions. It consists of three parts; measuring unit, signal
conditioning, and analog-to-digital (ADC) with personal
computer. The graphical results of the LC signal are
developed using Visual Basic (VB) software in the
personal computer. Visual Basic (VB) applies events
driven programming language and allows users to click on
a certain object randomly to make some action. More
specifically, the system is designed for interactive user
interface. The data is entered and then the result is viewed
on the computer screen. The front panel of the graphical
user interface (GUI) of the developed program is shown in
Fig. 5.
Figure 5. Front panel of leakage current monitoring system
III. TIME FREQUENCY DISTRIBUTION
Linear time frequency distribution (TFD) provides the
information about time variation as well as frequency
spectrum of leakage current simultaneously which using
spectrogram. From the time frequency representation
(TFR), characteristics of the signals can be calculated and
used as input for signals classification. The signal
characteristics are total harmonic distortion, total non
harmonic distortion, total waveform distortion and current
root mean square (IRMS).
Spectrogram is one of the time-frequency
representations (TFR) that represents a three-dimensional
of the signal energy with respect to time and frequency.
Spectrogram is squared magnitude of the short time
Fourier transforms (STFT). This technique roughly reflects
how frequency content changes over time. The
spectrogram defined as
2
2( , ) ( ) ( ) j fxS t f h w t e d
(1)
where h(τ) is the input signal and w(t) is the window
observation window. In this study, Hanning window is
selected because of its lower peak side lope [14] which is
narrow effect on other frequencies around fundamental
value (50Hz in this study) and other frequency
components.
IV. PARAMETER ESTIMATION
Parameters of the signal estimated from the time
frequency distribution to recognize the signal information
in time.
A. Instantaneous RMS Current
The instantaneous RMS current is
max
0
[ ] ( , )
f
rms xI t S t f df (2)
B. Instantaneous RMS Fundamental Current (I1rms)
Instantaneous RMS fundamental current I1rms(t) is
defined as the RMS current at power system frequency and
can be calculated as:
(3)
where f1 is the fundamental frequency that corresponds to
the power system frequency and Δf is the bandwidth which
is set to 50Hz.
C. Total Harmonic Distortion (THD)
THD is the relative signal energy present at
non-fundamental frequencies and written as:
2,2
1
( )( )
( )
H
h rmshTHD
rms
I tI t
I t
(4)
where Ih,RMS(t) is RMS harmonic current and H is the
highest measured harmonic component.
D. Total Non-Harmonic Distortion (TnHD)
Non-harmonics or inter harmonic are not multiple
integer signal components frequency of the power system
frequency. Therefore, TnHD is referred as distinguishing
between non-harmonic and noise, and is calculated as:
2 2,0
1
( ) ( )( )
( )
H
rms h rmshTnHD
rms
I t I tI t
I t
(5)
E. Total Waveform Distortion (TWD)
TWD consists of harmonic distortion and non-harmonic
distortion. It can define as:
2 2( ) ( ) ( )TWD THD TnHDI t I t I t
(6)
International Journal of Electrical Energy, Vol. 3, No. 2, June 2015
©2015 International Journal of Electrical Energy 107
1
1 1
( ) 2 ( , )
,2 2
hi
lo
f
rms x
f
hi lo
I t S t f df
f ff f f f
V. RESULT
The effectiveness and reliability of the online
monitoring system is tested by conducting an experiment
according to the IPT method. During experiment, it is
observed that at low voltages level between 0-0.5kV, there
is no erosion and arcing activities occurred on the
insulation material. The capacitive state from both
materials shows the same capacitive sinusoidal shape
depicts in Fig. 6. All patterns amplitudes of RMS current
have been normalized by dividing by the RMS value of the
normal signal (resistive signal). This is because at resistive
signal only fundamental frequency exists.
Figure 6. Capacitive pattern
Figure 7. Resistive pattern
At this resistive state only fundamental frequency 50Hz
exist. This condition appeared at voltage stresses 1kV until
1.5kV. The average instantaneous RMS current is
approximately 1.0pu and it has a smallest TWD value that
is below 3% because of low distortion on the signal. In Fig.
7, resistive pattern is shown.
At 1.8kV, waveform of the non linear pattern with
distortion exists depict in Fig. 8 and the sounds of arcing
without spark are produced. At 2.5kV until 4kV arcing
sound and spark exist, this also created nonlinear pattern.
The RMS current pu for this state is higher than 1.0pu.
THD, TnHD and TWD value around 100%. On non linear
pattern there are ‘rest’ times requisite for a full
hydrophobic condition to take place again. It is the
temporary loss of hydrophobicity during discharge.
Hydrophobicity transfer or recovery occurs because of
absorption of low molecular weight to pollution layer. Arc
current dissipated energy large enough to cause rupture in
the polymer chains and is transferred to the surface directly.
The severities because of degradation on the surface
become higher [17], [19]. The non linear pattern during
discharge on polymeric insulation material also indicated
that erosion is occurred on the surface of the insulator.
Figure 8. Nonlinear pattern
VI. SURFACE CONDITION CLASSIFICATION
In order to find out the relation between RMS current pu
and total waveform distortion (TWD) to the surface
condition of the insulator signal classification have been
conducted. Analysis results were made based on leakage
current parameter estimation obtained from spectrogram.
The rules based are used to classify the surface condition
event instanteously are summaries in Table I.
TABLE I. RULED BASED ON SURFACE CONDITION CLASSIFICATION
Pattern Parameter High Density
Polyethylene Polypropylene Condition
Capacitive IRMS <1.0 pu <1.0 pu No arching Harmonic
exist at 500-600Hz TWD 50-3% 50-3%
Resistive
IRMS 1.0 pu 1.0 pu Arching sound and
only fundamental
frequency, 50Hz exist TWD <3% <3%
Nonlinear
IRMS varies
1.0≤x≥1.0 varies Dry band with high
arcing which lead to
erosion and harmonic
exist at 3rd, 150Hz TWD
Percentage
varies around
50 %-200%
Percentage
varies around
50 %-200%
International Journal of Electrical Energy, Vol. 3, No. 2, June 2015
©2015 International Journal of Electrical Energy 108
VII. CONCLUSION
This study using new approach to presents analysis of
polymeric insulation leakage current patterns using
parameter estimation to identify the surface condition of
the insulator. In conclusion, it was found that late aging
period state with unsymmetrical discharge pattern on
polymeric insulation material can be use as erosion
indicator occurred on the surface of the insulator. Instead
of using leakage current amplitude, ruled based obtained
from spectrogram can be used to determine the surface
condition of the insulator instantaneously. Moreover, the
classification of surface condition of insulators can be
automatically done by using this new system.
ACKNOWLEDGMENT
The authors gratefully acknowledge the Ministry of
Higher Education (MOHE) and Universiti Teknikal
Malaysia Melaka (UTeM) for providing the research grant
RACE/F1/TK3 /UTEM/10.
REFERENCES
[1] M. H. Ahmad, H. Ahmad, N. Bashir, Y. Z. Arief, R. Kurnianto, and
F. Yusof, “New statistical ranking of tree inception voltage
distribution of silicone rubber and epoxy resin under AC voltage
excitation,” International Review of Electrical Engineering, vol. 6,
no. 4, pp. 1768-1774, Aug. 2011.
[2] R. Hackam, “Outdoor HV composite polymeric insulators,” IEEE
Transactions on Dielectrics and Electrical Insulation, vol. 6, no, 5,
pp. 557-585, 1999.
[3] N. Bashir and H. Ahmad, “Insulator leakage current studies in
artificial climatic chambers using dimensional analysis,”
International Review of Electrical Engineering, vol. 4, pp.
1433-1439, 2009.
[4] D. Pylarinos, K. Siderakis, E. Thalassinakis, E. Pyrgioti, I. Vitellas,
and S. L. David, “Online applicable techniques to evaluate field
leakage current waveforms,” Electric Power Systems Research, vol.
84, pp. 65-71, 2012.
[5] T. Zuo, T. Liu, K. Chen, and X. Hu, “On-Line monitoring system of
insulator leakage current based on ARM,” in Proc. 10th IEEE Int.
Conference on Industrial Informatics (INDIN), 2012, pp. 75-79.
[6] J. W. Chang and R. S. Gorur, “Surface recovery of silicone rubber
used for HV outdoor insulation,” IEEE Transactions on Dielectrics
and Electrical Insulation, vol. 1, pp. 1039-1046, 1994.
[7] M. A. R. M. Fernando and S. M. Gubanski, “Leakage currents on
non-ceramic insulators and materials,” IEEE Transactions on
Dielectrics and Electrical Insulation, vol. 6, pp. 660-667, 1999.
[8] A. H. El-Hag, “A new technique to detect dry-band arcing,” IEEE
Transactions on Power Delivery, vol. 20, pp. 1202-1203, 2005.
[9] N. Bashir and H. Ahmad, “Odd harmonics and third to fifth
harmonic ratios of leakage currents as diagnostic tools to study the
ageing of glass insulators,” IEEE Transactions on Dielectrics and
Electrical Insulation, vol. 17, pp. 819-832, 2010.
[10] H. H. Kordkheili, H. Abravesh, M. Tabasi, M. Dakhem, and M. M.
Abravesh, “Determining the probability of flashover occurrence in
composite insulators by using leakage current harmonic
components,” IEEE Transactions on Dielectrics and Electrical
Insulation, vol. 17, pp. 502-512, 2010.
[11] H. C. Lin, C. H. Chen, and L. Y. Liu, “Harmonics and
interharmonics measurement using group-harmonic energy
distribution minimizing algorithm,” Engineering Letters, vol. 19,
pp. 265-270, 2011.
[12] O. Rioul and M. Vetterli, “Wavelets and signal processing,” IEEE
Signal Processing Magazine, vol. 8, pp. 14-38, 1991.
[13] Electrical Insulating Materials Used under Severe Ambient
Conditions - Test Method for Evaluating Resistance to Tracking
and Erosion, BS EN 60857:2007.
[14] H. I. S. Jayasundara, W. P. S. Sudarshani, and M. A. R. M.
Fernando, “Leakage current patterns on high voltage insulators:
analysis on frequency and time-frequency domains,” in Proc. IEEE
Region 10 and the Third International Conference on Industrial
and Information Systems, 2008, pp. 1-6.
[15] A. Aman, M. M. Yaacob, M. A. Alsaedi, and K. A. Ibrahim,
“Polymeric composite based on waste material for high voltage
outdoor application,” International Journal of Electrical Power
and Energy Systems, vol. 45, pp. 346-352, 2013.
[16]
[17] W. D. Callister, Fundamentals of Materials Science and
Engineering: An Integrated Approach, John Wiley & Sons, 2005.
[18] S. Naidu and V. Kamaraju, High Voltage Engineering 4E, Tata
McGraw Hill Education Private Limited, 2009.
[19] J. Kim, W. Song, et al., “Leakage current monitoring and outdoor
degradation of silicone rubber,” IEEE Transactions on Dielectrics
and Electrical Insulation, vol. 8, pp. 1108-1115, 2001.
A. N. Ramani was born in Kuala Lumpur,
Malaysia in 1981. She received his B.Eng. from
University Teknikal Malaysia Melaka (UTeM)
in 2006 and Master Degree from University of
Technology Malaysia in 2009.
1979. He received his B. Eng., Master, PhD
Degree from University of Technology
Malaysia in 2001, 2004 and 2011 in Electrical
Engineering and Digital Signal Processing. He
is currently a Senior Lecturer and Coordinator
of Center of Excellent Robotic and Industrial
Automation (CeRiA) in Universiti Teknikal
Malaysia Melaka (UTeM).
Malaysia in 1987. She received her Diploma
and B.Eng from Universiti Teknikal Malaysia
Melaka (UTeM) in 2008 and 2011, respectively,
in electronic power and drives. She is now
pursuing her Master Degree in High Voltage
engineering in Universiti Teknikal Malaysia
Melaka (UTeM).
in Perak, Malaysia, on November 25, 1985. She
received Diploma and B. Eng. degrees with
honours in electrical engineering from the
University Teknikal Malaysia Melaka (UTeM)
in 2006 and 2011, respectively. Currently, she
is master’s degree student researcher in the field
of high voltage engineering.
Malaysia in 1973. He received his B.Eng. and
Master Degree from University of Technology
Malaysia in 1998 and 2007, respectively, in
Control & Instrumentation and High Voltage
engineering. He is now pursuing his PhD in
same university in High Voltage engineering.
He is lecturer in Universiti Teknikal Malaysia
Melaka.
International Journal of Electrical Energy, Vol. 3, No. 2, June 2015
©2015 International Journal of Electrical Energy 109
M. S. Naidu and V. Kamaraju, High Voltage Engineering,
McGraw-Hill, 1999.
A. R Abdullah was born in Kedah, Malaysia in
Nurbahirah Binti Norddin was born in Kedah,
Nur Qamarina Binti Zainal Abidin was born
Aminudin Bin Aman was born in Melaka,