HVDC Faults Classification by Lab Color based Monitoring System
ZIYAD S. ALMAJALI
Department of Electrical Engineering
Faculty of Engineering
Mutah University
Mutah - Karak,
JORDAN
Abstract: - This paper presents a new method for faults classification of HVDC system by analyzing the
currents and voltages of the system under the Zero-Direct-Quadrature (DQ0) coordinates using Lab color
chromatic based monitoring approach. Repeated for several healthy and faulty scenarios in DC side and AC
side as well, The DQ0 transformed signals are evaluated for each case, Continuous chromatic monitoring is
applied on all signals. A chromatic Lab mapping of short time-slot windows result in providing an early fault
detection. The preliminary results show high accuracy in fault type classification as well with short processing
time and without the need for any pretraining requirement.
Key-Words: - HVDC, Converter, Rectifier, Fault, Lab color, Chromatic monitoring, Classification.
Received: May 12, 2021. Revised: February 21, 2022. Accepted: March 25, 2022. Published: April 29, 2022.
1 Introduction
The adoption of the HVDC system appears because
of its significant benefits in connecting different
systems, especially those with different frequencies
or in the integration of wind and solar power
resources to the main AC grid. Scientific researchers
continue to work on improving various properties of
the system with a special focus on reliability, but
because faults are inevitable; protection researchers
are concerned with minimizing and avoiding losses,
and their first important task here is to have correct
diagnosis of the fault.[1]
The high voltage direct current (HVDC) system
consists of various integrated components, each
with its own function that cannot be ignored, such as
rectifiers whose functions is to convert AC to DC
and inverters to convert DC to AC, in addition to, of
course, the long transmission lines extending
between the terminals. In view of the multiplicity of
parts, a variety of faults can occur in the system in
terms of type, i.e., faults on AC side or DC side in
addition to the possibilities of different fault
locations on the transmission line or the fault
resistance whose value cannot be predicted.[2] For
the purpose of rapid detection and fast clearance of
faults its necessary and important to identify its
characteristics such as its type, direction and
location.[3]
Before choosing the method and deciding which
actions to take, it is necessary to have accurate and
fast detection of the event.
Protection is usually applied to each component
of power system with extra care for important vital
parts. Given the importance of the HVDC system,
researchers have devoted large efforts to diagnose
and classify its various faults to ensure the
continuity of power delivery and to avoid power
outages and blackouts.[4]
Several methods have been reported in the
literature, many are based on the traditional
travelling wave resulting from the system events
and its characteristics[5-8], and limitations of some
methods such as high processing requirements and
expensive equipment cost [9,10] are reported, and
some of the approaches are reported as efficient
only for far fault due to noise interference with its
detection abilities.[11,12]
Modern approaches have also been reported,
based on waveform [13,14], artificial neural
network[15] and fuzzy-logic support vector machine
[16-18] and some of which are based on a
combination of recent approaches such as wavelet
transform fusion with neural networks.[19]
The continuous motivation for innovative
approaches development or old method
modifications is the significant error reported in
traditional methods or the requirement of
comprehensive training in some modern approaches.
The Chromatic approach is one of the modern
monitoring methods that have proven their
efficiency and distinction in monitoring in many
different fields, including recent applications in
electrical power systems [20]. Due to its effective
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
186
Volume 17, 2022
ability, it was chosen as a tool for the classification
step in this proposed approach. Chromatic
monitoring has various forms, such as the HSL, Lab
and xyz.
The chromatic processing stage follows The
DQ0 Transform of a monitored HVDC system
currents and voltage waveforms. The Lab color
chromatic processing algorithm was adopted in this
work, and its implementation and general procedure
are explained in the next section after the principles
of DQ0 coordinate transformation are introduced
and described. While classification results and
discussion are presented in section (3).
2 Method Description
2.1 Overall Procedure
Novel approach for classifying the types of errors
that occur in the HVDC system by applying
chromatic monitoring is presented in this paper. The
flowchart in Figure 1 illustrates the steps of the
approach. The figure shows a few steps before the
direct application of Lab chromatic monitoring, and
these steps will be covered in the following
subsections, such as the description of the simulated
system in section (2.2). The DQ0 coordinates will
be given in subsection (2.3), then samples of the
results are given in subsection (2.4). The final step
is the implementation of chromatic monitoring
which includes the implementation of the RGB
processor, the Lab color transformation and the final
mapping step, the details of the algorithm and its
various steps that define its parameters are
explained in subsection (2.5).
2.2 Simulated System
To prepare a set of data for various potential fault
conditions in the HVDC system, a bipolar 12-pulse
transmission system is considered and simulated
using MATLAB® software and its associated
Simulink® and Simpowersystem® toolboxes. The
standard model of HVDC system is selected from
MATLAB libraries is utilized in this work and
shown in Figure 2.
Start
Input signals
System currents and
voltages collection
DQ0 Coordinate
RGB processors
Lab transformation
Mapping
Detection and
classification
Fig. 1: Method Overall Procedure flowchart
Fig. 2: The simulated system
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
187
Volume 17, 2022
The selected model represent part of a typical
HVDC system, A 300 km long DC line with 0.5 H
smoothing reactors is utilized to transmit power
from a 500kv, 60 Hz 5000MVA convertor station to
a 345kv, 50 Hz, 10000MVA station.
Both sides of the AC are represented as infinite
sources, and simple parallel R-L branch are used to
represent source impedances. AC filters are
included to limit the harmonics to the level required
by the network. The used HVDC transmission Line
is of distributed model and its parameters are as
given in Table 1.
Table 1. Transmission line parameter
Resistance Ω/km
0.015
Inductance mH/km
0.0792
Capacitance nf/km
14.4
Line length km
300
For the purpose of testing the proposed method,
data is collected from the system in various fault
conditions such as single line to ground fault, line to
line to line, line to line to ground fault, and the
three-phase fault in the AC side in addition to the
DC side faults. The simulations involve various
fault parameter variation such as the fault location
and fault resistance for the DC faults and various
fault types for the AC faults with different fault
resistance and in addition to the normal system
operation. The results will then be used to verify the
effectiveness and robustness of the method
algorithm under different operating conditions.
2.3 Zero-Direct-Quadrature (DQ0)
Signal transformations have useful and proven uses
in research in many areas. It would transform
ordinary system signals that carry a lot of invisible
meanings or information into signals that are
different in form but equal in content, but in a way
that is more capable of extracting meanings and
information from them in a clearer way and with the
least amount of trouble.
Sometimes it is necessary to pass the results to a
second transformation and processing stage with
different equations to get better results. The
complexity and repetition of the transformation and
mapping process depends on the complexity of the
observed phenomenon. Therefore, researchers may
have to pass the signal through several
transformations before the information begins to
become clear. The multiple transformations based
technique forms the idea of the proposed method in
this research. Two types of transformations have
been proposed; the first type is the DQ0
transformation, its algorithm and equations will be
illustrated in this section. While the second
transformation is the chromatic transformation
which will be explained in detail in subsection 2.5.
Starting by the DQ0 transformation, the used
algorithm can generate a new set; a rotating two-
axis reference waveforms from the original three
phase voltage waveforms as given in equations (1)
and (2).[21]
The same equations are used but for the calculation
of Iq and Id as given in equations (3) and (4). With
its simple calculation the algorithm can be used to
reverse the transformation successfully.[21]
2.4 Simulation Results
For the purpose of testing the proposed method, data
is collected from the system in various fault
conditions. Figure 3 illustrates a sample of the
recorded currents, voltages waveforms for a single
line to ground fault encountered in the AC side of
the simulated HVDC system while Figure 4
illustrates another sample for the recorded currents,
voltages waveforms for a line to line to ground fault
waveforms.
Equations (5) and (6) are used for the dq-axis
magnitude calculations of the system voltage and
the current and Figures 5 and 6 illustrate a sample of
the dq-axis magnitudes of both current and voltage
waveform variations respectively for a single line to
ground fault case.
(5)
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
188
Volume 17, 2022
(6)
Fig. 3: Single line to ground fault
Fig. 4: Line to line to ground fault
Fig. 5: The dq-axis magnitudes of current waveform
changes for a single line to ground fault case
Fig. 6: The dq-axis magnitudes of voltage waveform
changes for a single line to ground fault case
Same procedure is followed for all fault
conditions as well as normal condition including
various fault parameter variation such as the fault
location and fault resistance for the DC faults and
various fault types for the AC faults with different
fault resistance as well.
2.5 Chromatic Monitoring
The used method in this work is based on exploring
the nature of the event whose type and location
leave fingerprints on the generated signals. In
summary, signals from different events show great
similarity, however, each event, due to its location
and/or type, has the effect of altering the
configuration of the system and thus leaves a trace
that may not be clearly visible. So, to show this,
transformations can reveal a hidden feature that
enables different classes of events to be
distinguished.
The initial idea of chromatic monitoring came
from the working principle of the human eye; the
sensitive observation element in humans’ system.
The eye contains powerful sensors whose purpose is
to transmit the complete visual image for analysis in
the brain, which includes a complete and accurate
analysis of its spectrum, colors and contents.
In human vision system, covering the entire
visible color spectrum is performed by Color-
distinctive cells through a defined set of overlapping
filters. Inspired by the same idea, it is possible to
create a similar working principle system, in Figure
7, a set of filters is applied named R, G and B, may
be similar in shape, but differ in their coverage
range with the possibility of overlapping to monitor
the signal P. For full coverage monitoring, the signal
range P determines the beginning and end of the
coverage range by all the different filters combined.
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
189
Volume 17, 2022
The response of each processing filter depends
on the details and magnitude of the monitored signal
in the covered range by its covering processing
filter. The outcomes of three signal filters are given
by equations 7, 8 and 9:
R=∫R(ϕ)P(ϕ)dϕ (7)
G=∫G(ϕ)P(ϕ)dϕ (8)
B=∫B(ϕ)P(ϕ)dϕ (9)
R(ϕ), G(ϕ) and B(ϕ) are the filters profiles and ϕ
is the monitored signals domain, and P(ϕ) is the
monitored signal. The gaussian profile shape of the
filters in Figure 7 can be replaced by other shapes
such as the triangular shape and the number of
processors can be altered as well depending on the
application and the monitored signal. In this work,
three processors with triangular shapes were
selected.
Fig. 7: Three gaussian filters coverage of monitored
signal domain
For implementation of the adopted Lab color
chromatic processing algorithm is in this work, a
group of equations 4-6 are used for processing of
the three R, G, and B processors outcomes.
Equations 10, 11 and 12 produce three relative
magnitudes (L, a, and b) respectively [20]
L=116(G/Gn)*1/3-16 (10)
a=500[(R/Rn)*1/3-(G/Gn)*1/3)] (11)
b=200[(G/Gn)*1/3-(B/Bn)*1/3)] (12)
The Rn, Gn, Bn in color science are the processor
outputs when addressing the illumination source
directly. In this work, these terms will be used for
normalization purpose.
The outcomes of the transformation process
illustrated by previous equations are used for the
detection and classification stage. Some parameters
can be used to characterize and determine the
occurrence of faults. Other parameters can be used
for the classification stage.
3 Monitoring Results
The overall procedure presented in subsection (2.1)
and as shown in the flowchart of Figure 1 is applied
on the computed DQ0 transform signals waveforms
under different condition for AC faults, DC faults
cases in addition to the healthy condition of the
simulated HVDC system shown in Figure 2.
The proposed scheme has been thoroughly
tested. Test cases are repeated under different
normal operating in addition to faulty conditions
with different fault resistance and diverse locations
for the purpose of sensitivity investigation.
Variation of the L, a and b parameters are evaluated
for both voltage and current waveform as well for
all the different cases.
Figure 8 shows the relationship plot for the L
parameter from the Vq waveform on the x axis, the
L parameter from the Iq waveform on the y axis
versus the b parameter from the Iq waveform on the
z axis. Preliminary analysis of this variation for
these selected chromatic monitoring transformation
parameters shows successful indication of the
condition occurrence detection upon fault incidence.
Mapping between the selected parameters can
also serve the purpose of distinguishing between
different conditions as the results identify a
significant defining characteristic of well-defined
clear boundaries for various clusters defining
different conditions.
Fig. 8: LVq with LIq and bIq parameter plot for all
cases
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
190
Volume 17, 2022
Although Figure 8 shows a possibility of
ambiguate between the normal condition and the
DC fault but there is a clear defining characteristic
of well-defined clear boundaries between these
conditions, and this is illustrated by zooming and as
shown in Figure 9. But and to increase the proposed
system sensitivity, further investigation by
utilization of diverse color transformation is
encouraged with monitoring of different domain
such as the frequency and this is left as a future
work suggestion.
Fig. 9: LVq with LIq and bIq parameter plot for the
Normal operation and the DC fault case
4 Conclusion
The preliminary results obtained with Lab color
monitoring of the Zero-Direct-Quadrature DQ0
waveforms from a HVDC system under various
faulty conditions have been presented. The proposed
method has the potential to successfully distinguish
between different cases without the need for prior
training.
This paper presents a new method for classifying
HVDC system faults. The analyses are applied on
currents and voltages of the system under (DQ0)
coordinates using Lab color chromatic based
monitoring approach. Repeated for several healthy
and faulty scenarios in DC side and AC side as well,
The DQ0 transformed signals are evaluated for each
case, Continuous chromatic monitoring is applied
on all signals. A chromatic Lab mapping of short
time-slot windows result in providing an early fault
detection. Initial results show high accuracy in fault
type classification as well with short processing
time.
References:
[1] H. D. Al-Majali, “Abnormal Harmonics
Generated by Modified HVDC
Converter”, Electric Power Components and
Systems, 35:6, 639-654, 2007
[2] Chan-Ki Kim, Vijay K. Sood, Gil-Soo Jang,
Seong-Joo Lim, Seok-Jin Lee “HVDC
Transmission: Power Conversion Applications
in Power Systems” 2009, Wiley-IEEE Press,
ISBN 13:9780470822951
[3] L.L.Lai, F.Ndeh-Che, K.S.Swarup,
H.S.Chandrasekharaiah “Fault Diagnosis in
HVDC Systems with Neural Networks” IFAC
Proceedings Volumes, Volume 26, Issue 2,
Part 5, July 1993, Pages 827-830
[4] P. Giradkar and A. Deshmukh, “Fault
Detecting in High Voltage Transmission Line
Based on Protection Scheme” International
Journal of Electrical, Electronics and Data
Communication (IJEEDC) , pp. 82-85 June
2015
[5] Iraide López, Jon Andreu, Salvador Ceballos,
Iñigo Martínez de Alegría, Iñigo Kortabarria,
“Review of wave energy technologies and the
necessary power-equipment”, Renewable and
Sustainable Energy Reviews, Volume 27,
2013, Pages 413-434,
[6] Nanayakkara, O.M.K.K.; Rajapakse, A.D.;
Wachal, R. Location of DC line faults in
conventional HVdc systems with segments of
cables and overhead lines using terminal
measurements. IEEE Trans. Power Deliv.
2012, 27, 279–288.
[7] Nanayakkara, O.M.K.K.; Rajapakse, A.D.;
Wachal, R. Traveling wave based line fault
location in star connected multi terminal HVdc
systems. IEEE Trans. Power Deliv. 2012, 27,
2286–2294.
[8] Farshad, M.; Sadeh, J. A novel fault location
method for HVdc transmission lines based on
similarity measure of voltage signals. IEEE
Trans. Power Deliv. 2013, 28, 2483–2490.
[9] Glik, K.; Kowalik, R.; Rasolomampionona, D.;
Anwar, S. Travelling wave fault location in
power transmission systems: An overview. J.
Electr. Syst. 2011, 7, 287–296.
[10] L. Tang, “Control and protection of multi-
terminal DC transmission systems based on
voltage-source converters,” 2003, PhD thesis,
McGill University
[11] Ji, Z.; Zeng, Q.; Liao, J.; Wu, Q.H. A novel
mathematical morphology filter for the
accurate fault location in power transmission
lines. In Proceedings of the TENCON 2009—
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
191
Volume 17, 2022
2009 IEEE Region 10 Conference, Singapore,
23–26 November 2009; pp. 1–6.
[12] X. Liu, A. H. Osman and O. P. Malik, "Real-
Time Implementation of a Hybrid Protection
Scheme for Bipolar HVDC Line Using
FPGA," in IEEE Transactions on Power
Delivery, vol. 26, no. 1, pp. 101-108, Jan. 2011
[13] F. Unal and S. Ekici, “A fault location
technique for HVDC transmission lines using
extreme learning machines,” ICSG 2017 - 5th
Int. Istanbul Smart Grids Cities Congr. Fair,
pp. 125–129, Jun. 2017, doi:
10.1109/SGCF.2017.7947616.
[14] Y. M. Yeap, N. Geddada and A. Ukil,
“Analysis and validation of wavelet transform
based DC fault detection in HVDC system”,
Applied Soft Computing, Volume 61, 2017,
Pages 17-29,
[15] C. Venkatesh, P. Venugopal Rao, “Wavelet-
ANN based Classification of HVDC Converter
Faults”, International Conference on Power
Systems (ICPS), 2016 IEEE, pp. 1-5.
[16] J. Y. Wu, S. Lan, S. J. Xiao, and Y. Bin Yuan,
“Single Pole-to-Ground Fault Location System
for MMC-HVDC Transmission Lines Based
on Active Pulse and CEEMDAN,” IEEE
Access, vol. 9, pp. 42226–42235, 2021.
[17] B. Paily, S. Kumaravel, M. Basu and M.
Conlon, "Fault analysis of VSC HVDC
systems using fuzzy logic," 2015 IEEE
International Conference on Signal Processing,
Informatics, Communication and Energy
Systems (SPICES), 2015, pp. 1-5.
[18] J. P. Keshri and H. Tiwari, “Fault
Classification in VSC-HVDC Transmission
System using Machine Learning Approach,”
2019 8th Int. Conf. Power Syst. Transit.
Towar. Sustain. Smart Flex. Grids, ICPS 2019,
Dec. 2019
[19] S. A. Gafoor and P. V. Ramana Rao, “Wavelet
based fault detection, classification and
location in transmission lines,” First Int. Power
Energy Conf. (PECon 2006) Proc., pp. 114–
118, 2006, doi: 10.1109/PECON.2006.346630.
[20] G. R. Jones, J. W. Spencer “Advanced
Chromatic Monitoring” ISBN 9780367409470,
2020 by CRC Press
[21] C. J. O’Rourke, M. M. Qasim, M. R. Overlin, J.
L. Kirtley Jr. "A Geometric Interpretation of
Reference Frames and Transformations: dq0,
Clarke, and Park." IEEE Transactions on
Energy Conversion 34, 4 (December 2019):
2070 - 2083.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on SYSTEMS and CONTROL
DOI: 10.37394/23203.2022.17.21
Ziyad S. Almajali
E-ISSN: 2224-2856
192
Volume 17, 2022
