Sie sind auf Seite 1von 7

1

Abstract--While the importance of wide area measurement


system (WAMS) consisting of phasor measurement units
(PMUs) is recognized, there are still unidentified applications
that can be addressed with the advance of technology.
Traditionally, power system operators diagnosed the fault point
through the status of protective relays and circuit breakers of
the protection system. In large blackouts, however, the
malfunction of protection system has itself often been among the
suspects of the disaster. This paper proposes an alternative fault
diagnosis approach independent of the function of protection
system, utilizing PMU data and bus impedance matrix (Zbus).
First, the proposed method diagnoses the fault area, and next
the fault line is diagnosed. Fault location along the diagnosed
line is also estimated by wide-area measurements. The proposed
method is applied to the IEEE 118-bus test system where the
results indicate successful diagnosis of the faulted lines
throughout the grid as well as wide area fault location along the
faulted line.

Index Terms-- Fault Diagnosis, Power System, Phasor
Measurement Unit (PMU), Synchrophasor, Wide Area
Measurement System (WAMS).
I. INTRODUCTION
IDE area measurement system (WAMS) utilizing
fast and accurately extracted synchrophasors by phasor
measurement units (PMUs) is considered as the basis for
smart transmission grid [1]. Although the importance of
synchrophasors in power systems is recognized [2], there are
still undisclosed applications that can be applied by the
development of the required infrastructure.
The current practice in power system fault diagnosis [3]-
[5] is to use the status data of protective relays (PR) and
circuit breakers (CB) so as to infer the faulted section based
on past experiences. Traditionally, the data associated with
the function of the protection system are provided by
supervisory control and data acquisition (SCADA) system,
which are communicated to the energy management system
(EMS) every 2-10 seconds [6]. However, in large blackouts, it
is typical of PRs and/or CBs to misoperate, which is leading
to the expansion of outage area [7]. Therefore, while there is
an urgent need to pinpoint the fault point for system

The authors are with the Center of Excellence in Power System Management
and Control, Department of Electrical Engineering, Sharif University of
Technology (SUT), Tehran 11365-11155 Iran. (e-mail: asalehi@ee.sharif.edu;
amranjbar@sharif.edu).
restoration [8], fault diagnosis grows problematic or even may
be made incorrectly [9]. Therefore, another approach for fault
diagnosis independent of the protection system performance
should be pursued.
Although current PMU applications and standard concern
steady-state or quasi-stationary conditions of the power
system, huge effort has been made to employ PMUs with
high-sampling features for transient phenomena in the
network [10]. In fact, any relay or digital fault recorder (DFR)
capable of producing synchronized measurements by the
global positioning system (GPS) signal is considered as a
PMU device [11].
This paper presents a novel application of synchronized
measurements provided by WAMS for fault diagnosis of a
power transmission grid. While the need for smart
transmission grid fault diagnosis by synchrophasors is
emphasized [12], there are a few methods addressing this
problem. This paper introduces a novel analytic method for
wide area fault diagnosis and location in transmission grids.
The proposed method uses the online data of network
topology and state, which are provided by SCADA and EMS,
as well as real-time synchronized measurements of fault
currents, which are provided by PMUs, to diagnose the fault
point in the grid by circuit rules. The proposed measurement-
based fault diagnosis approach is expected to be a new
application of WAMS in smart transmission grids, as the
communication technology is penetrating power systems
increasingly.
II. PROBLEM FORMULATION
The problem addressed in this paper is determining the
fault section in a power network using only network topology
and state information, which can be acquired by SCADA
system in an hourly basis, synchrophasors data of voltage and
current signals during the fault, which are available by
WAMS, and parameter data of transmission lines,
transformers and generators. The proposed scheme for
transmission grid fault diagnosis is depicted in Fig. 1. This
approach is completely different from that of conventional
methods where the status of CBs and PRs determine the
faulted part of the network while no measurement is utilized.
Following a fault in the system, voltage and current
phasors at different buses and branches vary according to the
fault location and type. The main idea is to fit the measured
Transmission Grid Fault Diagnosis by
Wide Area Measurement System
A. Salehi Dobakhshari, Graduate Student Member, IEEE, and A. M. Ranjbar
W
2012 3rd IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Berlin
978-1-4673-2597-4/12/$31.00 2012 IEEE
2














Fig. 1. Proposed measurement-based wide area fault diagnosis scheme.

phasors by PMUs to the analytically produced phasors for
faults at different points of the network. In this way, the fault
candidate point which best matches the measured phasors
indicates the fault section in the network, as follows:


( , ) ,

T
V
T
I
F
F
Res F T =

meas
meas
f ( ) V
I f ( )
(1)
* *
( , ) arg min{ ( ( , )}),
F
T
F T Res F T

e
e
= (2)
where V
meas
and I
meas
are respectively vectors of voltage and
current phasors measured by several PMUs across the
network. The set of candidate fault points and various fault
types are represented by and , respectively.
T
V
F f ( ) and
T
I
F f ( ) are complex vector functions which relate the
expected voltage and current phasors after fault of type T at
point F to the measured voltage and current phasors, i.e.
V
meas
and I
meas
, respectively. F* is the faulted point diagnosed
by fitting the observed phasors to the calculated phasors in
(1). Fig. 2 shows a two-dimensional depiction of (1), where
analytically calculated vectors are compared with the
measured vector. It must be noted, however, that vectors in
(1) are M-dimensional, where M is the number of
measurements.
Online
(Updated Periodically)
Real Time
(During Fault Incident)
Database
EMS WAMS


Fault Current Phasors
by PMUs
Proposed Fault Diagnosis Method
Fault Diagnosis
Grid Impedance Data Generators Impedance Data
System State
Network Topology

3









Fig. 2. Depiction of the proposed fault diagnosis method by matching the
observed phasors to the calculated phasors.

For each of the fault types,
T
V
F f ( ) and
T
I
F f ( ) can be
formed as follows, comprising of phase voltages and currents,
respectively.

A,
1
B,
1
C,
1
A,
B,
C,
( ) ,
dur
dur
dur
V
dur
N
dur
N
dur
N
V
V
V
F
V
V
V
(
(
(
(
(
=
(
(
(
(
(

f

A,
12
B,
12
C,
12
A,
( 1)
B,
( 1)
C,
( 1)
( ) ,
dur
dur
dur
I
dur
N N
dur
N N
dur
N N
I
I
I
F
I
I
I

(
(
(
(
(
=
(
(
(
(
(

f
(3)
where
A,dur
i
V and
A,dur
jk
I are the expected voltage phasor of
phase A of bus i and current phasor of phase A of branch j-k
calculated during the fault, respectively; N is the number of
system buses. For fault type T at bus F of the network,
T
V
F f ( )
and
T
I
F f ( ) can be calculated based on the bus impedance
matrix (Z
bus
). Symmetrical components are also utilized for
unbalanced fault analysis.

III. TRANSMISSION LINE FAULT DIAGNOSIS
Operators of power systems are mostly confronted by the
faults along transmission lines. If fault diagnosis has to be
done for fault point m along a transmission line between
buses p and q, it is sufficient to define a virtual bus at point
m. Next, point m can be added to the set of fault candidates
() so that if a fault occurs at point m along line p-q, then the
proposed method can pinpoint the fault point by (1)-(2).
Since this approach for transmission line fault diagnosis
needs infinite fault point candidates along any line of the
system, a three-stage method is developed to firstly diagnose
the faulted area in the network and next determine the faulted
line, and finally pinpoint the fault point along the faulted
line.
A. Diagnosis of Fault Area
In the first stage, includes the midpoint of transmission
lines as fault candidate points. It is expected that if a fault
occurs along transmission line p-q, then fault candidate m at






Fig. 3. Similar fault currents for the actual fault point (f) and faulted line
midpoint (m) in Stage I.











Fig. 4. Determining Fault Location by defining virtual buses in Stage II.

the midpoint of this line shall have the least fitting residual,
as shown in Fig 3.
In this way, the fault line may be determined. However, to
account for uncertain factors leading to error in this method,
in a conservative manner, all the lines connected to the
diagnosed line in the first stage, will also be examined in the
second stage to diagnose the fault line.
B. Diagnosis of Fault Line and Fault Location
In the second stage, fault area determined by the first stage
is explored to diagnose the faulted line in the fault area.
This is achieved by examining the line which midpoint
had the least fitting residual as well as all the lines connected
to it. Fault diagnosis is achieved by defining virtual buses
along the suspected lines so that these buses account for fault
point candidates in this stage (See Fig. 4). The line with the
least electrical distance (i.e., impedance) among fault
candidate lines is selected and a virtual bus is defined at the
middle of it. Next, all of the remaining suspected lines are
sectionalized by new virtual buses so that the length of each
subsection does not exceed l/2, where l is the length of the
shortest line which has been divided into two sections. As
such, it is guaranteed that unequal impedances of neighboring
lines would not result in wrong diagnosis.
Once more, by applying (1) with including the virtual
buses defined along the suspected transmission lines as
explained, in case of a fault along one of the suspected lines,
the faulted line as well as fault location is pinpointed by the
virtual bus which is the output of (2).
The proposed two-stage transmission line fault diagnosis
algorithm is summarized as following steps.
Step 1- Define a virtual bus in the midpoint of each line of
the network.
Step 2- Run bus fault diagnosis algorithm as explained in
Sections II, where includes the defined virtual buses.
m f
p q
I
pq

I
pq
I
qp

m
8

m
9

s
q
m
1

w
p
I
qp

m
11

m
10

f
m
5
m
4
m
3
m
2
m
6
m
7

t
, 3
( )
a
SLG
V I
f F
V
meas
,I
meas

*
,
( )
a
SLG
V I
f F
, 1
( )
a b
DLG
V I
f F


, 2
( )
Sym
V I
f F
, 3
( )
a b
LL
V I
f F


4
Step 3- Find F
*
from (2) and name the line at which F
*
is
located as L
*
.
Step 4- Determine the set of suspected transmission lines
as {L
*
, L
1
, L
2
, , L
S
}, i.e. L
*
together with the lines connected
to it.
Step 5- Pick the suspected line L
S
with the least electrical
distance (l) and define a virtual bus m
1
at its midpoint.
Step 6- Define more virtual buses as {m
2
, m
3
, , m
NV
} at
equal distances along suspected transmission lines other than
L
S
so that the length of each section created by these virtual
buses does not exceed l/2.
Step 7- Run (1) and (2) with ={ m
1
, m
2
, , m
NV
} and
determine the fault point m
*
and the fault type T
*
.
Step 8- Diagnose the fault line as the one including the
virtual bus m* obtained in Step 7. Moreover, the actual fault
location f along the diagnosed line can be approximated by
m*.
C. Fault Location by Fine-Tuning
To fine-tune the fault location estimation in stage II, more
virtual buses along the diagnosed fault line can be defined.
Fine-tuning can be achieved as follows.
Step 1- Define two virtual buses n
1
and n
2
at either side of
the virtual bus approximated as the fault point (m
*
) in Stage
II.
Step 2- Run bus fault diagnosis algorithm (1)-(2) with
including m
*
, n
1
and n
2
and designate the diagnosed fault
point as n
*
.
Step 3- If n
*
=n
1
or n
*
=n
2
go to Step 4, else terminate the
process and locate the fault point as m
*
.
Step 4- Go to Step 1, but replace the approximate fault
point m
*
with n
*
.
It must be noted that the fault location may be further
improved by defining more virtual buses along the diagnosed
line. As such, fault point estimation can be fine-tuned at the
expense of more computation time.

IV. SIMULATION RESULTS
In order to demonstrate the effectiveness of the proposed
fault diagnosis scheme, it is applied to the IEEE 118-bus test
systems with 177 transmission lines, 9 transmission
transformers and 19 generators. Dynamic data for the
generators are selected according to the typical data given in
[13]. The bus impedance matrix (Z
bus
) for positive, negative
and zero sequences are constructed for this system in
MATLAB

environment, where generators are modeled with


a voltage source behind the subtransient reactance. In this
way, expected voltage and current phasors emerging
following a fault may be analytically calculated [14].
To simulate the fault, the network has been modeled in
DIgSILENT Power Factory software as the power system
simulator, considering the distributed model of transmission
lines and the frequency dependency of their parameters [15].

TABLE I. PMU BUSES IN THE IEEE 118-BUS TEST SYSTEM
3, 9, 11, 12, 17, 21, 23, 28, 34, 37, 41, 45, 49, 53, 56, 62,
71, 75, 77, 80, 85, 86, 90, 94, 102, 105, 110, 115
PMU Buses
28 No. of PMUs

TABLE II. TRANSMISSION LINE FAULT DIAGNOSIS USING THE PROPOSED
ALGORITHM FOR ACTUAL FAULT AT LINE 30-38
2nd Stage 1st Stage
Fitting
Residual
Virtual Bus
Distance
(%)
Virtual
Buses Along
Lines
Fitting
Residual
Mid-Line
Virtual Bus
Rank
6.46 33%
30-38
(SLG
a
)
1.87 30-38 (SLG
a
) 1
1.86 66%
11.03 25%
38-65
(SLG
a
)
19.04 38-65 (SLG
A
) 2
19.04 50% 26.44 26-30 (SLG
A
) 3
27.45 75%
39.25 8-30 (SLG
A
) 4
33.01 25%
26-30
(SLG
a
)
26.44 50%
57.89 38-65 (DLG
CA
) 5
22.47 75%
39.25 50%
8-30
(SLG
a
)
58.87 38-65 (DLG
AB
) 6

The sampling rate of PMUs is 4800 Hz (96 samples per
cycle). The input signals to PMUs are first filtered by a
second-order analog (Butterworth) anti-aliasing filter with a
cut-off frequency of 350 Hz. Next, a digital mimic filter is
utilized to remove decaying dc offset. Finally a full-cycle
Fourier transform is applied to extract the fundamental
component of the input signal in MATLAB

environment.
Table I shows the WAMS configuration considered for the
studied power system. This configuration consists of 28
PMUs providing complete observability for this system [16].
The a-phase voltage angle at bus 49 is considered as the
reference phase angle for PMUs measurements.
A single-line-to-ground (SLG) fault at line 30-38 at the
distance of 60% of the line length near bus 30 is simulated to
illustrate the application of the proposed two-stage method.
Figs. 5 and 6 show the fault current flowing through line 37-
40 in time and phasor domains, respectively.
As shown in Table II, at the first stage, the midpoint of
line 30-38 is found to have the significantly least fitting
residual. In the second stage, transmission lines 8-30, 26-30
and 30-38 are examined as fault line candidates, based on the
proposed two-stage algorithm. It can be seen that not only is
the fault on line 30-38 diagnosed, the location of the fault is
also estimated as the 66% of the line length near bus 30.
To compare the accuracy of the diagnosis for different
situations, an accuracy index (AI) is defined as the ratio of the
least residual error associated with virtual buses along the
fault candidate lines in stage II other than the actual fault line
to the residual error of actual fault point, as follows.


{ ( )}
{ ( )}
j
l
i
bus j along lines other than the diagnosed line l
bus i along diagnosed line l
Res F
AI
Res F
Min
Min
=
(4)

5
0.96 0.98 1 1.02 1.04 1.06 1.08 1.1
-0.4
-0.2
0
0.2
0.4
0.6
Time [Sec.]
I
3
3
-
3
7

(
t
)

[
p
u
]


Input to PMU
Output of
Anti-Aliasing Filter

Fig. 5. Fault current through line 37-33 observed by the PMU at bus 37 after a
fault at line 30-38.


Fig. 6. The phase angle and magnitude of fault current phasor through line 37-33
after a fault at line 30-38.

TABLE III. FAULT DIAGNOSIS AND LOCATION FOR ACTUAL FAULT AT LINE 30-38 WITH DIFFERENT FAULT INCEPTION ANGLES

Fault Inception
Angle (Deg)
Actual Fault
Distance (p.u,)
0 30 60 90 120 150 180

FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
0.1 0.1094 11.964 0.1094 11.763 0.1100 11.413 0.1126 11.564 0.1146 12.038 0.1250 11.471 0.1094 11.974
0.3 0.3333 7.4078 0.3333 7.3033 0.3333 7.4312 0.3333 7.2056 0.3333 7.4312 0.3333 7.0391 0.3333 7.4135
0.5 0.5 12.209 0.5 12.05 0.5 11.784 0.5 11.883 0.5 12.291 0.5 11.55 0.5 12.22
0.7 0.666 6.7816 0.666 6.7005 0.666 6.5443 0.666 6.5278 0.666 6.6054 0.666 6.0131 0.666 6.787
0.9 0.8542 7.1241 0.8542 7.0427 0.8333 6.679 0.8333 6.7884 0.8333 7.0386 0.8333 6.7077 0.8542 7.13

TABLE IV. FAULT DIAGNOSIS AND LOCATION FOR ACTUAL FAULT AT LINE 30-38 WITH DIFFERENT SAMPLING RATES FOR PMUS

PMUs Sampling Rate
(samples/cycle)
Actual Fault
Distance (p.u,)
24 48 96 192 384

FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
FL
[pu]
AI
0.1 0.125 3.2196 0.125 3.5011 0.1094 11.964 0.1094 15.171 0.1094 17.079
0.3 0.333 3.8331 0.333 5.1693 0.333 7.4078 0.333 9.3332 0.333 10.49
0.5 0.5 5.9223 0.5 8.2101 0.5 12.209 0.5 15.825 0.5 18.109
0.7 0.666 3.545 0.666 4.704 0.666 6.7816 0.666 8.6355 0.666 9.8081
0.9 0.8333 3.691 0.8333 4.8874 0.8542 7.1241 0.8542 9.0457 0.8594 10.324

TABLE V. FAULT DIAGNOSIS AND LOCATION FOR ACTUAL FAULT AT LINE 30-38 WITH DIFFERENT TYPES OF BALANCED AND UNBALANCED FAULTS

Fault
Type

Actual Fault
Distance (p.u.)
Single Line to Ground Line to Line Double Line to Ground 3 Phase Balanced
Diagnosed Line
(Fault Type)
FL
[pu]
AI
Diagnosed Line
(Fault Type)
FL
[pu]
AI
Diagnosed Line
(Fault Type)
FL
[pu]
AI
Diagnosed Line
(Fault Type)
FL
[pu]
AI
0.1 30-38 (SLG
A
) 0.1094 11.964 30-38 (LL
BC
) 0.1198 7.256 30-38 (DLG
BC
) 0.1198 10.807 30-38 (3PH) 0.1250 19.409
0.3 30-38 (SLG
A
) 0.3333 7.4078 30-38 (LL
BC
) 0.3333 6.62 30-38 (DLG
BC
) 0.3333 8.5063 30-38 (3PH) 0.3333 15.506
0.5 30-38 (SLG
A
) 0.5 12.209 30-38 (LL
BC
) 0.5000 13.364 30-38 (DLG
BC
) 0.5000 16.63 30-38 (3PH) 0.5000 32.228
0.7 30-38 (SLG
A
) 0.666 6.7816 30-38 (LL
BC
) 0.6667 7.5845 30-38 (DLG
BC
) 0.6667 9.1386 30-38 (3PH) 0.6667 17.24
0.9 30-38 (SLG
A
) 0.8542 7.1241 30-38 (LL
BC
) 0.8750 7.8569 30-38 (DLG
BC
) 0.8620 9.3589 30-38 (3PH) 0.8594 18.124

TABLE VI. FAULT DIAGNOSIS AND LOCATION FOR THREE TRANSMISSION LINES EQUIPPED WITH DIFFERENT NUMBER OF PMUS

Line 75-77 (Equipped with Two PMUs) Line 49-69 (Equipped with One PMU) Line 30-38 (Equipped with No PMU)
Actual
FL
[pu]

Stage III Stage II Stage III Stage II Stage III Stage II
Diagnosis
Accuracy (AIl)
FL [pu]
Diagnosis
Accuracy (AIl)
FL [pu]
Diagnosis
Accuracy (AIl)
FL [pu]
Diagnosis
Accuracy (AIl)
FL [pu]
Diagnosis
Accuracy (AIl)
FL [pu]
Diagnosis
Accuracy (AIl)
FL [pu]
63.301 0.1212 63.301 0.1212 317.4487 0.1034 17.0945 0.0769 11.964 0.1094 1.5067 0.3333 0.1
73.854 0.3106 72.850 0.3030 16.353 0.3077 16.353 0.3077 7.4078 0.333 7.4078 0.3333 0.3
81.27 0.5000 79.785 0.4848 36.766 0.5000 32.169 0.5385 12.209 0.5 4.4445 0.3333 0.5
40.1 0.6818 38.4 0.6667 26.214 0.6923 26.214 0.6923 6.7816 0.666 6.7816 0.6666 0.7
108.4 0.8892 84.9 0.8788 2.9195 0.9038 2.7633 0.9231 7.1241 0.8542 2.2898 0.6666 0.9

Fault Inception
Undefined
Phasor
Pre-fault
Phasor
Post-fault
Phasor
Fault Clearance
6
In this way, the capability of the proposed algorithm in
distinguishing the actual fault line from the neighboring lines
can be measured. In order to examine the robustness of the
proposed diagnosis method under different system conditions,
several analyses in terms of different fault inception angles,
different sampling rates for PMUs, different fault types, and
the occurrence of fault along different lines are carried out.
A. Different Fault Inception Angles
Table III investigates the performance of the proposed
algorithm when the fault occurs at different phase angles. The
obtained results for fault location (FL) estimation and
accuracy index (AI) show that fault diagnosis and location is
not affected by the fault inception angle, confirming
successful removal of DC offset in phasor estimation.

B. Different Sampling Rates for PMUs
Table IV examines the impact of PMUs' sampling rate on
the proposed fault diagnosis and location scheme. It is
expected that by increasing the sampling rate more accurate
results are obtained. The accuracy can be measured by the AI
defined in (4). It is evident that both fault location estimation
and AI improve when the sampling rate excesses 96 samples
per cycle. Nonetheless, even with low sampling rates
acceptable results are obtained. It can be concluded that many
PMUs with relatively low sampling rates currently in use for
stationary or dynamic applications can be employed for wide
area fault diagnosis and location as well.
C. Different Fault Types
Various possible unbalanced and balanced faults are
imposed on line 30-38 to investigate the capability of the
proposed method in distinguishing the fault type besides fault
diagnosis and location. Different possible faults, namely
single-line-to-ground, line-to-line, double-line-to-ground, and
three-phase faults, are studied in Table V. The obtained
results indicate successful diagnosis of the fault type, together
with fault diagnosis and location.
D. Different Number of PMUs for Transmission Lines
In order to study the performance of the proposed method
in the cases of PMU-equipped lines, fault diagnosis and
location for two other transmission lines, i.e. lines 49-69 and
75-77, which are respectively equipped with one and two
PMUs, are studied. The results of fault diagnosis for stage II
is shown in Table VI, besides those of final fine-tuned fault
location in stage III. It can be clearly seen that although fault
line is correctly diagnosed in stage II, the fault location is not
accurate enough. However, in stage III, the fault location
estimation is enhanced significantly by the fine-tuning
process. This is also evident by comparing the corresponding
AI indexes for these two stages.
Comparing the diagnosis accuracy reflected in the AI
index, we can conclude that more reliable results are obtained
when one or both sides of the fault line are equipped with
PMUs. Nevertheless, different faults along non-PMU line 30-
38 are also accurately diagnosed by the proposed scheme.

V. COMPARISON WITH CONVENTIONAL FAULT DIAGNOSIS
Conventional fault diagnosis approach is based on the
function of the protection system. If a fault occurs in each
point of the network, the associated protective relays are
expected to trip off the corresponding circuit breakers,
thereby isolating the fault section from the remaining healthy
part of the system. If the protection system was 100% reliable,
there would be no need of fault diagnosis, as the fault section
is obviously the branch whose circuit breakers have operated.
However, the malfunction of the protective relays or/and
circuit breakers are not unusual in power systems, as they
may not operate correctly when required or may misoperate
when not required [3]. In such conditions, the outage area
may largely be extended as the operation of backup relays,
especially the operation of the third zone of distance relays
[17], is expected in order to isolate the fault.
In fact, most of the large blackouts often involve the
malfunction of the protection system, where the outages put
severe stress on the system and eventually the cascading
outages lead to the whole system collapse, as was the case in
Irans 2002 major blackout [18]. The conventional fault
diagnosis approach uses symptom-based schemes. In so
doing, the problem is often modeled as an optimization
problem in terms of the operations of different protective
relays and circuit breakers so that the most reasonable
scenario with the minimum malfunctions of the protection
system elements is obtained as the actual course of the event
[18]. With too many protective relays and/or circuit breakers
misoperating in a major blackout, conventional fault
diagnosis relying on the status of the protection systems
elements results in a false solution to the problem [9].
As such, in major blackouts where accurate fault diagnosis
is crucial to the system restoration, the conventional fault
diagnosis approach fails with the inadequate performance of
the protection system. In contrast, in such conditions where
the fault persists in the network, the proposed method finds
the fault section solely based on the measurements of PMUs
across the grid. It must be noted that, as discussed in Section
IV, the proposed method in this paper needs at most the
sampling of the during-faults signals over two cycles of power
frequency, when circuit breakers are not activated yet.
Therefore, regardless of the true or false operation of the
protective relays and/or circuit breakers, the proposed
measurement-based method diagnoses the fault point,
successfully. The only requirement for the implementation of
the proposed method is sophisticated information and
communication technology (ICT) infrastructure in the bulk
power transmission grid as depicted schematically in Fig. 1,
which is ongoing in many power systems around the world.

7
VI. CONCLUSION
A novel fault diagnosis approach utilizing wide area
measurement system is proposed in this paper. Voltage and
current synchrophasors provided by PMUs during the fault
can help to diagnose the fault in the transmission grid. Not
only can the proposed method diagnose the faulted line in the
grid, but also the fault location along the faulted line can be
determined by the synchronized measurements. This is
expected to be a new application of WAMS in the
transmission grid with the development of communication
infrastructure in smart transmission grids. The proposed
scheme has shown successful fault diagnosis while the
network is completely observable. As in many power systems
the buses equipped with PMUs still constitute a small percent
of the total system buses, future research will focus on fault
observability when the WAMS coverage is limited.
VII. ACKNOWLEDGMENT
The first author appreciates financial support for this work
by Irans National Elites Foundation.

VIII. REFERENCES
[1] F. Li, W. Qiao, H. Sun, H. Wan, J. Wang, Y. Xia, Z. Xu, and P. Zhang,
"Smart Transmission Grid: Vision and Framework," IEEE Trans. Smart
Grid, vol. 1, no. 2, pp. 168-177, Sept. 2010.
[2] A. G. Phadke and J. S. Throp, Synchronized Phasor Measurements and
Their Applications; 1st ed. New York:Springer, 2008.
[3] H. Monsef, A. M. Ranjbar, and S. Jadid, Fuzzy rule-based expert system
for power system fault diagnosis, IEE proc. on Gen., Transmiss.,
Distrib., vol. 144, no. 2, pp. 186-192, Mar. 1997.
[4] L. Xu and M. Kezunovic, Implementing fuzzy reasoning Petri-Nets for
fault section estimation, IEEE Trans. Power Del., vol. 23, no. 2, pp. 676
685, Apr. 2008.
[5] X. Lin, S. Ke, Z. Li, H. Weng, and X. Han, A fault diagnosis method of
power systems based on improved objective function and genetic
algorithm-tabu Search, IEEE Trans. Power Del., vol. 25, no. 3, pp.
12681274, Jul. 2010.
[6] A. J. Wood and B. F. Wollenberg, Power Generation, Operation and
Control, 2nd ed. New York: Wiley, 1996.
[7] D. Novosel, G. Bartok, G. Henneberg, P. Mysore, D. Tziouvaras, and S.
Wasrd, IEEE PSRC report on performance of relaying during wide-area
stressed conditions, IEEE Trans. Power Del., vol. 25, no. 1, pp. 316,
Jan. 2010.
[8] S. A. N. Sarmadi, A. S. Dobakhshari, S. Azizi, and A. M. Ranjbar, A
sectionalizing method in power system restoration based on WAMS,
IEEE Trans. Smart Grids, vol. 2, no. 1, pp. 190-197, Mar. 2011.
[9] A. Beschta, O. Dressler, H. Montag, and P. Struss, A model-based
approach to fault localisation in power transmission networks, Intelligent
Systems Engineering, vol. 2, no. 1, pp. 3-14, 1993.
[10] A. G. Phadke, and B. Kasztenny, Synchronized phasor and frequency
measurement under transient conditions, IEEE Trans. Power Del., vol.
24, no. 1, pp. 89-95, Jan 2009.
[11] W. Premerlani, B. Kasztenny, and M. Adamiak, Development and
implementation of a synchrophasors estimator capable of measurements
under transient conditions, IEEE Trans. Power Del., vol. 23, no. 1, pp.
109-123, Jan. 2008.
[12] M. Kezunovic, Smart fault location for smart grids, IEEE Trans. Smart
Grid, vol. 2, no. 1, pp. 11-22, Mar 2011.
[13] P. M. Anderson and A. A. Fouad, Power System Control and Stability,
2nd ed. New Jersey: John Wiley & Sons, 2003.
[14] H. Saadat, Power System Analysis, 2
nd
ed, New York: McGraw-Hill,
2002.
[15] DIgSILENT Power Factory Software, version 14, [Online]. Available:
http://www.digsilent.de
[16] S. Azizi , A. S. Dobakhshari, S. A. N. Sarmadi, and A. M. Ranjbar,
Optimal PMU Placement by an Equivalent Linear Formulation for
Exhaustive Search, IEEE Trans. Smart Grids, vol. 3, no. 1, pp. 174-182,
Mar. 2012.
[17] S. H. Horowitz and A. G. Phadke, Third zone revisited, IEEE Trans.
Power Del., vol. 21, no. 1, pp. 2329, Jan. 2006.
[18] M. Sanaye-pasand, Scrutiny of the Iranian national grid, IEEE Power
Energy Mag., vol. 5, no. 1, pp. 3139, Jan./Feb. 2007.
[19] W. Guo, F. Wen, G. Ledwich, Z. Liao, X.He, and J. Liang, An analytic
model for fault diagnosis in power systems considering malfunctions of
protective relays and circuit breakers, IEEE Trans. Power Del., vol. 25,
no. 3, pp. 13931401, Jul. 2010.

IX. BIOGRAPHIES


Ahmad Salehi Dobakhshari (S08) received his B.Sc. and M.Sc. degrees in
electrical engineering from Sharif University of Technology (SUT), Tehran,
Iran, in 2006 and 2008, respectively, where he is currently pursuing the Ph.D.
degree in electrical engineering. His current research interests include developing
and utilizing wide area measurement system in power systems, fault diagnosis
and Smart Grid.
Since 2009, he has also been with the System and Energy Study Center,
Monenco Iran Consulting Engineers Co., Tehran, Iran, where he is working on
the application of modern controllers to transmission grid.
Mr. Dobakhshari is a member of Irans National Elites Foundation.




Ali Mohammad Ranjbar received the M.Sc. and Ph.D. degrees in electrical
engineering from Tehran University and Imperial College of London in 1967
and 1975, respectively. Since then, he has been with the Department of Electrical
Engineering, Sharif University of Technology, where he is currently a full
professor. His research interests include power system protection and operation,
electric machines and Smart Grid.
Dr. Ranjbar was the director of Niroo Research Institute (NRI) from 1996
to 2007. He was elected as a Memorable Figure in 2004 for 30 years of
involvement in both academia and industry. He was awarded the outstanding
faculty of Sharif University of Technology in 1996 and the distinguished
professor in the Department of Electrical Engineering in 1998.

Das könnte Ihnen auch gefallen