You are on page 1of 6

Optimization of Directional Overcurrent Relay Times Using Laplace Crossover

Particle Swarm Optimization (LXPSO)

Kusum Deep

Jagdish Chand Bansal

Department of Mathematics
Indian Institute of Technology Roorkee
Roorkee, India
E-mail: kusumfma@iitr.ernet.in

Department of Mathematics
Birla Institute of Technology and Science Pilani
Pilani, India
E-mail: jcbansal@gmail.com

Abstract An important problem in electrical engineering is to


determine the optimal directional overcurrent relay times. The
problem is modeled as a constrained nonlinear continuous
optimization problem in which the decision variables are the
devices that control the act of isolation of faulty lines from the
system without disturbing the healthy lines. Two models are
considered namely IEEE-3 bus system and IEEE-4 bus system.
The optimization methodology is the Laplace Crossover
Particle Swarm Optimization (LXPSO), which has recently
been introduced by the authors. LXPSO uses a new
information sharing strategy amongst the particles of the
swarm using a new crossover, called Laplace Crossover (LX),
based on Laplace distribution. The results obtained by LXPSO
are compared with the results available in the literature. It is
shown that LXPSO is able to provide superior results in terms
of optimality and reliability in comparison to other methods.

power systems can be modeled as a non-linear constrained


optimization problem. Objective function for this problem is
the sum of the operating times of all the primary relays,
which are expected to operate in order to clear the faults of
their corresponding zones. The constraints of this problem
are bounds on all decision variables (called selectivity
constraints) and restrictions on each term of the objective
function to be within the specified limits.
The rest of the paper is organized in the following
manner. In section 2 review of the literature for this problem
is given. The general formulation of the problem is stated in
section 3. In section 4, details of particle swarm optimization
and Laplace Crossover Particle Swarm Optimization is
given. The methods of solution and discussion of results are
given in section 5. Finally the conclusions are given in
section 6.

Keywords- Particle Swarm Optimization, LXPSO, Crossover


Operator, Directional Overcurrent Relay Times.

I.

INTRODUCTION

The problem considered in this paper has its origin in


electrical power systems. It requires finding the optimal
values of decision variables subject to intricately
interconnected non-linear inequality constraints. The
problem is about computing the values of the decision
variables of the devices called Relays, which control the
act of isolation of faulty lines from the system without
disturbing the healthy lines.
Directional Overcurrent Relays (DOCRs) are provided in
electrical power systems to isolate only the faulty lines, in
the event of the faults in the system. These relays are placed
at both ends of each line. Thus, the number of directional
overcurrent relays in an electrical power system is twice the
number of the lines. To maintain the continuity of supply to
healthy sections and to isolate the faulty sections only, relays
are coordinated. This ensures that minimum lines are
disrupted when fault occurs. This is done in DOCRs by
properly fixing the two adjustable parameters of each relay
called settings. The two settings of each relay are plug
setting (PS) and time dial setting (TDS). There can be many
relays in the system depending on the size of the system.
Thus, each relay introduces two decision variables (one TDS
and one PS) in the problem. The above stated problem of
coordinating each DOCR with one another in electrical

c
978-1-4244-5612-3/09/$26.00 2009
IEEE

II.

PREVIOUS WORK

As the dimension of the problem increases for the


modern interconnected power systems, the complexity of the
problem increases. Also, due to the complexities of nonlinear programming techniques, most of the researchers have
solved the problem in linear environment by assuming the
values of decision variables (all plug settings), which make
the problem non-linear. This assumption is made based on
the basis of engineering experience e.g. [5], [14], [22] and
[23] etc.
The use of optimization techniques in relay coordination
was first suggested by Urdaneta et al. [21]. Elrafie and Irving
[14] used Sparse Dual Revised Simplex method of linear
programming suggested by Irving and Sterling [15] to
optimize TDS settings for assumed non-linear PS settings.
Laway and Gupta [17] applied Simplex and RosenbrockHillclimb methods to optimize TDS and PS settings
respectively, in a similar way, as used by Urdaneta et al.
[21]. These approaches were further followed by simplexbased approaches with more and more sophistications about
finer aspects of the relays [23], Chattopadhyay et al. [5],
Urdaneta et. al [21], Abdelaziz et al. [1], So and Li [20] used
evolutionary programming. Deep et. al [10] used RST2 of
[18] to solve the relay coordination problem for IEEE 3-bus
and IEEE 4 bus models. Dipti [11], applied genetic
algorithm (GA), self organizing migrating algorithm
(SOMA), self organizing migrating genetic algorithm
(SOMGA) which is a genetic algorithm hybridized with self

288

organizing migrating algorithm to solve the problem. Bansal


and Deep [2], used three versions of PSO, namely PSO with
time varying inertia weight, PSO with constriction factor and
Chaotic PSO [9], for the solution of this problem. A survey
of all coordination philosophies used by various researchers
in the past has been presented recently by Birla et al. [4] and
Bansal and Deep [2].
III.

GENERAL FORM OF THE PROBLEM

The operating time (T) of a DOCR is non-linear function


of the relay settings (Time Dial Settings (TDS) and Plug
Settings (PS) and the fault current (I) seen by the relay).
Therefore, Relay operating-time equation for a directional
overcurrent relay is given by a non-linear equation as given
below:

T=

TDS

(1)


PS CT

pri _ rating

N cl

Minimize OBJ =

i
T pri
_ cl _ in

i =1

N far

T prij _ far _ bus

(2)

j =1

Where,
N cl is number of relays responding for close-in fault.

N far is number of relays responding for far-bus fault.


T pri _ cl _ in is primary relay operating-time for close-in fault.
T pri _ far _ bus is primary relay operating-time for far-bus
fault.
The constraints are:
a) Bounds on variables TDSs:
i
i
TDS min
TDS i TDS max
; where, i varies

Only TDS and PS are unknown variables in the above


equation. These are the decision variables of the problem.
, and are the constants representing the behavior of
characteristic in a mathematical way, in which operating
time of the DOCR varies and are given as 0.14, 0.02 and 1.0
respectively as per [IEEE std. (1997)]. Value of CTpri_rating
depends upon the number of turns in the equipment CT
(Current Transformer). CT is used to reduce the level of the
current so that relay can withstand it. With each relay one
Current Transformer is used and thus, CTpri_rating is known
in the problem. Value of I (Fault current passing through the
relay) is also known, as it is a system dependent parameter
and continuously measured by measuring instruments.
Number of constraints for systems of bigger sizes will be
dependent upon the number of lines in the system (see Table
1). In practice, electrical engineering power systems may be
of even bigger size and there are other types of relays also
besides DOCRs. Coordinating DOCRs with other types of
relays generate even larger number of constraints than shown
in Table 1. It is evident from Table 1 that simultaneous
optimization of both the settings (TDS and PS) of each
DOCR of the system is a complex problem.
TABLE 1: The Complexity of the DOCR Problem as the Bus Size
Increases

No. of lines
No. of DOCRs (relays)
No. of decision variables
No. of selectivity constraints

Conventionally, objective function in coordination studies is


constituted as the summation of operating-times of all
primary relays, responding to clear all close-in and far-bus
faults. The objective function is as follows:

IEEE
3-bus
3
6
12
8

IEEE
4-bus
4
8
16
9

A. The Optimization Problem


The relay, which is supposed to operate first to clear the
fault, is called primary relay. A fault close to relay is known
as the close-in fault for the relay and a fault at the other end
of the line is known as a far-bus fault for this relay.

from 1 to Ncl .
i
i
TDS min
is lower limit and TDS max
is upper limit
of TDSi. These limits are 0.05 and 1.1,
respectively.
b) Bounds on variables PSs:

j
j
PS min
PS j PS max
; where, j varies from 1
to Nfar.
j
j
PS min
is lower limit and PS max
is upper limit of

PS j . These are 1.25 and 1.50, respectively.


c)

Limits on primary operation times:


This constraint imposes constraint on each term
of objective function to lie between 0.05 and 1.0.
d) Selectivity constraints for all relay pairs:
Tbackup T primary CTI 0 ; where Tbackup is

operating time of backup relay and Tprimary is operating time


of primary relay. Value of CTI is known.
B. Model I (The IEEE 3-Bus Model)
For the coordination problem of IEEE 3 bus model, value
of each of Ncl and Nfar is 6 (equal to number of relays or
twice the lines). Accordingly, there are 12 decision variables
(two for each relay) in this problem i.e. TDS1 to TDS6 and
PS1 to PS6. The 3 bus system can be visualized as shown in
Fig. 1.
2

Line

1
DOCR Relay
6

Generator 4
3

Fig. 1: A typical IEEE 3-bus DOCR coordination problem model.

2009 World Congress on Nature & Biologically Inspired Computing (NaBIC 2009)

289

Table 3: Values of Constants ei, fi, gi and hi for Model-I

Objective function (OBJ) to be minimized as given by


equation (2) is:
N cl

Minimize OBJ =

i =1

i
T pri
_ cl _ in +

N far

j =1

p
5
6
4
2
5
6
2
4

j
T pri
_ far _ bus

Here,
i
T pri
_ cl _ in =

0.14 TDS i
i

a
i
i
PS b

0.02

0.14 TDS j

Tprij _ far _ bus =

0.02

c
1

j
j
PS d
The values of constants a i , b i , c j and d j are given in the
Table 2.

Table 2: Values of Constants ai, bi, ci and di for Model-I

TDS
TDS1
TDS2
TDS3
TDS4
TDS5
TDS6

Tipri cl in
ai
9.4600
26.9100
8.8100
37.6800
17.9300
14.3500

b
2.0600
2.0600
2.2300
2.2300
0.8000
0.8000

TDS
TDS2
TDS1
TDS4
TDS3
TDS6
TDS5

Tipri far bus


ci
100.6300
14.0800
136.2300
12.0700
19.2000
25.9000

di
2.0600
2.0600
2.2300
2.2300
0.8000
0.8000

Tibackup
ei
fi
14.0800 0.8000
12.0700 0.8000
25.9000 2.2300
14.3500 0.8000
9.4600 0.8000
8.8100 0.8000
19.2000 2.0600
17.9300 2.2300

C. Model II (The IEEE 4-Bus Model)


The second coordination problem is of IEEE 4 bus model,
value of each of Ncl and Nfar is 8 (equal to number of relays
or twice the lines). Accordingly, there are 16 decision
variables (two for each relay) in this problem i.e. TDS1 to
TDS8 and PS1 to PS8. The value of CTI in 4 bus model is
0.3. The 4 bus system can be visualized as shown in Fig. 2.
The objective function and constraints for this model will be
of same form as in the case of Model-I (with N cl = 8)
described in section 3.2. The values of constants

a i , b i , c j , d j and e i , f i , g i , h i for Model-II are given in


Table 4 and Table 5 respectively.
1

i
i
TDS min
TDS i TDS max
;where, i varies from 1 to Ncl.
i
i
TDS min
is lower limit and TDS max
is upper limit of TDSi.
These limits are 0.05 and 1.1, respectively.
b) Bounds on variables PSs:

These are 1.25 and 1.50, respectively.


c) Limits on primary operation times:
This constraint imposes constraint on each term of
objective function to lie between 0.05 and 1.0.
d) Selectivity constraints for all relay pairs:
Tbackup T primary CTI 0 ; where Tbackup is operating
time of backup relay and Tprimary is operating time of primary
relay. Value of CTI is 0.3.
Here,

0.14 TDS p
0.02

i
; Tprimary =

0.14 TDS q
i

0.02

e
q i 1

1
p
i
PS h
PS f
i
i
i
The values of constants e , f , g and h i are given in the
Table 3.

290

XD

FX

XC

E X

j
j
PS min
PS j PS max
; where, j varies from1 to Nfar.
j
j
PS min
is lower limit and PS max
is upper limit of PS j .

A
X
1
6

B
X
2
4

The constraints are:


a) Bounds on variables TDSs:

i
backup

q
1
3
5
6
1
3
6
5

Tiprimary
gi
hi
14.0800 2.0600
12.0700 2.2300
25.9000 0.8000
14.3500 2.0600
9.4600 2.0600
8.8100 2.2300
19.2000 0.8000
17.9300 0.8000

X
8 H

X
G 7

Fig. 2: A typical IEEE 4-bus DOCR coordination problem model


Table 4: Values of Constants ai, bi, ci and di for Model-II
i

TDS
TDS1
TDS2
TDS3
TDS4
TDS5
TDS6
TDS7
TDS8

Tipri cl in
ai
20.3200
88.8500
13.6100
116.8100
116.7000
16.6700
71.7000
19.2700

b
0.4800
0.4800
1.1789
1.1789
1.5259
1.5259
1.2018
1.2018

TDS
TDS2
TDS1
TDS4
TDS3
TDS6
TDS5
TDS8
TDS7

Tipri far bus


ci
23.7500
12.4800
31.9200
10.3800
12.0700
31.9200
11.0000
18.9100

2009 World Congress on Nature & Biologically Inspired Computing (NaBIC 2009)

di
0.4800
0.4800
1.1789
1.1789
1.5259
1.5259
1.2018
1.2018

Table 5: Values of Constants ei, fi, gi and hi for Model-II

p
5
5
7
7
1
2
2
4
4

Tibackup
ei
fi
20.32 1.5259
12.48 1.5259
13.61 1.2018
10.38 1.2018
1.16
0.48
12.07
0.48
16.67
0.48
11
1.1789
19.27 1.1789

IV.

q
1
1
3
3
4
6
6
8
8

Tiprimary
gi
hi
20.32
0.48
12.48
0.48
13.61 1.1789
10.38 1.1789
116.81 1.1789
12.07 1.1789
16.67 1.5259
11
1.2018
19.27 1.2018

PARTICLE SWARM OPTIMIZATION

The particle swarm optimization (PSO) algorithm,


originally introduced in terms of social and cognitive
behavior by Kennedy and Eberhart in 1995 [16], can
effectively solve optimization problems in many fields,
especially engineering and computer science. The power of
the technique is its fairly simple computations and sharing
of information within the algorithm as it derives its internal
communications from the social behavior of individuals.
The individuals, called particles henceforth, are flown
through the multi-dimensional search space with each
particle representing a possible solution to the multidimensional optimization problem. Each solutions fitness is
based on a performance function related to the optimization
problem being solved.
The movement of the particles is influenced by two
factors using information from iteration-to-iteration as well
as particle-to-particle. As a result of iteration-to-iteration
information, the particle stores in its memory the best
solution visited so far, called pbest, and experiences an
attraction towards this solution as it traverses through the
solution search space. As a result of the particle-to-particle
information, the particle stores in its memory the best
solution visited by any particle, and experiences an
attraction towards this solution, called gbest, as well. The
first and second factors are called cognitive and social
components, respectively. After each iteration the pbest and
gbest are updated for each particle if a better or more
dominating solution (in terms of fitness) is found. This
process continues, iteratively, until either the desired result
is converged upon, or its determined that an acceptable
solution cannot be found within computational limits.
Working of PSO may be briefly described as under:
Suppose the search space is n dimensional, then the i-th
particle of the swarm can be represented by a ndimensional vector, Xi = (xi1, xi2, ,xin)T . The velocity
(position change) of this particle can be represented by
another n- dimensional vector Vi = (vi1, vi2,,vin)T . The best
previously visited position of the i-th particle is denoted as
Pi = (pi1, pi2, ,pin)T . Defining g as the index of the best
particle in the swarm, the swarm is manipulated according
to the following two equations:

Velocity Update Equation:

vid = vid + c1r1( pid xid ) + c2r2 ( pgd xid )

(3)

Position Update Equation:

xid = xid + vid

(4)
where d = 1, 2 n; i = 1, 2,, S, where S is the size of the
swarm; c1 and c2 are constants, called cognitive and social
scaling parameters respectively (usually, c1= c2; r1, r2 are
random numbers, uniformly distributed in [0, 1] ). Equations
(3) and (4) define the initial version of PSO algorithm. A
constant, Vmax, was used to arbitrarily limit the velocities of
the particles and improve the resolution of the search. The
maximum velocity Vmax, serves as a constraint to control
the global exploration ability of particle swarm. A larger
Vmax facilitates global exploration, while a smaller Vmax
encourages local exploitation. The concept of an inertia
weight was also developed to better control exploration and
exploitation. The motivation was to be able to eliminate the
need for Vmax. The inclusion of an inertia weight in the
particle swarm optimization algorithm was first reported in
the literature in 1998 [19].
After some experience with the inertia weight, it was
found that although the maximum velocity factor, Vmax,
couldnt always be eliminated, the particle swarm algorithm
works well if Vmax set to the value of the dynamic range of
each variable (on each dimension). The resulting velocity
update equation becomes:
vid = w * vid + c1r1 ( pid xid ) + c2r2 ( pgd xid )
(5)
Eberhart and Shi, [13] indicate that the optimal strategy
is to initially set w to 0.9 and reduce it linearly to 0.4,
allowing initial exploration followed by acceleration toward
an improved global optimum.
Clerc, [6] has introduced a constriction factor, ,
which improves PSOs ability to constrain and control
velocities. is computed as:

(6)

2 ( 4)

Where = c1 + c 2 , > 4 , and the velocity update


equation is then
vid = * vid + c1r1 ( pid xid ) + c 2 r2 ( p gd xid )
(7)

Eberhart and Shi, [13] found that , combined with


constraints on Vmax, significantly improved the PSO
performance.
Much of the research has focused on designing the
influences for a particular particle under consideration. In
[3], a crossover based on Laplace distribution was introduced
in PSO that develops an interaction model between any two
randomly chosen particles.
A. Laplace Crossover PSO (LXPSO)
In LXPSO, a crossover for PSO is proposed which uses
Laplace Distribution. This parent centric operator is called

2009 World Congress on Nature & Biologically Inspired Computing (NaBIC 2009)

291

Laplace

Crossover

(LX)

Operator. Two

off-springs

y1 = ( y11 , y12 ,..., y1D ) and y2 = ( y21 , y22 ,..., y2 D ) are


generated from a pair of parents x1 = ( x11 , x12 ,..., x1D ) and
x2 = ( x21 , x22 ,..., x2 D ) using LX as follows (For details

refer [3]):
First, a uniformly distributed random number ui (0,1)
is generated. Then, from Laplace distribution function, the
ordinate i is calculated so that the area under the
probability curve excluding area from a (location parameter)
to i is equal to chosen random number u i .

ui
a b log e (1 2ui ),
2
i =
1
a b log e (2ui 1),
ui >
2

Here, b is called the scale parameter. The off-springs are


then given by the equations:
y1i = x1i + i x1i x 2 i
(8)

y2i = x2i + i x1i x2 i

(9)

i = 1,2,..., D
Based on the Laplacian operator described as above, two
new particles are formed. The best particle (in terms of
fitness) is selected. This new particle, called Laplacian
particle, can replace one of the particles from which it is
formed or replace the worst performing particle in the
swarm. LXPSO analyze swarms behavior if the worst
particle (in terms of fitness) is replaced by this Laplacian
particle. PSO with Laplace crossover is called as Laplace
Crossover PSO (LXPSO).
Two versions of PSO are considered, namely PSO-W
(PSO with time varying inertia weight) and PSO-C (PSO
with constriction factor) in order to evaluate the performance
of LXPSO. Thus two versions of LXPSO come in existence,
LXPSO-W (LXPSO with time varying inertia weight) and
LXPSO-C (LXPSO with constriction factor).

V.

METHOD OF SOLUTION AND DISCUSSION OF RESULTS

Four variants of PSO are used to solve IEEE 3-bus and


IEEE 4-bus model problems. These are PSO with time
varying inertia weight (PSO-W), PSO with constriction
factor (PSO-C), LXPSO with time varying inertia weight
(LXPSO-W), LXPSO with constriction factor (LXPSO-C).
Swarm size S is set to 50. The inertia weight w is set to
reduce linearly from 0.8 to 0.4. Constriction coefficient is
calculated from equation (6). For PSO-C the cognitive and
social scaling parameters c1 and c2 are 2.8 and 1.3
respectively. And for PSO-W both are set to 2. Global
variants of PSO are considered here. Maximum velocity,
Vmax is set equal to 0.5*(Xmax-Xmin), where Xmax and
Xmin are the upper and lower bounds of the decision
variables. The location and scale parameters a and b for
Laplace crossover are 1 and 0.9 respectively (as suggested
in [3]). The criterion to terminate the simulation of the

292

algorithms is reaching maximum number of iterations,


which is set as 1000. For fair comparison we set the
parameters of LXPSO-W and LXPSO-C are set as in their
respective classical versions i.e. parameters for PSO-C and
LXPSO-C are same and for PSO-W and LXPSO-W are
same. Static penalty function approach is used for constraint
handling [8]. Each problem has been run 30 times and the
best solution obtained out of the 30 runs is reported as the
global optimal solution. The mean objective function value
is also reported to compare the reliability of the algorithms
considered here. For comparison, previously quoted results
by RST2, GA, SOMA, and SOMGA [12] are also
reproduced here.
Table 6 shows the results for IEEE 3 Bus model obtained
by PSO-W, PSO-C, LXPSO-W, and LXPSO-C. Table 6
also contains minimum objective function value (Min OBJ)
out of 30 runs obtained by RST2, GA, SOMA, and SOMGA
for IEEE 3-Bus model are reported [12]. It can be observed
from Table 6 that SOMGA provides best results among all
previously quoted results as the minimum objective function
value obtained by SOMGA is minimum. From Table 6 it is
also clear that all versions of PSO considered here provides
better results than previously quoted results. Also we see
that LXPSOs are better than all other versions of PSO as
well as previously quoted results. If we compare LXPSOs,
then we see that LXPSO-C provides best results. This
verifies the conclusions drawn in [3], which states that
LXPSO-C is a better choice than LXPSO-W.
Table 6: Optimal Values of Decision Variables, Minimum Objective
Function Value and Mean Objective Function Value for IEEE 3 Bus Model
obtained by RST2, GA, SOMA, SOMGA, PSOs and LXPSOs.

D.V.
TDS1
TDS2
TDS3
TDS4
TDS5
TDS6
PS1
PS2
PS3
PS4
PS5
PS6
Min OBJ
Mean OBJ
Previous
Results [12]
Min OBJ

PSO-W
0.05
0.197647
0.05
0.209413
0.183754
0.180676
1.25
1.5
1.25
1.499142
1.450163
1.5
4.783847
4.796408

LXPSO-W
0.05
0.197709
0.05
0.209035
0.18121
0.180678
1.250002
1.499178
1.25
1.49998
1.499914
1.499967
4.780764
4.79570048

PSO-C
0.05
0.200942
0.05
0.209566
0.181411
0.182515
1.25
1.5
1.251405
1.5
1.497092
1.467093
4.78217
4.794473

LXPSO-C
0.05
0.197647
0.05
0.209032
0.181205
0.180676
1.25
1.5
1.25
1.5
1.5
1.5
4.7806508
4.793628

GA
5.07616

SOMA
8.01016

RST2
4.835427

SOMGA
4.78989

In Table 7, the optimal values of decision variables


(D.V.) and minimum objective function (OBJ) obtained by
RST2, GA, SOMA, and SOMGA for IEEE 4-Bus model are
reported [12]. Table 7 shows the results for IEEE 4 Bus
model obtained by PSO-W, PSO-C, LXPSO-W, and
LXPSO-C. The same conclusion i.e. SOMGA is better than

2009 World Congress on Nature & Biologically Inspired Computing (NaBIC 2009)

all other previously quoted results and LXPSO-C is better


than all other PSOs, drawn from here.
Table 7: Optimal Values of Decision Variables, Minimum Objective
Function Value and Mean Objective Function Value for IEEE 4 Bus Model
obtained by RST2, GA, SOMA, SOMGA, PSOs and LXPSOs.

D.V.
TDS1
TDS2
TDS3
TDS4
TDS5
TDS6
TDS7
TDS8
PS1
PS2
PS3
PS4
PS5
PS6
PS7
PS8
Min OBJ
Mean OBJ
Previous
Results [12]
Min OBJ

PSO-W
0.05
0.212169
0.05
0.151576
0.126479
0.05
0.133786
0.05
1.284329
1.5
1.25
1.5
1.5
1.25
1.5
1.25
3.670242
3.678875

LXPSO-W
0.05
0.212169
0.05
0.151576
0.1264
0.05
0.133786
0.05
1.273327
1.5
1.25
1.5
1.5
1.25
1.5
1.25
3.669327
3.675315

PSO-C
0.050201
0.212509
0.05
0.15202
0.128223
0.05
0.134236
0.05
1.499937
1.5
1.25
1.492643
1.5
1.252359
1.5
1.25
3.677945
3.69403

LXPSO-C
0.05
0.212175
0.05
0.151578
0.1264
0.05
0.133813
0.05
1.273329
1.499859
1.25
1.499953
1.5
1.25
1.499405
1.25
3.669327
3.675038

GA
3.85874

SOMA
3.78922

RST2
3.70502

SOMGA
3.67453

[6]

[7]
[8]
[9]

[10]

[11]

[12]
[13]

[14]

[15]

VI. CONCLUSION
In this paper, electrical engineering power system DOCR
coordination problem is solved by modeling it as constrained
non-linear continuous optimization problem using four
different versions of PSO. The problem is to determine the
optimal value of Time Dial Setting and Plug Setting so that
the relay operating time can be minimized. Two models of
this problem namely IEEE 3-Bus and IEEE 4-Bus are solved
using PSO. The complexities of both the models are different
due to different decision variables and constraints. The
results obtained by four versions of PSO are compared with
GA, SOMA, SOMGA and RST2 algorithm. In both models,
LXPSO outperforms all other algorithms. Thus LXPSO is
found to be a robust technique for such type of constrained
nonlinear continuous optimization problems.

[16]
[17]

[18]

[19]

[20]

[21]

[22]

REFERENCES
[1]

[2]

[3]

[4]

[5]

Abdelaziz, A. Talaat, H. Nosseir, A. and Hajjar, A. (2002), 'An


adaptive protection scheme for optimal coordination of overcurrent
relays', Electric Power Systems Research 61(1), 1-9.
Bansal, J. C. and Deep, K. (2008), 'Optimization of directional
overcurrent relay times by particle swarm optimization' 'Swarm
Intelligence Symposium, 2008. SIS 2008. IEEE', 1-7.
Bansal, J. C. Deep, K. Kalyan, V. and Osadciw, L. (2009),
'Information Sharing Strategy among Particles in Particle Swarm
Optimization Using Laplacian Operator' 'accepted, Swarm
Intelligence Symposium, 2009. SIS 2009. IEEE'.
Birla, D. Maheshwari, R. and Gupta, H. (2005), 'Time-Overcurrent
Relay Coordination: A Review', International Journal of Emerging
Electric Power Systems 2(2), 1039.
Chattopadhyay, B. Sachdev, M. Sidhu, T. Inc, M. and Eldridge, I.
(1996), 'An on-line relay coordination algorithm for adaptive

[23]

protection using linear programming technique', IEEE Transactions


on Power Delivery, 11(1), 165-173.
Clerc, M. (1999), 'The swarm and the queen: towards a deterministic
and adaptive particle swarm optimization', Proceedings of the
Congress on Evolutionary Computation, 1999. CEC 99.
Deb, K. (2001), Multi-Objective Optimization Using Evolutionary
Algorithms, Wiley.
Deb, K. (1998), Optimization for Engineering Design: Algorithms
and Examples, Prentice Hall of India.
Deep, K. and Bansal, J. C. (2008), 'A New Chaotic Particle Swarm
Optimization Algorithm', International Journal of Mathematical
Modeling, Simulation and Applications 2(2).
Deep, K. Birlar, D. Maheshwari, R. Gupta, H. and Thakur, M. (2006),
'A population based heuristic algorithm for optimal relay operating
times', World Journal of Modelling and Simulation 2(3), 167-176.
Deep, K. and Thakur, M. (2007), 'A new crossover operator for real
coded genetic algorithms', Applied Mathematics and Computation
188(1), 895-911.
Dipti (2007), 'Hybrid Genetic Algorithm and Their Applications',
Ph.D. Thesis, Indian Institute of Technology Roorkee, India.
Eberhart, R. and Shi, Y. (2000), 'Comparing inertia weights and
constriction factors in particleswarm optimization' Proceedings of the
Congress on Evolutionary Computation, 2000.
LRAFIE, H. and IRVING, M. (1993), 'Linear programming for
directional overcurrent relay coordination in interconnected power
systems with constraint relaxation', Electric power systems research
27(3), 209-216.
Irving, M. and Sterling, M. (1983), 'Economic dispatch of active
power with constraint relaxation' 'IEE Proceedings C. Generation,
Transmission and Distribution', 172-7.
Kennedy, J. and Eberhart, R. (1995), 'Particle swarm optimization',
Proceedings of IEEE International Conference on Neural Networks.
Laway, N. and Gupta, H. (1993), 'A method for adaptive coordination
of directional relays in an interconnected power system',
Developments in power system protection 368, 240-243.
Shankar, K. and Mohan, C. (1987), 'A Random Search Technique for
the Global Minima of Constrained Nonlinear Optimization Problems'
'Proceedings of the International Conference on Optimization
Techniques and Applications, Singapore', 905-918.
Shi, Y. and Eberhart, R. (1998) A modified particle swarm
optimizer, Proceedings of the IEEE International Conference on
Evolutionary Computation, pp. 69 73.
So, C. and Li, K. (2000), 'Overcurrent relay coordination by
evolutionary programming', Electric Power Systems Research 53(2),
83-90.
Urdaneta, A. Nadira, R. and Perez Jimenez, L. (1988), 'Optimal
coordination of directional overcurrent relays in interconnected power
systems', IEEE Transactions on Power Delivery 3(3), 903-911.
Urdaneta, A. Prez, L. Gmez, J. Feijoo, B. and Gonzlez, M. ( 1) ,
'Presolve analysis and interior point solutions of the linear
programming coordination problem of directional overcurrent relays',
International Journal of Electrical Power and Energy Systems 23(8),
819-825.
Urdaneta, A. Restrepo, H. Marquez, S. and Sanchez, J. (1996),
'Coordination of directional overcurrent relay timing using linear
programming', IEEE Transactions on Power Delivery 11(1), 122-129.

2009 World Congress on Nature & Biologically Inspired Computing (NaBIC 2009)

293

You might also like