Anda di halaman 1dari 12

J Control Autom Electr Syst

DOI 10.1007/s40313-015-0219-x

A Modified Cuckoo Search Algorithm for Improving Voltage


Profile and to Diminish Power Losses by Locating Multi-type
FACTS Devices
Siva Sankar Akumalla1

Sujatha Peddakotla2 Seetha Rama Anjaneyulu Kuppa2

Received: 11 June 2015 / Revised: 10 September 2015 / Accepted: 15 October 2015


Brazilian Society for AutomaticsSBA 2015

Abstract This paper proposes a cuckoo search (CS) algorithm, a stochastic heuristic algorithm, combined with the
most familiar genetic algorithm (GA) to seek optimal location(s) of FACTS devices in a multi-machine power system.
The intention of hybridizing GA with CS algorithm is to
improve the quality of solution through expanding the search
space and speed of convergence. The run time and the
required function evaluation number(generations) for acquiring optimum by the modified algorithm are generally smaller
than the basic algorithm. Identification of the best location
for FACTS is a vital task as they are expensive to use.
Here, three emerging and dissimilar kinds of FACTS devices,
namely Unified Power Flow Controller, Thyristor Controlled
Series Capacitor and Interline Power Flow Controller, are
chosen for optimum locations and are modeled for steadystate studies. The optimal location and size of these FACTS
devices, including installation costs, are computed utilizing
the real power losses of the system as the objective function
to be minimized. The feasibility of the proposed method is
demonstrated for IEEE 30 bus power system network using
MATLAB working platform. The results show that the proposed approach, with good stability, has better convergence
and the simultaneous use of several kinds of FACTS controllers is the most efficient solution to improve the voltage
profile with minimum power loss of the system.
Keywords Cuckoo search Genetic algorithm UPFC
TCSC IPFC FACTS power losses

Siva Sankar Akumalla


sankarelectrical@gmail.com

Department of Technical Education, Hyderabad, AP, India

Department of Electrical Engineering, JNTUCEA,


Anantapuramu, AP, India

1 Introduction
As the modern electric power system is a heterogeneous,
globally interconnected, and a widely dispersed network,
controlling such a greatly complex system is a serious technological challenge. So modern power systems are prone to
widespread failures due to its complexity in structure, rapid
growth, and operating methods which leave the power system
exposed to instabilities (Baghaee et al. 2008). The technical
aspects of the challenges demand improving existing technology through engineering and inventing new technologies.
The Flexible AC Transmission System (FACTS)a new
technology based on power electronicsoffers an opportunity to enhance controllability, stability, and power transfer
capability of AC transmission systems (Hingorani and Gyugyi 2000). FACTS devices can regulate the active and reactive
power control as well as adaptive to voltage-magnitude
control simultaneously because of their flexibility and fast
control characteristics. Placement of these devices at suitable
location can alter line power flows and keep bus voltages at
desired level and so improve voltage stability margins and
enhance network security (Baghaee et al. 2008). Finding
the optimal location, type, and size of these devices is very
vital due to their considerable installation costs (Tabatabaei
et al. 2011). Optimal location of FACTS devices is a complex
combinatorial problem, differently defined according to the
goals to be achieved, as well as the types of devices considered. A comparative study on optimal placement of FACTS
devices, using different optimization methods, according to
multiple criteria such as the type of FACTS device considered, the specific purpose of the device in the power system,
and the optimization technique applied in the methodology,
is reported in the review article (Hernandez et al. 2013).
An approach to discover the optimal location of FACTS
devices such as Unified Power Flow Controller (UPFC),

123

J Control Autom Electr Syst

STATCOM, SSSC, Thyristor Controlled Series Capacitor


(TCSC), and SVC in electrical power transmission system
which enhanced the active and reactive power flow of transmission lines, voltage regulation, and reduce power loss
consequently upholding system constancy has been recommended by Kumaran and Allirani (2011). Optimal locations
of FACTS devices and their influences in a multi-machine
system are studied in Gerbex et al. (2003) using three heuristic methods: simulated annealing (SA), tabu search (TS), and
genetic algorithms (GA). Here, increasing security margin
of a power system with help of five types of FACTS devices
such as TCSC, TCVR, TCPST, SVC, and UPFC has been
demonstrated.
A number of deterministic and stochastic global optimization methods have been discussed in Pardalosa et al.
(2010) with their capabilities and demerits. But the stochastic algorithms have been found more promising than
the deterministic global optimization methods. However,
they often enforce a trade-off between speed and reliability in optimization calculations. Till date, there is no single
stochastic method capable of solving all optimization problems of different types and structures (Rashedi et al. 2009).
As optimization in most engineering and other real-world
applications concerns multiple objectives under complex,
highly nonlinear constraints, which results in computationally expensive solutions (Yang et al. 2014), and there is a
need to develop more research in the optimization methods
dealing such problems, and parameter-tuning for algorithms
is also important to ensure the best performance of an algorithm (Yang 2013). These necessitated developing many new
metaheuristic algorithms, which is a hot research topic in
current literature of nonlinear and non-differentiable optimization problems.
Recently, a new metaheuristic search algorithm of some
cuckoo species, called cuckoo search (CS) algorithm, based
on the obligate brood parasitism has been developed by Yang
and Deb (2009a, b) and the performance of this technique
was proved to be potentially more efficient than other evolutionary algorithms such as GA and PSO (Moravej and
Akhlaghi 2013). According to Rajabioun (2011) and Adnan
and Razzaque (2013), the CS outperforms PSO in view of
computational efficiency when applied to problem-specific
distance functions (PSDF). In Yang and Deb (2010), CS
was compared with PSO over 100 trials for each objective function and allowed the algorithms to run until the
fitness was smaller than 105 , which resulted in large numbers of objective function evaluations. These many number
of objective function evaluations would not be feasible for
application to practical engineering problems with costly
objective functions. Although CS can solve even the complicated nonlinear multi-modal problems effectively than other
traditional or modern optimization methods, its performance
is still not as good as expected. Also, seeking the global

123

optima of multi-modal function by any optimization method


will always call for higher convergence rate. These reasons
call for further study on speed-up strategies for CS, which
may reduce the computational time of the algorithm while
obtaining better solutions and increasing the probability of
getting the global optimum solution. These motivations lead
to the development of modified cuckoo search (MCS), a possible improvement over the standard CS.
This paper presents a strategy to modify the CS algorithm
by combining it with the most popular GA. In this modified approach, updating each chromosome/nest is performed
by GA to generate an offspring through selection, crossover,
and mutation operations thus providing the offspring with
best qualities of parent chromosome. With this hybridization technique, the enhanced capabilities of the algorithm
can effectively deal several real-world problems involving
complexity.
This paper utilizes the superior performance of the modified CS algorithm for best locations of single and multi-type
FACTS devices in improving the voltage profile and reducing
power losses of a transmission system while satisfying the
power system constraints. The controlling of these FACTS
devices is beyond the scope of this paper, but the main contribution is optimally locating and sizing of multi-type FACTS
devices. The effectiveness of this technique is clearly evident
from the result shown when applied on IEEE 30 bus network
and validated it against original CS algorithm. The design of
the proposed algorithm is detailed clearly in Sect. 3.
The rest of the paper is organized as follows: Sect. 2
discusses the mathematical modeling of the mentioned
FACTS devices for steady-state studies. Section 3 provides an
overview of CS and Genetic algorithms. However, the main
focus of Sect. 3 is on the detailed description of modified
CS algorithm. Then, the simulation results are presented and
discussed clearly in Sect. 4. Finally, conclusions are drawn
in Sect. 5.

2 Modeling of FACTS Devices


Appropriate mathematical models are required to investigate
the effects of FACTS devices on power flows in steady state.
Several modeling ways of FACTS devices are proposed in
the literature (Wang 2003) for several purposes. Several models for UPFC, TCSC, and Interline Power Flow Controller
(IPFC) are conceivable and applied in different studies. In
this section, the modeling of FACTS devices in steady state
and their incorporation into the power flow calculations are
described. Since the power-injection model (Wang 2003) will
not alter the existing impedance matrix required for load flow
computation problem, it is regarded as a convenient model
for a power system with FACTS devices.

J Control Autom Electr Syst



b
cos (se A )]
A ) + bAB +
2
VA VB (gAB sin AB bAB cos AB )

(2)

where AB = A B and PAB and Q AB are the active


and reactive powers flow from bus A to B through the line
joining them. VA , VB , Vse and Ish are the voltages of buses
A and B, voltage of series transformer injection and current
of shunt transformer, respectively. A , B and se are the
phase angles of VA , VB and Vse , respectively. gAB , bAB and
b/2 are the line conductance, susceptance, and half line shunt
susceptance, respectively. The altering injected power values
in the line due to UPFC can also be represented (Verma and
Gupta 2006) as,

Pinj,A = Vse2 gAB 2VA Vse


 gAB cos (se A )

+ VB Vse gAB cos 

se j
+ bAB sin se j



sin se j
Q inj,A = VAIsh + VA V se [gAB

b
+ bAB + 2 cos se j ]
 


g cos se j 

Pinj,B = VB Vse AB

sin

b
AB
se
j
 


g sin se j 

Q inj,B = VB Vse AB
+bAB cos se j

Fig. 1 The UPFC: a circuit diagram and b injection model

2.1 UPFC
The structure of UPFC and its model for representing power
flow are depicted in Fig. 1. UPFC is used to control the real
power flow in the power transmission line using its both voltage source converters (VSCs).
In Fig. 1, the circuit and its equivalent diagram of the
UPFC connected between two buses A and B is presented.
It consists of two voltage source converters connecting the
UPFC in series and shunt manner to the transmission line
using respective transformers. Both VSCs are connected
back to back through a DC link for receiving/transferring
the real power from sending bus to receiving bus without
consuming it. The power flow between those buses solely
depends on the degree of bus voltages, their phase difference, and the impedance of the transmission line (Lubis and
Hadi 2012) in the power system. It is controlled to alter the
basic parameters of the voltage, impedance, or phase angle
of a transmission line to run the power flow as desired. Now,
the altered active and reactive power flows from bus A to B
with UPFC in the system is given by the Eqs. (1) and (12).

(3)

(4)

2.2 TCSC
TCSC is used to control the impedance of transmission line to
increase its total power transfer capability. In Caizares and
Faur (1999), a TCSC model suitable for voltage and angle
stability applications and power flow studies is presented.
Here, the structure of TCSC, from steady-state perspective,
is illustrated in Fig. 2.



PAB = VA2 + Vse2 gAB + 2VA Vse gAB cos (se A )
VA Vse [gAB cos (se A ) + bAB sin (se A )]

Q AB = VA Ish VA2

b
bAB +
2

(1)


VA Vse [gAB sin(se

Fig. 2 The TCSC: a basic structure and b injection model

123

J Control Autom Electr Syst

It consists of a capacitor inserted directly in series with


the transmission line, and the thyristor-controlled inductor
is mounted directly in parallel with the capacitor. The bidirectional thyristor valves {T  and T  } are gated as per the
requirement to vary impedance of the transmission line. The
effective line impendence of the transmission line after TCSC
connected is,
X eff = X AB + kTCSC X AB = X AB (1 + kTCSC )

(5)

where X eff , X AB , X TCSC , and kTCSC = X TCSC / X AB are the


effective line reactance with TCSC connected, actual line
reactance, reactance of TCSC, and coefficient of compensation, respectively. So, as per the requirement of compensation, kTCSC is varied between [0.7 0.2] by thyristor valves
to vary the line impedance; thus, the power flow is affected.
Now, the power flow equations of the TCSC (Tripathy and
Mishra 2011) connected in a line are given as

PAB = VA2 G AB VA VB [gAB cos AB + bAB sin AB


 ]
g sin AB
Q AB = VA2 (bAB + bsh ) VA VB AB

bAB cos AB
(6)
where gAB =
At bus A,

rAB
2 +(X
2
rAB
AB X TCSC )

Pinj,A =

kTCSC
VA VB
(kTCSC 1) x X line

sin AB

Q inj,A =

kTCSC
VA
(kTCSC 1) x X line

[VA VB cos AB ]

(7)

Fig. 3 The IPFC: a basic structure and b injection model

presented. IPFC is connected to the lines {(Line-1) = AB


and (Line-2) = AC} through series injection transformers.
From its equivalent structure, the respective injections by
these transformers are given as VL1 /_L1 and VL2 /_L2 . The
active and reactive power flow equations with IPFC in the
system (Zhang 2003) are as given as follows:
At bus A,
PA = VA2 gAA VA VB (gAB cos AB + bAB sin AB )

At bus B,

VA VC (gAC cos AC + bAC sin AC )

Pinj,B =

kTCSC
VA VB
(kTCSC 1) x X line

sin BA

Q inj,B =

kTCSC
VB
(kTCSC 1) x X line

[VB VA cos BA ]

VA VL1 (gAB cos AL1 + bAB sin AL1 )


(8)

VA VL2 (gAC cos AL2 + bAC sin AL2 )


QA =

The Interline Power Flow Controller (IPFC) is used for


addressing the problem of compensating a number of transmission lines in a given substation by balancing power flows
among them (Gyugyi et al. 1999). The building block of
the IPFC is VSC, and based on those VSCs, the power flow
among the multiple lines are compensated through a DC link
for various purposes. The basic structure and model of the
IPFC connected between three buses are illustrated in Fig. 3.
In Fig. 3, the circuit diagram of the IPFC connected
between three buses {A, B and C} and its equivalent are

123

(9-i)

VA VB (gAB sin AB bAB cos AB )

VA VC (gAC sin AC bAC cos AC )

Actually, TCSC does not exchange real power with the AC


system and only generates or absorbs the reactive power
required for compensation by the capacitor or reactor banks.
2.3 IPFC

VA2 bAA

VA VL1 (gAB sin AL1 bAB cos AL1 )


VA VL2 (gAC sin AL2 bAC cos AL2 )

(9-ii)

where
AB = A B ; AC = A C AL1 = A L1 ;
AL2 = A L2
At bus B
PB = VB2 gBB VA VB (gAB cos BA + bAB sin BA )
+VB VL1 (gAB cos BL1 + bAB sin BL1 )
QB =

VB2 bBB

(10-i)

VA VB (gAB sin BA bAB cos BA )

VB VL1 (gAB sin BL1 bAB cos BL1 )

(10-ii)

J Control Autom Electr Syst

And at bus C,
PC = VC2 gCC VA VC (gAC cos CA + bAC sin CA )
+VC VL2 (gAX cos CL2 + bAC sin CL2 )
QC =

VC2 bCC

(11-i)

VA VC (gAC sin CA bAC cos CA )

VC VL2 (gAC sin CL2 bAC cos C L )

(11-ii)

3 Description of the Proposed Algorithm


This section details the formulation of an algorithm based on
a modified cuckoo search for optimal locating and sizing the
FACTS in the power system. The proposed algorithm uses
the total power losses as the objective function to optimize the
locations and sizes of the employed FACTS devices: UPFC,
TCSC, and IPFC.
The Conventional CS algorithm is based on surviving
and reproduction manner of Cuckoo bird (Yang and Deb
2009a, b). In the conventional CS algorithm, a new nest or
solution is generated in a traditional manner, i.e., a Lvy flight
is performed starting at the position of a randomly selected
egg. Upon evaluation of the objective function at the new
nest, the cuckoo is moved to the new nest if its value is better than at randomly selected nest. The scale of this random
search is controlled by multiplying the generated Lvy flight
by a step size. This solution of the CS algorithm will always
find the optimum (Yang and Deb 2010), but, as the search
relies entirely on random walks, a fast convergence cannot be
guaranteed. This necessitates the CS algorithm to combine
with another algorithm called hybrid algorithm which uses
some combination of deterministic and randomness so as to
design more efficient algorithms. In the proposed controller,
CS is having a modification of replacing the Lvy distribution
by GA for updating the nests for next generation.
3.1 Overview of Conventional CS Algorithm
CS algorithm is one of the swarm algorithms which are
defined as the emergent collective intelligence of groups
of simple agents (ants colony, artificial bees colony, bird
flocking, and fish schooling). CS is inspired by the brood
parasitism of some cuckoo species by laying their eggs in the
nests of other host birds. Each egg (nest or cuckoo) represents a solution, and a cuckoo egg represents a new solution.
It mainly consists of four elements; cuckoo, nests, objective, and probability. Here, cuckoo searches for new nests in
accordance to improve the objective/quality of nests in iteration manner. The basic steps of CS algorithm are given as
follows:
Initialize the parameters to be optimized,

Generate the nests randomly within the range of parameters,


Check for termination and execute the algorithm until it
is satisfied,
Select a cuckoo with nest and update it with levy distribution (random walk),
If the updated nest is better in quality/fitness than the
present nest, then swap it,
Find the worst nests with a probability value and abandon
them,
Generate new nests randomly in the place of abandon
nests and go for termination.
This algorithm has the drawback of random walk which guarantees no optimal solutions. So to improve its performance,
it is altered with genetic algorithm. The description of GA is
as follows:
3.2 Overview of Genetic Algorithm
Genetic algorithm, inspired by genetics of living organisms, is one of the most familiar evolutionary algorithm in
which a population of individuals evolves through selection,
crossover, and mutation (Melanie 1999). Generally, GA consists of three elements: genes, chromosomes, and objective.
Here, the genes are called as parameters to be optimized
for objective function, and chromosomes are the strings of
genes representing a single solution. For optimizing any
objective function, it had five phases: initialization, selection, crossover, mutation, and termination. The procedure of
GA in simple terms is as follows:
Initialize the parameters/genes to be optimized,
Randomly generate the chromosomes, i.e., initial population,
Check for termination and execute the following phases
until it is satisfied,
Evaluate the fitness of all chromosomes,
Select the best chromosomes for reproduction based on
the fitness value,
Crossover and mutate the selected parent chromosomes
to generate a offspring,
Create a next-generation chromosomes using deletion
operator and go for termination.
3.3 The Proposed Hybrid Algorithm
In this subsection, the proposed algorithm based on CS with
the hybridization of GA is designed. The purpose of GA in
this algorithm is to avoid the random walk by a cuckoo by
inclusion of genetic operators (mutation and crossover). It
helps the cuckoo to go for next nest position from the current
nest position with sure improvement in its quality/fitness. So

123

J Control Autom Electr Syst

with elimination of Levy flight in CS algorithm, it is improved


to obtain the optimal nests with faster convergence and better
nest quality. Structure of the proposed algorithm showing
their basic step is as follows:
Step 1 Define the input data
In this step, select the number of FACTS devices to be
placed on the power system and their ranges, and select the
number of random variables and their ranges. Here, the number of random variables is equal to the number of parameters
(locations and sizes of selected FACTS devices).
Step 2 Initialize the nests randomly
In this step, the nests are randomly initialized of population size as I . The initialization of the nests parameter is
derived in Eq. (12).


xi j = xi j,min + xi j,max xi j,min

(12)

with 1 i I and 1 j n.
Where, xi j is the jth parameter of the ith nest to be optimized, where is the scaling factor with a value in the range
of [0, 1] and xi j,min and xi j,max are the minimum and maximum values of the jth parameter of the X i nest. The total
number of parameters to be optimized is n. The parameters to
be optimized are the locations and sizes of selected FACTS
devices. At the end of step 2, first generation of the nests is
obtained and iteration count starts at t = 1.
Step 3 Check for termination condition?
In this step, the termination condition for the proposed
MCS algorithm is defined and checked. The termination condition is given as,
t T?

(13)

Here, T is the maximum number of iterations to be performed


by the proposed algorithm to give the optimized results. If
the Eq. (13) is satisfied, the proposed algorithm stops and
provides the final results as output. The final result is the
optimized location and size of the selected FACTS devices.
Otherwise, go to step 4.
Step 4 Select a cuckoo and evaluate its fitness
In this step, a cuckoo is selected randomly and its fitness
function is evaluated. For this, the selected FACTS devices
are placed at the location specified by the ith nest with corresponding size. Then, using the NR method, the load flow
of the power system is performed and power loss at each bus
is computed. The fitness function to be evaluated is
Nb

 
Pb,Loss
F X it = Min
b=1

123

(14)

 
where, F X it is the total power loss due to the nest i during
the tth iteration. The total number of buses in the power
system is Nb , and Pb,Loss is the power loss at busb .
Step 5 Generate a new nest for cuckoo i using Genetic Algorithm (GA)
In this step, a new nest is generated for ith cuckoo which
lives in its ith nest. Here, the updating to the next nest is
performed by GA which makes the convergence faster. It is
derived in Eq. (15).
X it+1 = X it + Rm

(15)

where, X it+1 is the updated nest position of ith cuckoo. X it


is the present nest position and Rm is the random vector
generated by GA which minimizes the power losses surely.
Now, the GA-based updating to the next nest position
using Rm is as follows:
Phase 1 Initialize the chromosomes randomly
In this phase, the chromosomes of size N are initialized
randomly. It is worth mentioning that the number of genes
(random variables) is equal to the number of parameters in the
cuckoo search algorithm, i.e., n. The initialization of chromosomes is given as,


rm j = rm j,min + rm j,max xm j,min

(16)

where, rm j is the jth gene of mth chromosomes (Rm ), where


is the scaling factor with a value in the range of [0, 1], and
rm j,min and rm j,max are the minimum and maximum values of
the jth parameter/gene of the chromosome (Rm ). It is a random vector of n random values in their respective minimum
and maximum range of values. This is the first generation of
chromosomes and generation count starts, i.e., = 1.
Phase 2 Determine the multiple nests positions for ith cuckoo
In this phase, the multiple nests positions are determined
using the Eq. (17) for ith cuckoo.

( )
= X it + Rm
X i,m

(17)

where, X i,m
is the obtained nest due to mth chromosome
for ith cuckoo. It is performed for all chromosomes and N
number of new nests are obtained and they are given as

, X i,2
, . . . , X i,m
, . . . , X i,N
}
{X i,1

Phase 3 Compute the fitness of obtaining nests and check for


updating
In this phase, the fitness of obtaining multiple nests positions for ith computed by Eq. (17) are computed using
Eq. (14). From the obtained nests fitness evaluations, the
updating check of the ith cuckoo is given as,

J Control Autom Electr Syst

Fig. 4 The flow chart of the proposed approach

 
 
F X it > F X i,m

(18)

If the Eq. (18) is satisfied for any of the obtained nests


, X , . . . , X , . . . , X } then it
positions, i.e., for {X i,1
i,2
i,m
i,N
is swapped and given as the updated nest position for ith
) has lower
cuckoo. For example, obtained nest position (X i,m
power losses than the X i then it is placed in X it+1 , i.e.,
and go to step 6. Otherwise, go to Phase 4.
X it+1 X i,m

Phase 4 Generate an offspring


In this phase, an offspring is generated based on the
selection, crossover, and mutation operations on chromosomes (random vectors). Here, based on the fitness/power
loss value, best chromosomes are selected for reproduction.
Based on the crossover rate, a new chromosome is generated
and it is mutated using a mutation rate, thus provides an offspring with the best qualities of parent chromosomes. The
purpose of generating an offspring from the best chromo-

123

J Control Autom Electr Syst

somes is to achieve the best nest position which makes the


minimization of power loss. At the end of the reproduction,
an offspring is developed.
Phase 5 Update the population of chromosomes and go to
Phase 2.
In this phase, a new generation of chromosomes is generated using a deletion operator and keeps the size constant.
Therefore, the generation count increments, i.e., = + 1.
After the generation of a new population of chromosomes,
go to Phase 2.
Step 6 Abandon the worst nests and go to step 3.
In this step, the nests with probability values less than
the worst probability value (Pw ) are deleted, and in lieu of
them, new nests are generated using Eq. (12) for keeping the
size of nest constant (I ). The probability value of the nest is
determined as,
Pi =  I

fi

i=1 f i

with f i =

1
1 + F (X i )

(19)

I
where f i the fitness entity x, i=1
f i is the total fitness of all
entities, and F(X i ) is the fitness function. As this is a minimization problem, f i is necessary to be determined. Using
the Pi < Pw , the algorithm deletes the ith nest position which
has probability value less than the given Pw value. And again,
generates new nests positions randomly, thus iteration count
increments, i.e., t = t + 1 and go to step 3.
This is the procedure of the proposed algorithm for optimally locating and the sizing of the selected FACTS devices.
The flow chart of the proposed algorithm is illustrated in
Fig. 4.
The flow chart of the proposed algorithm is detailed in
Fig. 4 wherein the portion of the contribution of GA is mentioned in the dotted rectangular box. Once the algorithm is
finished, it will be able to determine the optimal locations
and sizes of the multi-FACTS devices using the minimum
power losses as its objective.

Fig. 5 IEEE 30 bus system test structure

4 Results and Discussions


In this section, we discuss the performance of the modified
cuckoo search algorithm to improve the voltage profile of the
power system. For the evaluation of the algorithm, a standard IEEE 30 bus test system is implemented in MATLAB
working platform. Applying the algorithm, optimal locations
and sizes of the multi-FACTS devices in the chosen complex
power system are computed. The determination of optimal
location and size depends on the maximum power loss of the
corresponding bus.
During the simulation, both the modified and the original
CS algorithms were run in a trial with the same initial popu-

123

Fig. 6 Convergence histories of best solutions of modified and original


CS algorithms for UPFC at 1215

lation through seeding it with uniformly distributed random


values within the given boundary constraints. The population
size and the maximum generation were specified to be the
same for the purpose of comparison of both the algorithms.
All the running trials were stopped in a given maximum generation of 100, with a population size of 30.

J Control Autom Electr Syst


Table 1 Comparison of results
of original and modified CS
algorithms for the 30-bus real
system

Algorithm

FACTS device

Best location

Value of fitness after


100 iterations (MW)

Run time (CPU time) (s)

CS

UPFC

1215

8.3732

98.56

Modified CS

UPFC

1215

7.3835

36.55

Table 2 Power loss evaluations and installation costs of the test system with and without FACTS device
Whether FACTS
devices connected?

Type of FACTS devices

Location

Total power losses (in MW)

Cost of the
FACTS devices (in $/kVAR)

NO

10.8049

YES

UPFC

5&7

8.0963

158.3440

12&15

7.3835

184.9191

15&13

7.4839

183.9235

12&15

9.8871

147.1803

15&23

9.9974

148.6250

18&19

9.7236

147.0325

5,7& 6

7.0675

188.4015

12,15&16

6.9123

188.1005

22,24&25

6.6235

187.9179

UPFC@5&7,

5.4601

498.3265

5.3256

474.7164

5.5409

493.7533

TCSC

IPFC

Multi-FACTS devices

TCSC@12&15,
IPFC@24,25&26
UPFC@5&7,
TCSC@15&23,
IPFC@12,15&16
UPFC@5&7,
TCSC@22&24,
IPFC@12,15&16

Figure 5 presents the structure of IEEE 30 bus test system


for evaluating the proposed system performance. The optimal
locations of the FACTS devices and their sizes are evaluated
in four cases as follows:
1.
2.
3.
4.

Optimal locating and sizing of UPF C,


Optimal locating and sizing of TCSC,
Optimal locating and sizing of IPFC,
Optimal locating and sizing of the combined operation
{UPFC + TCSC + IPFC}.

Figure 6 illustrates the graphical analysis results of convergence characteristics (power loss evaluation vs. iteration)
of both CS algorithms for the IEEE system with UPFC at
location 1215, and Table 1 gives the comparison results of
both the algorithms in terms of computational elapsed time.
Here, run time denotes the time of the best function evaluation number of the experiment. The graph is self-explanatory:
the modified CS has much better convergence rate than the
original algorithm within the given number of generations.

This again proved the supremacy of the modified CS over its


conventional version in terms of convergence performance.
In Table 2, the total real power losses (the fitness value)
for optimal locations of different FACTS devices by the
proposed algorithm are presented along with actual power
losses of the test system. It also shows the capacity of
the FATCS devices in terms of its installation cost for the
respective power loss minimization. A study of Table 2
reveals that the total power losses of the test system are
reduced by connecting FACTS devices at the optimally
specified locations. The normal total power loss of the test
system without FACTS device is 10.8049 MW, and with optimally located FACTS devices, this power loss reduces to
7.3835, 9.7236, 6.6235, and 5.3256 MW, respectively, for
the four cases. The best places corresponding to the given
reduced power losses are (12&15), (18&19), (22,24&25) and
{UPFC@5&7, TCSC@15&23, IPFC@12,15&16} respectively.
In Fig. 7, the voltage profile of the IEEE 30 bus system is presented for the best possible locations and sizes

123

J Control Autom Electr Syst

Fig. 7 Voltage profile of the test system with/without FACTS devices

of the multiple FACTS devices. Here, the red line is the


indication of the voltage profile without FACTS devices.
It had some deviations than the limited range (1.06 pu)
for 11th and 13th bus. These voltage deviations are made
to within the limits by the proposed technique by placing the FACTS in appropriate locations. From the Fig. 7,

123

for UPFC@12&15, TCSC@12&15, IPFC@22,24&25, and


multiple FACTS case, the voltages of the real test system are
improved and limited to within tolerable pu values. In other
words, the normal voltage of the buses has less deviations
from the slack bus and nearer to the boundary limits of the
good voltage profile, i.e., 0.9866 pu< Vi < 1.06 pu.

J Control Autom Electr Syst

Fig. 8 Convergence histories of the best locations and sizes of FACTS devices by the proposed controller

In Fig. 8, the power loss (fitness function) evaluation of


the nests in the proposed algorithm is presented for each
case until the best nest is achieved. It is apparent from
Fig. 8 that the power loss is gradually minimized as the
iteration count increases and at the end of the algorithm,
it got best nest with lower power losses, i.e., the optimal
location and size of FACTS for each case. Among the four
cases, we have less power loss of 5.3256 MW for the case

of different FACTS devices at the selected optimal locations. So, from this analysis, we can conclude that the
combined operation of these FACTS device results in best
minimization of the power losses(i.e., best fitness value of
the function) in the test system. However, as is evident from
Table 2, the main drawback of this case is that its overall
cost of installation is more than their independent operation.

123

J Control Autom Electr Syst

5 Conclusion
In this paper, we have formulated a modified cuckoo search
(CS) algorithm and applied it to place dissimilar FACTS
devices in a 30-bus standard power transmission system with
the aim to improve voltage profile while diminishing real
power losses. For this purpose, we have considered three
FACTS devices: UPFC, TCSC, and IPFC, and the employment of these devices in IEEE 30 bus test system substantial
reduction in line losses are achieved. The controller used
power losses of the system as objective function for optimally
locating and sizing of FATCS devices. The validation of the
proposed system is performed by simulating in MATLAB
platform. It was tested for four cases, independently placing
the mentioned three FACTS devices and combined operation
of them. As shown from simulation results, we achieved a
considerable reduction in the total real power losses from
10.8049 to 5.3256 MW for the simultaneous placement of
the three different FACTS devices at the optimal locations,
and the voltage profile of buses within tolerable limits by the
proposed CS algorithm. Hence, from the discussions of minimum power losses and good voltage profile, the proposed
algorithm in optimal locating and sizing of different FACTS
devices validated. The modified CS also demonstrates its
superiority over the standard CS algorithm in reducing power
losses(solution quality) of the given test system and having
faster convergence rate.

References
Adnan, M. A., & Razzaque, M. A. (2013). A comparative study of particle swarm optimization and cuckoo search techniques through
problem-specific distance. In International conference of information and communication technology (ICoICT). doi:10.1109/
ICoICT.2013.6574619.
Baghaee, H. R., Jannati, M., Vahidi, B. Hosseinian, S. H., & Rastegar,
H. (2008). Improvement of voltage stability and reduce power system losses by optimal GA-based allocation of multi-type FACTS
devices. In IEEE international conference on optimization of
electrical and electronic equipment. doi:10.1109/OPTIM.2008.
4602368.
Caizares, C. A., & Faur, Z. T. (1999). Analysis of SVC and TCSC controllers in voltage collapse. IEEE Transactions on Power Systems,
14(1), 158165.
Gerbex, S. Cherkaoui, R., & Germond, A. J. (2003). Optimal location of
FACTS devices to enhance power system security. In Power tech
conference proceedings, IEEE Bologna. doi:10.1109/PTC.2003.
1304363.
Gyugyi, L., Sen, K. K., & Schauder, C. D. (1999). The interline power
flow controller concept: A new approach to power flow management in transmission systems. IEEE Transactions on Power
Delivery, 14(3), 11151123.
Hernandez, A., Rodriguez, M. A., Torres, E., & Eguia, P. (2013). A
review and comparison of FACTS optimal placement for solving
transmission system issues. Renewable Energy and Power Quality
Journal (RE&PQJ), 11.

123

Hingorani, N., & Gyugyi, L. (2000). Understanding FACTS: Concepts


and technology of flexible AC transmission systems. New York:
Wiley-IEEE Press.
Kumaran, A., & Allirani, A. (2011). Location of FACTS devices for
real and reactive power flow control. In International conference
on signal, image processing and applications with workshop of
ICEEA (Vol. 21, pp. 242248).
Lubis, R. S., & Hadi, S. P. (2012). Tumiran. Selection of suitable location of the FACTS devices for optimal power flow. International
Journal of Electrical and Computer, 12(3), 38.
Mitchell, M. (1999). Introduction to genetic algorithms (Vol. 158).
Cambridge: MIT Press.
Moravej, Z., & Akhlaghi, A. (2013). A novel approach based on cuckoo
search for DG allocation in distribution network. International
Journal of Electrical Power and Energy Systems. doi:10.1016/j.
ijepes.2012.08.009.
Pardalosa, M., Edwin Romeijna, H., & Tuyb, H. (2010). Recent
developments and trends in global optimization. Journal of Computational and Applied Mathematics, 124, 209228. doi:10.1016/
S0377-0427(00)00425-8.
Rajabioun, R. (2011). Cuckoo optimization algorithm. Applied Soft
Computing, 11(8), 55085518.
Rashedi, E., Nezamabadi-pour, H., & Saryazdi, S. (2009). GSA: A
gravitational search algorithm. Information Sciences, 179(13),
22322248.
Tabatabaei, N. M., Aghajani, G. H., Boushehri, N. S., & Shoarinejad, S.
(2011). Optimal location of FACTS devices using adaptive particle
swarm optimization mixed with simulated annealing. International Journal on Technical and Physical Problems of Engineering,
3(2), 6070.
Tripathy, M., & Mishra, S. (2011). Interval type-2-based thyristor controlled series capacitor to improve power system stability. IET
Generation, Transmission and Distribution, 5(2), 209222.
Verma, K. S., & Gupta, H. O. (2006). Impact on real and reactive power
pricing in open power market using unified power flow controller.
IEEE Transactions on Power Systems, 21(1), 365371.
Wang, H. F. (2003). Modelling of multiple FACTS devices into
multi-machine power systems and applications. International
Journal of Electrical Power and Energy Systems. doi:10.1016/
S0142-0615(02)00033-9.
Yang, X. S., & Deb, S. (2009a). Cuckoo search via Levy flights. In
Proceedings of world congress on nature and biologically inspired
computing (NaBIC 2009 India) (pp. 210214). USA: IEEE Publications.
Yang, X. S., & Deb, S. (2009b). Cuckoo search via lvy flights. In World
congress on nature and biologically inspired computing (NaBIC)
(pp. 210214). IEEE Publications.
Yang, X. S., Koziel, S., & Leifsson, L. (2014). Computational optimization, modelling and simulation: Past, present and future. In
ICCS 2014. 14th international conference on computational science (Vol. 29, pp. 754758). Elsevier.
Yang, X. S. (2013). Multiobjective firefly algorithm for continuous optimization. Engineering with Computers, 29(2), 175184. doi:10.
1007/s00366-012-0254-1.
Yang, X. S., & Deb, S. (2010). Engineering optimisation by cuckoo
search. International Journal of Mathematical Modelling and
Numerical Optimisation, 1(4), 330343.
Zhang, X. P. (2003). Modelling of the interline power flow controller
and the generalised unified power flow controller in Newton power
flow. IEEE Proceedings on Generation, Transmission and Distribution, 150(3), 268274.

Anda mungkin juga menyukai