See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357047672

Application of Meta-Heuristic Algorithm for Finding the Best Solution for the
Optimal Power Flow Problem

Article · December 2021

DOI: 10.22266/ijies2021.1231.47

CITATIONS READS

0 43

3 authors:

Thanhlong Duong Ngọc Anh Nguyễn

Industrial University of Ho Chi Minh Industrial University of Ho Chi Minh City
25 PUBLICATIONS 231 CITATIONS 4 PUBLICATIONS 8 CITATIONS

SEE PROFILE SEE PROFILE

Thuan Thanh Nguyen

Industrial University of Ho Chi Minh
48 PUBLICATIONS 578 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Application FACTS devices in power system View project

All content following this page was uploaded by Thanhlong Duong on 15 December 2021.

The user has requested enhancement of the downloaded file.

Received: September 3, 2021. Revised: October 7, 2021. 528

Application of Meta-Heuristic Algorithm for Finding the Best Solution for the
Optimal Power Flow Problem

Thanh Long Duong1* Ngoc Anh Nguyen1 Thuan Thanh Nguyen1

1
Faculty of Electrical Engineering Technology, Industrial University of Ho Chi Minh City, Vietnam
* Corresponding author’s Email: [email protected]

Abstract: Optimal power flow (OPF) is an important problem in the power system operation. The purpose of the OPF
problem is to optimize a defined objective function by modifying variables of control such as real power at generator
buses except slack bus generator, voltages at all generator buses, reactive power of compensators and tap ratio of
transformers while all constraints are satisfied. This paper is proposed an approach based on the artificial ecosystem
optimization (AEO) to solve problem of optimal power flow. The suggested algorithm is tested on the IEEE-30 bus
systems with five target functions consisting of fuel cost, emission, power loss, voltage deviations and L_index. The
results obtained of the suggested AEO approach compared with equilibrium optimizer (EO), particle swarm
optimization (PSO), sunflower optimization (SFO), genetic algorithm (GA) and other exiting methods. The results
simulation shows that, standard deviation obtained value after 50 independent runs by the proposed AEO algorithm is
better compared with EO, PSO, SFO and GA method. The fuel cost, emissions, active power loss and voltage
deviations levels are reduced by 11.21%, 44.06%, 46.44%, and 92.13% respectively, compared to the initial case.
Furthermore, for other exiting methods the improvement level percentage (IL) of the proposed AEO algorithm can be
up to 0.2285 % for fuel cost objective, 0.137% for emission objective, 7.618% for total power loss objective, 89.85%
for voltage deviation objective and 0.652% for L_index objective. Thus, the proposed AEO method is also one of
effective and reliable algorithms for handling OPF problem.
Keywords: Artificial ecosystem optimization, Optimal power flow, Power loss, Generator cost.

always a nonlinear optimization problem and may be

1. Introduction a non-differentiable one, thus it is an actual challenge
for optimization methods for dealing with, especially
Optimal power flow (OPF) plays important role
the conventional methods
in operating and planning of power system. The OPF
To over the limitations of classical methods,
aims to optimize a defined objective by modifying
heuristic methods have been considered as alternative
variables of control such as real power at generator
approaches to solve the OPF problem with the
buses except slack bus generator, voltages at all
advantages of obtaining nearly optimum solution
generator buses, reactive power of compensators and
whether the problem is differentiable or not. Many
tap ratio of transformers while satisfying constraints.
heuristic optimization methods have applied for
Many conventional approaches have been
solving OPF problem such as tabu search (TS) [6],
implemented for handling the OPF problem such as
evolutionary programming (EP) [7], differential
linear programming (LP) [1], nonlinear programming
evolution (DE) [8], biogeography optimization
(NLP) [2], newton-based technique [3], quadratic
(BOA) [9], teaching learning optimization algorithm
programming (QP) [4], and interior point (IP)
(TLOA) [10], stud krill herd algorithm (SKHA) [11],
methods [5]. However, the objective functions of the
water wave optimization algorithm (WWOA) [12],
OPF problem, which was solved by these
gravitational search algorithm (GSA) [13], artificial
conventional methods, is simple and differentiable.
bee colony approach (ABCA) [14], moth swarm
In fact, the OPF problem in modern power systems is
algorithm (MSA) [15], Jaya algorithm [16]. Besides,
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 529

number of improved version of heuristic optimization (ii) The OPF method based on AEO have been
algorithms have been proposed to improve the successfully implemented for finding the optimal
performance as well as robustness such as self- solution on IEEE 30-bus systems.
adaptive differential evolution (SADE) [17], (iii) The effectiveness of the AEO technique is
modified differential evolution algorithm (MDEA) compared to the implemented methods and different
[18], enhanced genetic algorithm (EGA) [19], exiting methods that prove the effectiveness of AEO
adaptive real coded biogeography-based for the problem of OPF.
optimization (ARCBOA) [20], improved stud krill
herd algorithm (ISKHA) [21], improved grey wolf 2. Problem formulation
optimization (IGWO) [22], modified shuffle frog
OPF is an optimization issue in electric power
leaping algorithm (MSFLA) [23], modified
system operation which minimizes the defined
imperialist competitive algorithm (MICA) [24],
objective functions by adjusting controlled variables
modified artificial bee colony approach (MABCA)
while satisfying all security constraints of electric
[25], improved electromagnetism mechanism
power system [11]. The problem of OPF is
approach (IEMA) [26], modified of sine-cosine
mathematically presented as below
approach (MSCA) [27], hybrid particle swarm
optimization and differential evolution (HPSO-DE)
𝑚𝑖𝑛 = 𝐹(𝑥, 𝑢) (1)
[28], hybrid particle swarm optimization and
gravitational search approach (HPSO-GSA) [29].
Subject to
In generally, these methods have successfully
applied for the OPF problem, however they have
𝑔(𝑥, 𝑢) = 0 (2)
always been a trade-off exploration and exploitation
problem. Therefore, it might be challenging for many
algorithms to obtain balance between exploration and ℎ(𝑥, 𝑢) ≤ 0 (3)
exploitation abilities. Recently, an artificial
ecosystem optimization (AEO) approach developed Where F is the objective function; 𝑔(𝑥, 𝑢) and
based on the flow of energy in the ecosystem is ℎ(𝑥, 𝑢) are equality and inequality constraints,
introduced in [30, 31]. The AEO method utilizes respectively. The state variables vector x and control
three mechanisms in the ecosystem to keep a problem variables vector u can be described as Eq. (4) and Eq.
of balancing exploration and exploitation ability that (5) respectively.
can over local minima. The ecosystem is considered
as a population containing of a production organism, 𝑥 = [𝑃𝐺,𝑠𝑙𝑎𝑐𝑘 , 𝑉𝐿1 , … , 𝑉𝐿𝑁𝐿 , 𝑄𝐺1 , … , 𝑄𝐺𝑁𝐺 ,
a decomposition organism and consumption 𝑆𝑙 . . . , 𝑆𝑁𝑙 ] (4)
organisms. The exploration mission is performed
during the processing of consumption organisms via 𝑢 = [𝑃𝐺𝑖 , … , 𝑃𝐺𝑁𝐺 , 𝑉𝐺1 , … , 𝑉𝐺𝑁𝐺 , 𝑇1 , … , 𝑇𝑁𝑇 ,
updating new solution and selecting the smallest 𝑄𝑐1 , . . . , 𝑄𝑐𝑁𝑐 ] (5)
energy level value. Unlike different many algorithms,
the AEO does not need special control parameters in 2.1. OPF objective functions
the calculation process. The AEO only requires two
external parameters to control is that population size In this study, five target functions including of
and maximum iterations number, so it is simple to fuel cost, emission cost, power loss, voltage
implement and smooth execution. From this deviations and L_index are considered as follows
viewpoint, this paper proposed the AEO technique
2.1.1. Fuel cost
for dealing with the OPF problem with five different
𝑁𝐺
objective functions. The suggested technique is
2
simulated on IEEE-30 bus system. The achieved 𝑂𝐹𝐹 = 𝑓𝑖 (𝑃𝐺𝑖 ) = ∑ 𝑎𝑖 + 𝑏𝑖 𝑃𝑖 + 𝑐𝑖 𝑃𝐺𝑖 (6)
result values of suggested technique compared with 𝑖=1
different techniques that shows the AEO also is an
effective method to solve OPF problem in large scale Where 𝑎𝑖 , 𝑏𝑖 , 𝑐𝑖 are the fuel cost coefficients of
and complex systems. the 𝑖𝑡ℎ generator.
The main contributions of the study can be
briefed as follows: 2.1.2. Emission
(i) The AEO is successfully adjusted for handling the Two important types SOx and NOx of emission
OPF problem with five other target functions. gasses are calculated as the pollutant gasses. The

International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 530

emission gasses generated by each generating unit L index at load bus j can be calculated as follows
may be approximated by a combination of a quadratic
cost and an exponential function of generator active 𝑁𝐺
𝑉𝑖
power output. The emission is defined as Eq. (7) 𝐿𝑗 = |1 − ∑ 𝐹𝑗𝑖 ∠𝜃𝑖𝑗 + 𝜃𝑖 − 𝜃𝑗 |, (12)
𝑉𝑗
𝑖=1
𝑁𝐺
𝑗 = 1, . . . , 𝑁𝐿
2
𝑂𝐹𝐸 = ∑(𝛼𝑖 + 𝛽𝑖 𝑃𝑖 + 𝛾𝑖 𝑃𝐺𝑖 + 𝜉𝑖 (𝜆𝑖 𝑃𝐺𝑖 )) (7)
𝑖=1 Where 𝑉𝑖 , 𝛿𝑖 are voltage and voltage angle at
generator bus 𝑖; 𝑉𝑗 , 𝛿𝑗 are voltage and voltage angle
Where 𝛽𝑖 , 𝛾𝑖 , 𝜉𝑖 are the emission coefficients of at load bus 𝑗; 𝜃𝑖𝑗 is the phase-angle of 𝐹𝑗𝑖 , 𝑁𝐿 is the
the 𝑖𝑡ℎ generator
load buses. The objective function can be given as
2.1.3. Total transmission loss
𝑂𝐹𝐼𝐿 = 𝑚𝑎𝑥( 𝐿𝑗 ) (13)
Total power loss is presented as Eq. (8)
2.2 Constraints
𝑁𝐺 𝑁𝐿

𝑂𝐹𝐿 = ∑ 𝑃𝐺𝑖 − ∑ 𝑃𝐷𝑗 (8) 2.2.1. Equality constraints

𝑖=1 𝑗=1 Constraints on real and reactive power balance
2.1.4. Voltage deviation
𝑃𝑖 = 𝑃𝐺𝑖 − 𝑃𝐷𝑖 =
This objective is to minimize voltage deviation 𝑁
at all load buses and is described as Eq. (9) 𝑉𝑖 ∑|𝑌𝑖𝑗 ||𝑉𝑗 | 𝑐𝑜𝑠( 𝛿𝑖 − 𝛿𝑗 − 𝜃𝑖𝑗 ) (14)
𝑗=1
𝑁𝐿

𝑂𝐹𝑉 = ∑|𝑉𝐿𝑖 − 1.0| (9) 𝑄𝑖 = 𝑄𝐺𝑖 − 𝑄𝐷𝑖 =

𝑁
𝑖=1
𝑉𝑖 ∑|𝑌𝑖𝑗 ||𝑉𝑗 | 𝑠𝑖𝑛( 𝛿𝑖 − 𝛿𝑗 − 𝜃𝑖𝑗 ) (15)
2.1.5. Voltage stability enhancement 𝑗=1
Voltage stability is one of the important
problems which needs to consider for operating of
electric power system. To evaluate voltage stability, 2.2.2. Inequality constraints
L-index known as voltage collapse proximity The limits of power generation
indicator. The bus with the highest L-index value will
be the most vulnerable bus in the system. The L-index 𝑚𝑖𝑛
𝑃𝐺𝑖 𝑚𝑎𝑥
≤ 𝑃𝐺𝑖 ≤ 𝑃𝐺𝑖 , 𝑖 = 1,2, . . . , 𝑁𝐺 (16)
calculation for a power system is presented as follow
[14]. 𝑚𝑖𝑛 𝑚𝑎𝑥
𝑄𝐺𝑖 ≤ 𝑄𝐺𝑖 ≤ 𝑄𝐺𝑖 , 𝑖 = 1,2, . . . , 𝑁𝐺 (17)
𝐼 𝑌 𝑌𝐺𝐿 𝑉𝐺
[ 𝐺 ] = [ 𝐺𝐺 ][ ] (10) The voltage limits of generator buses and load buses
𝐼𝐿 𝑌𝐿𝐺 𝑌𝐿𝐿 𝑉𝐿
𝑚𝑖𝑛 𝑚𝑎𝑥
In which: 𝑉𝐺𝑖 ≤ 𝑉𝐺𝑖 ≤ 𝑉𝐺𝑖 , 𝑖 = 1,2, . . . , 𝑁𝐺 (18)
𝐼𝐺 : Current at generator bus
𝐼𝐿 : Load bus load 𝑉𝐿𝑖𝑚𝑖𝑛 ≤ 𝑉𝐿𝑖 ≤ 𝑉𝐿𝑖𝑚𝑎𝑥 , 𝑖 = 1,2, . . . , 𝑁𝐿 (19)
𝑉𝐺 : Generator bus voltage
𝑉𝐿 : Load bus voltage The capacity of switchable capacitor
Rearrange Eq. (10)
𝑚𝑖𝑛 𝑚𝑎𝑥
𝑄𝑐𝑖 ≤ 𝑄𝑐𝑖 ≤ 𝑄𝑐𝑖 , 𝑖 = 1,2, . . . , 𝑁𝑐 (20)
𝑉 𝑍 𝐹𝐿𝐺 𝐼𝐿
[ 𝐿 ] = [ 𝐿𝐿 ][ ] (11)
𝐼𝐺 𝐾𝐺𝐿 𝑌𝐺𝐺 𝑉𝐺 The limits of transformer tap

With 𝑇𝑘𝑚𝑖𝑛 ≤ 𝑇𝑘 ≤ 𝑇𝑘𝑚𝑎𝑥 , 𝑘 = 1,2, . . . , 𝑁𝑇 (21)

𝐹𝐿𝐺 = −[𝑌𝐿𝐿 ]−1 [𝑌𝐿𝐺 ] The line flow limits of transmission line

International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 531

𝑆𝑙 ≤ 𝑆𝑙𝑚𝑎𝑥 , 𝑙 = 1,2, . . . , 𝑁𝑙 (22) Where, OF is the objective function of each case

(OFF, OFE, OFL, OFV, OFIL) that is defined by
3. Implementation of AEO for solving the equation (6)-(9) and (13). Ip, Iq, Iv, and Is are the
OPF problem penalty coefficients for the inequality constraints of
the state variables.
AEO is a method that are inspired by the flow of Step 4: Update the production organism
energy in a food chain. The AEO utilizes three All rearrange organisms in direction of increasing
mechanisms in the ecosystem to keep a problem energy level. The first organism which has lowest
balancing exploration and exploitation ability. The energy level is chosen as production organism. The
ecosystem is considered as a population containing of production organism is updated as below
a production organism, a decomposition organism
and consumption organisms. The energy level of each 𝑡
organism is based on its fitness value. The organism 𝑠𝑜𝑙1𝑛𝑒𝑤 = 𝑟𝑎𝑛𝑑1 (0,1). [1 − (1 − )] . 𝑠𝑜𝑙𝑏𝑒𝑠𝑡
𝑡𝑚𝑎𝑥
with better fitness value has higher energy level. The 𝑡
step by step of applying of AEO for the OPF problem +(1 − ). [𝑟𝑎𝑛𝑑2 (0,1). (𝑠𝑜𝑙𝑑𝑚𝑎𝑥 − 𝑠𝑜𝑙𝑑𝑚𝑖𝑛 ) +
𝑡𝑚𝑎𝑥
is presented as follow 𝑠𝑜𝑙𝑑𝑚𝑖𝑛 ] (26)
Step 1: Choose control parameter: Ecosystem
size 𝑛, max iteration 𝑡𝑚𝑎𝑥 Step 5: Update the consumption organisms
Step 2: Initialize the ecosystem Update the consumption organisms is based on
Each solution (sol) is considered as an organism three types of herbivores, carnivore and omnivorous.
in the ecosystem and is initialized as follows If the consumption organism is an herbivore, it will
update with the production organism as Eq. (27)
𝑠𝑜𝑙𝑖,𝑑 = 𝑠𝑜𝑙𝑑,𝑚𝑖𝑛 + 𝑟𝑎𝑛𝑑(0,1)(𝑠𝑜𝑙𝑑,𝑚𝑎𝑥 −
𝑠𝑜𝑙𝑑,𝑚𝑖𝑛 ) (23) 𝑠𝑜𝑙𝑖𝑛𝑒𝑤 = 𝑠𝑜𝑙𝑖 + 𝛽𝑐 . (𝑠𝑜𝑙𝑖 − 𝑠𝑜𝑙1𝑛𝑒𝑤 ) ; 𝑖
∈ [2, . . . , 𝑛] (27)
Where, sold,min is the ith solution, d is the number
of control variables, sold,max and sold,min are the upper Where, 𝛽𝑐 is the consumption coefficient
and lower limits of 𝑑th control variable which are determined based on the Levy distribution as follow
defined as Eq. (4).
1 𝑢
𝛽𝑐 = 2 . 𝑢1 ; 𝑢1 , 𝑢2 ~N(0,1) (28)
𝑚𝑖𝑛 𝑚𝑖𝑛 𝑚𝑖𝑛 𝑚𝑖𝑛
𝑠𝑜𝑙𝑑,𝑚𝑖𝑛 = [𝑃𝐺2 , … , 𝑃𝐺𝑁 𝐺
, 𝑉𝐺1 , … , 𝑉𝐺𝑁 𝐺
, 2

𝑚𝑖𝑛 𝑚𝑖𝑛 𝑚𝑖𝑛 𝑚𝑖𝑛

𝑇1 , … , 𝑇𝑁𝑇 , 𝑄𝑐1 , … , 𝑄𝑐𝑁𝑐 ] Where, 𝑁(0,1) is a standard normal distribution.
𝑚𝑎𝑥 𝑚𝑎𝑥 𝑚𝑎𝑥 𝑚𝑎𝑥
(24)
𝑠𝑜𝑙𝑑,𝑚𝑎𝑥 = [𝑃𝐺2 , … , 𝑃𝐺𝑁 𝐺
, 𝑉𝐺1 , … , 𝑉𝐺𝑁 𝐺
If the consumption organism is a carnivore, it will
𝑇 𝑚𝑎𝑥
, … , 𝑇 𝑚𝑎𝑥
, 𝑄 𝑚𝑎𝑥
, … , 𝑄 𝑚𝑎𝑥
] update with another carnivore with higher energy
{ 1 𝑁𝑇 𝑐1 𝑐𝑁𝑐
level as Eq. (29)
Step 3: Solve power flow and evaluate the fitness 𝑠𝑜𝑙𝑖𝑛𝑒𝑤 = 𝑠𝑜𝑙𝑖 + 𝛽𝑐 . (𝑠𝑜𝑙𝑖 − 𝑠𝑜𝑙𝑗 ) (29)
function using Eq. (25) based on the energy level of
the organism soli. The organism with the smallest
If the consumption organism is omnivorous, it
fitness value is considered as the best organism in the
will update with a producer and a carnivore with
ecosystem (𝑠𝑜𝑙𝑏𝑒𝑠𝑡 ).
higher energy level as Eq. (30)
(0) 𝑙𝑖𝑚 2
𝑂𝐹𝑖 = 𝑂𝐹 + 𝐼𝑃 (𝑃𝐺1 − 𝑃𝐺1 ) 𝑠𝑜𝑙𝑖𝑛𝑒𝑤 = 𝑠𝑜𝑙𝑖 + 𝛽𝑐 . [𝑟𝑎𝑛𝑑2 (0,1). (𝑠𝑜𝑙𝑖 − 𝑠𝑜𝑙1𝑛𝑒𝑤 )
𝑁𝐺
+1 − 𝑟𝑎𝑛𝑑2 (0,1). (𝑠𝑜𝑙𝑖 − 𝑠𝑜𝑙𝑗 )] (30)
𝑙𝑖𝑚 2
+ 𝐼𝑞 ∑(𝑄𝐺𝑖 − 𝑄𝐺𝑖 )
𝑖=1 Step 6: Solve power flow and evaluate the fitness
𝑁𝐿
function using Eq. (25) based on the energy level of
+ 𝐼𝑣 ∑(𝑉𝐿𝑖 − 𝑉𝐿𝑖𝑙𝑖𝑚 )2 the new organisms sol𝑛𝑒𝑤
𝑖
𝑖=1 Step 7: Replace all organisms if new organisms
𝑁𝑇𝐿
have better fitness values
+ 𝐼𝑠 ∑(𝑆𝑙 − 𝑆𝑙,𝑚𝑎𝑥 )2 (25) If new organisms have the better quality than the
𝑖=1 corresponding ones in the ecosystem, the selective
mechanism is used to update the ecosystem and the

International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 532

Step 8: Update the whole ecosystem by

decomposition mechanism
Organisms that die will be decomposed by a
decomposition organism. Therefore, each organism
in the ecosystem will update with the decomposition
organism as follows

𝑠𝑜𝑙𝑖𝑛𝑒𝑤 = 𝑠𝑜𝑙𝑖 + 3. 𝛽𝑑𝑟 . (𝜎1 . 𝑠𝑜𝑙𝑏𝑒𝑠𝑡 − 𝜎2 . 𝑠𝑜𝑙𝑖 )

(31)

Where, 𝛽𝑑𝑟 is the decomposition rate determined

by 𝛽𝑑𝑟 ~𝑁(0,1). 𝜎1 and 𝜎2 are weight factors which
are determined by the below equations

𝜎 = 𝑟𝑎𝑛𝑑3 (0,1). 𝑟𝑎𝑛𝑑𝑖([1,2]) − 1

{ 1 (32)
𝜎2 = 2. 𝑟𝑎𝑛𝑑3 (0,1) − 1

Step 9: Solve power flow and evaluate the fitness

function using Eq. (25). The ecosystem and the best
organism 𝑠𝑜𝑙𝑏𝑒𝑠𝑡 are also updated.
Step 10: Check stopping criteria. If (𝑡 < 𝑡𝑚𝑎𝑥 )
Figure. 1 The IEEE 30-bus system return to step 5 with 𝑡 = 𝑡 + 1, otherwise go to next
step.
Step 11: End. Export best organism
The best organism 𝑠𝑜𝑙𝑏𝑒𝑠𝑡 corresponding to the
value of the fitness function 𝐹𝑏𝑒𝑠𝑡 is considered as the
result of the OPF problem.

4. Numerical results
The OPF problem with five different objective
functions based on AEO method is developed on
Matlab software to determine the optimal solution for
IEEE-30 bus system. Furthermore, the obtained
results using AEO are also compared with other
Figure 2. Convergence rate of the AEO and other studies to prove the effectiveness of the OPF problem
methods for case 1 method based on AEO.
The IEEE 30-bus system consist of six
generators, 24 load buses and 41 lines as Fig 1. Bus
1, 2, 5, 8, 11 and 13 are generator buses. Line 6–9, 6–
10, 4–12 and 27–28 is tap changer of transformers. In
addition, bus 10, 12, 15, 17, 20, 21, 23, 24 and 29 are
reactive power sources. System data is given in [25,
32]. The generator buses voltage bound is within 0.95
and 1.1 p.u, while load bus voltage limits are 0.95 and
1.05 p.u, the tap ratio bound is 0.9 and 1.1 p.u, the
reactive power sources capacity is [0-5] MVAr. The
generation cost and emission coefficients of IEEE-30
bus system given in Table 1.
Figure 3. Fuel cost obtained in 50 runs using AEO and The Table 2 is presented the control parameters
other approaches and optimal value obtained using AEO method with
five targets including of fuel cost, emission, active
best organism 𝑠𝑜𝑙𝑏𝑒𝑠𝑡 is also updated after the power loss, voltage deviations and L_index. The
ecosystem updated. values obtained of AEO method compared with EO,
PSO, SFO and GA methods for every the objective
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 533

Table 1. IEEE 30-bus system: fuel cost and emissions constants [25, 32]
Generator Bus a b c α β γ ζ λ
G1 1 0 2 0.00375 4.091 −5.554 6.490 0.0002 2.857
G2 2 0 1.75 0.0175 2.543 −6.047 5.638 0.0005 3.333
G5 5 0 1 0.0625 4.258 −5.094 4.586 0.000001 8
G8 8 0 3.25 0.00834 5.326 −0.0355 3.380 0.002 2
G11 11 0 3 0.025 4.258 −0.05094 4.586 0.000001 8
G13 13 0 3 0.025 6.131 −0.05555 5.151 0.00001 6.667

Table 2. The control parameters and optimal value obtained using AEO method for five targets in IEEE 30-bus system
Limits Proposed AEO Method
Control Emissions Power Loss Voltage deviation L_index
Fuel cost
parameters Min Max Case 2 Case 3 Case 4 Case 5
Case 1
P1(MW) 50 200 177.1564 64.2336 51.6600 100.6993 162.9103
P2(MW) 20 80 49.1070 67.4432 79.9791 77.6726 45.1485
P5(MW) 15 50 21.3021 49.9976 49.9995 38.9237 19.8578
P8(MW) 10 35 21.0267 34.9994 34.9861 20.4162 24.8632
P11(MW) 10 30 11.8464 29.9978 29.9344 22.4628 18.5628
P13(MW) 12 40 12.0188 40.0000 39.9635 29.4494 20.6263
V1 (p.u) 0.95 1.1 1.0847 1.0600 1.0641 1.0065 1.0775
V2 (p.u) 0.95 1.1 1.0654 1.0565 1.0584 1.0027 1.0617
V5 (p.u) 0.95 1.1 1.0335 1.0376 1.0363 1.0177 1.0708
V8 (p.u) 0.95 1.1 1.0381 1.0434 1.0437 1.0098 1.0535
V11(p.u) 0.95 1.1 1.0843 1.0998 1.0959 1.0289 1.0873
V13(p.u) 0.95 1.1 1.0453 1.0650 1.0533 1.0102 1.0413
T11 0.9 1.1 1.0279 1.0733 1.0479 1.0387 1.0166
T12 0.9 1.1 0.9361 0.9315 0.9307 0.9060 0.9579
T15 0.9 1.1 0.9757 0.9999 0.9923 0.9910 0.9472
T36 0.9 1.1 0.9756 0.9825 0.9860 0.9714 0.9608
QC10 (MVAR) 0 5 2.3262 2.7505 1.4101 4.8911 1.6256
QC12 (MVAR) 0 5 3.9324 0.7183 3.6882 4.3993 0.2223
QC15 (MVAR) 0 5 4.4115 2.6664 3.2613 4.7698 0.9287
QC17 (MVAR) 0 5 2.9116 1.6783 2.7524 1.7478 0.7723
QC20 (MVAR) 0 5 4.6717 3.7098 1.5721 4.9801 3.6823
QC21 (MVAR) 0 5 4.9008 1.9602 4.9354 4.8403 2.6649
QC23 (MVAR) 0 5 3.5063 3.8141 4.7714 4.9578 1.9901
QC24 (MVAR) 0 5 4.8725 4.9986 4.6153 4.9911 0.6793
QC29 (MVAR) 0 5 2.6080 2.9194 3.1888 3.5958 0.1264
Total cost ($/h) - - 800.5454 944.2521 967.4142 874.4066 807.3090
Emissions - - 0.3664 0.2048 0.2073 0.2366 0.3264
PLoss (MW) - - 9.0574 3.2715 3.1225 6.2241 8.5690
VD - - 0.9215 0.8260 0.8736 0.0924 0.8711
L_index - - 0.1384 0.1394 0.1394 0.1491 0.1370

are detail descibled in Table 3. From Table 3, it can term of optimal value. For case 2, total emission of
be noted that, the total generator cost (case 1) is AEO approach is approximate EO, PSO approach
achieved 800.5454 ($/h) using AEO algorithm, and is reduced than SFO, GA method as shown in
which is better than EO, PSO, SFO and GA methods. Table 3. The AEO, EO, PSO approaches are total
The total generator achieved of EO method is emission 0.2048 (ton/h), while SFO and GA
800.6220 ($/h), PSO method is 800.5924 ($/h), SFO algorithm is 0.2163 (ton/h) and 0.2050 (ton/h)
method is 805.3635 ($/h) and GA method is 800.7742 respectively. As noted in Fig. 4, the standard
($/h). The convergence rate of the fuel cost function deviation in 50 runs of the AEO algorithm is better
and obtained value in 50 runs using the AEO and than compared with the PSO, SFO and GA method
other methods are demonstrated in Fig. 2, Fig.3. As and similar as the EO method.
observed from those Figure, ability convergence and With case 3, as shown in Table 3, the active
standard deviation of AEO algorithm is better than power losses is decreased to 3.1225 (MW) using the
compared with EO, PSO, SFO and GA methods in AEO. From Table 3, it can be seen that, the total
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 534

Table 3. The results obtained of AEO method compared with EO, PSO, SFO and GA method with case 1-5
Case 1 Fuel Cost ($/h)
Algorithm Min Average Max Standard deviation
AEO 800.5454 800.6614 800.9169 0.0742
EO 800.6220 801.2520 808.3032 1.1501
PSO 800.5924 3.8412 x107 1.2888 x108 3.1887 x1015
3
SFO 805.3635 909.4241 5.2008 x10 619.4435
GA 800.7742 802.2623 806.8618 1.4086
Case 2 Emission (Ton/h)
AEO 0.2048 0.2049 0.2052 6.1577 x10-5
EO 0.2048 0.2049 0.2051 4.5183 x10-5
7 7
PSO 0.2048 3.6719 x10 1.2698 x10 5.4718 x107
SFO 0.2163 0.2525 0.3101 0.0238
GA 0.2050 0.2060 0.2076 6.6487 x10-4
Case 3 Power Loss (MW)
AEO 3.1225 3.1980 3.3315 0.0505
EO 3.1048 3.3008 4.6138 0.2358
PSO 3.1012 3.1781 x107 1.2789 x108 5.2204 x107
3 5
SFO 4.8875 2.0842 x10 1.0389 x10 1.4691 x104
GA 3.1847 3.4366 5.2477 0.2927
Case 4 Voltage deviation
AEO 0.0924 0.1000 0.1149 0.0060
EO 0.1018 0.1207 0.1564 0.0140
PSO 0.0904 4.1989 x107 1.3034 x108 5.6051 x107
3 5
SFO 0.2108 8.8620 x10 4.4308 x10 6.2661 x104
4 6
GA 0.1183 6.7222 x10 3.3611 x10 4.7533 x105
Case 5 L_index
AEO 0.1370 0.1376 0.1388 4.0761 x10-4
EO 0.1372 0.1388 0.1408 8.2925 x10-4
7 8
PSO 0.1373 4.0918 x10 1.2667 x10 5.5296 x107
SFO 0.1399 0.1454 0.1562 0.0043
GA 0.1380 0.1406 0.1474 0.0017

Figure 4. Emission obtained in 50 runs using AEO and Figure 5. Power loss obtained in 50 runs using AEO and
other methods other methods

power loss of the AEO technique achieves better Fig. 7 are presented obtained values in 50 runs of the
minimum compared with the SFO and GA methods. AEO method for case 4 and case 5, respectively. It
Although power loss obtained from AEO is less than can be observed that from the Table 3 and those Fig,
EO and PSO method, however the AEO algorithm the AEO algorithm can obtained better voltage
has better average and standard deviation values deviation and L_index values with smaller standard
compared with EO and PSO method. This shows the deviation compared to EO, PSO, SFO and GA
suggested method’s effective wih ability to obtain methods. Furthermore, the results simulation show
optimized solution as shown in Fig. 5. Also, Fig. 6, that application of the AEO for the OPF problem that

International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 535

Table 6. The solution obtained from AEO for TPL

objective compared with different algorithms
Method TPL (MW) IL (%)
Initial 5.830 46.44
DE [8] 3.38 7.618
SKHA [11] 3.1409 0.586
ABCA [14] 3.1078 0
EGA [19] 3.2008 2.446
ARCBOA [20] 3.1009 0
Proposed AEO 3.1225 -

Table 7. The solution obtained from AEO for voltage

deviation objective compared with different algorithms
Method VD (Pu) IL (%)
Figure 6. Voltage profile obtained in 50 runs using AEO Initial 1.1747 92.13
and other methods GWO [8] 0.118736 22.18
BOA [9] 0.0951 2.839
GSA [13] 0.0932 0.858
Jaya [16] 0.1273 27.41
EGA [19] 0.911 89.85
ARCBOA [20] 0.092 0
ISKHA [21] 0.1029 10.20
MICA [24] 0.0952 2.941
MABC [25] 0.1017 9.145
IEMA [26] 0.1063 13.07
SCA [27] 0.1082 14.60
MSCA [27] 0.103 10.29
HPSO-GSA [29] 0.1267 27.07
Proposed AEO 0.0924 -

Figure 7. L_index obtained in 50 runs using AEO and Table 8. The solution obtained from AEO for L_index
other methods objective compared with different algorithms
Method L index (Pu) IL (%)
Table 4. The solution obtained from AEO for fuel cost ABCA [14] 0.1379 0.652
objective compared with different algorithms ARCBOA [20] 0.1369 0
Method Fuel Cost ($/h) IL (%) Proposed AEO 0.1370 -
Initial 901.6391 11.21
TS [6] 802.29 0.217 significantly enhances the performance of power
EP [7] 802.62 0.258 systems. The fuel cost, emissions, active power loss
DE [8] 801.23 0.085 and voltage deviations levels are reduced by 11.21%,
TLOA [10] 800.7257 0.022
44.06%, 46.44%, and 92.13% respectively, compared
SKHA [11] 801.4675 0.115
MDEA [18] 802.375 0.228 to the initial case.
IGWO [22] 801.259 0.089 With the purpose of evaluate effective of the
MSFLA [23] 802.287 0.217 susgested AEO method, the authors compare the
HPSO-DE [28] 802.248 0.212 objective functions including of fuel cost, emission,
Proposed AEO 800.5454 - power loss, voltage deviations and L_index from the
suggested AEO approach to those other methods. The
Table 5. The solution obtained from AEO for emission best values achieved for the objectives using the
objective compared with different algorithms proposed technique and other technique are listed in
Method Emission (Ton/h) IL (%) Table 4-8.
Initial 0.3661 44.06 As observed in the Table 4, total fuel cost
SKHA [11] 0.20508 0.136 achieved using the AEO method is reduced
ABCA [14] 0.204826 0.127
insignificantly as compared with TS [6], EP [7], DE
MSA [15] 0.20482 0.010
ARCBOA [20] 0.2048 0
[8], TLOA [10], SKHA [11], MDEA [18], IGWO
ISKHA [21] 0.204818 0.009 [22], MSFLA [23], and HPSO-DE [28] approaches.
Proposed AEO 0.2048 - The improvement level (IL) in % can be up to
0.2285 % for fuel cost objective. The four other
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 536

objectives including emission, power loss, voltage IEEE Transaction on Power System, Vol. 1, No.
deviation and L_index using the AEO have obtained 3, pp. 31-39, 1986.
value better or equal compared with diffrent [2] O. Alsac and B. Scott, “Optimal Load Flow with
approaches as seen in Table 5-7, respectively. The Steady State Security”, IEEE Transaction on
improvement level percentage of the proposed AEO Power Apparatus and System, Vol. 93, No. 3, pp.
algorithm to 0.137% for emission objective, 7.618% 745-751.
for total power loss objective, 89.85% for voltage [3] D. I. Sun, B. Ashley, B. Brewer, A. Hughes, and
deviation objective and 0.652% for L_index W. F. Tinney, “Optimal Power Flow by Newton
objective. The comparison results indicates that the Approach”, IEEE Transaction on Power
ability quickly convergence of AEO technique with Apparatus and System, Vol. 103, No. 10, pp.
the optimal value. This is demonstration the robust of 2864-2875, 1984.
the AEO technique in dealing with OPF problem. [4] R. C. Burchett, H. H. Happ and D. R. Vierath,
“Quadratically Convergent Optimal Power
5. Conclusion Flow”, IEEE Transaction on Power Apparatus
and System, Vol. 103, No. 11, pp. 3267-3276,
OPF is one of important issues for operating of
1984.
power system and it might be challenging for many
[5] X. Yan and V. H. Quintana, “Improving an
algorithms to handle with OPF problem, especially in
Interior Point Based OPF by Dynamic
complex systems. In this paper, the AEO is
Adjustments of Step Sizes and Tolerances”,
successfully adjusted for handling the problem of
IEEE Transaction on Power System, Vol. 14, No.
OPF with five other target functions. The optimal
2, pp. 709-717, 1999.
values and success rate obtained by the proposed
[6] M. A. Abido, “Optimal Power Flow Using Tabu
AEO algorithm are the better or same ompared with
Search Algorithm”, Electric Power Components
EO, PSO, SFO and GA method. Furthermore, the fuel
System, Vol. 30, No. 5, pp. 469-483, 2002.
cost, emissions, active power loss and voltage
[7] Y. Jason and P. W. Kit, “Evolutionary
deviations levels are reduced by 11.21%, 44.06%,
Programming Based Optimal Power Flow
46.44%, and 92.13% respectively, compared to the
Algorithm”, IEEE Transactions on Power
initial case. For other exiting methods, the
Systems, Vol. 14, No. 4, pp. 1245-1250,1999.
improvement level percentage (IL) of the proposed
[8] A. A. E. Fergany and H. M. Hasanien, “Single
AEO algorithm can be up to 0.2285 % for fuel cost
and Multi-Objective Optimal Power Flow Using
objective, 0.137% for emission objective, 7.618% for
Grey Wolf Optimizer and Differential Evolution
total power loss objective, 89.85% for voltage
Algorithms”, Electric Power Components
deviation objective and 0.652% for L_index
System, Vol. 43, No. 13, pp. 1548-1559, 2015.
objective. The simulation results demonstrate that,
[9] A. Bhattacharya and P. K. Chattopadhyay,
the AEO also is one of effective and reliable methods
“Application of Biogeography-Based
for dealing problem of OPF in large scale and
Optimisation to Solve Different Optimal Power
complex systems such as the OPF problem
Flow Problems”, IET Generator Transmission
incorporting renewable rnergy, FACTS.
Distribution, Vol. 5, No. 1, pp. 70-80, 2011.
[10] M. R. Nayak, C. K. Nayak, and P. K. Rout,
Conflicts of interest “Application of Multi-Objective Teaching
The authors declare no conflict of interest. Learning Based Optimization Algorithm to
Optimal Power Flow Problem”, Procedia
Author contributions Technology, Vol. 6, pp. 255-264, 2012.
[11] H. Pulluri, R. Naresh, and V. Sharma, “A
Conceptualization, T.L.D; methodology, T.L.D;
Solution Network Based on Stud Krill Herd
software, N.A.N; validation, T.L.D, N.A.N, and
Algorithm for Optimal Power Flow Problems”,
T.T.N; formal analysis, T.L.D and T.T.N;
Soft Computer, Vol. 22, No. 1, pp. 159-176,
investigation, N.A.N; writing-original draft
2018
preparation, T.L.D; writing-review and editing,
[12] M. Siva1, R. Balamurugan, and L.
T.L.D; visualization, N.A.N; supervision, T.L.D
Lakshminarasimman, “Water Wave
Optimization Algorithm for Solving Economic
References Dispatch Problems with Generator Constraints”,
[1] R. M. Palomino and V. H. Quintana, “Sparse International Journal of Intelligent Engineering
Reactive Power Scheduling by a Penalty- and Systems, Vol. 9, No. 4, pp. 31-40, 2016.
Function Linear Programming Technique”, [13] S. Duman, U. Gu¨venc, Y. So¨nmez, and N.
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
Received: September 3, 2021. Revised: October 7, 2021. 537

Yo¨ru¨keren, “Optimal Power Flow Using [24] A. K. Murtadha and A. B. Layth, “Modified
Gravitational Search Algorithm”, Energy Artificial Bee Colony Optimization Technique
Convers Manager, Vol. 59, No. 7, pp. 86-95, with Different Objective Function of Constraints
2012. Optimal Power Flow”, International Journal of
[14] M. R. Adaryani and A. Karami, “Artificial Bee Intelligent Engineering and Systems, Vol. 13,
Colony Algorithm for Solving Multi-Objective No. 4, pp. 378-388, 2020
Optimal Power Flow Problem”, International [25] M. Ghasemi, S. Ghavidel, M. M. Ghanbarian, M.
Journal Electric Power Energy System, Vol. 53, Gharibzadeh, and A. A. Vahed, “Multi-
No. 13, pp. 219–230, 2013. Objective Optimal Power Flow Considering the
[15] A. A. A. Mohamed, Y. S. Mohamed, A. A. E. Cost, Emission, VD and TPL Using Multi-
Gaafary, and A. M. Hemeida, “Optimal Power Objective Modified Imperialist Competitive
Flow Using Moth Swarm Algorithm”, Electric Algorithm”, Energy, Vol. 78, pp. 276-289, 2014.
Power System Resarch, Vol. 142, No. 1, pp. 190- [26] H. R. E. H. Bouchekara, M. A. Abido, and A. E.
206, 2017. Chaib, “Optimal Power Flow Using an
[16] W. Warid, H. Hizam, N. Mariun, and N. I. A. Improved Electromagnetism-Like Mechanism
Wahab, “Optimal Power Flow Using the Jaya Method”, Electric Power Components System,
Algorithm”, Energies, Vol. 9, No. 9, pp. 678, Vol. 44, No. 4, pp. 434-449, 2016.
2016. [27] A. F. Attia, R. A. E. Sehiemy, and H. M.
[17] C. Thitithamrongchai and B. E. Arporn, “Self- Hasanien, “Optimal Power Flow Solution in
Adaptive Differential Evolution Based Optimal Power Systems Using a Novel Sine-Cosine
Power Flow for Units with Nonsmooth Fuel Algorithm”, International Journal Electric
Cost Functions”, Journal Electric System, Vol. Power Energy System, Vol. 99, No. 7, pp. 331-
3, No. 2, pp. 88-99, 2007. 343, 2018.
[18] S. Sayah and K. Zehar, “Modified Differential [28] P. M. Le, T. L. Duong, D. N. Vo, T. T. Le, and
Evolution Algorithm for Optimal Power Flow S. Q. Nguyen, “An Efficient Hybrid Method for
with Non-smooth Cost Functions”, Energy Solving Security-Constrained Optimal Power
Convers Manager, Vol. 49, No. 11, pp. 3036- Flow Problem”, International Journal on
3042, 2008. Electrical Engineering and Informatics, Vol. 12,
[19] M. S. Kumari and S. Maheswarapu, “Enhanced No. 4, pp. 933-955, 2020.
genetic algorithm based computation technique [29] J. Radosavljević, D. Klimenta, M. Jevtić, and N.
for multi-objective optimal power flow Arsić, “Optimal Power Flow Using a Hybrid
solution”, International Journal Electric Power Optimization Algorithm of Particle Swarm
Energy System, Vol. 32, No. 6, pp. 736-742, Optimization and Gravitational Search
2010. Algorithm”, Electric Power Components and
[20] A. R. Kumar, and L. Premalatha, “Optimal Systems, Vol. 43, No. 17, pp. 1958-1970, 2015.
Power Flow for a Deregulated Power System [30] W. Zhao, L. Wang, and Z. Zhang, “Artificial
Using Adaptive Real Coded Biogeography- Ecosystem-Based Optimization: A Novel
Based Optimization”, International Journal Nature-Inspired Meta-Heuristic Algorithm”,
Electric Power Energy System, Vol. 73, No. 12 , Neural Computing and Applications, Vol. 32,
pp. 393-399, 2015. No. 13, pp. 9383-9425, 2020.
[21] G. Chen, Z. Lu, and Z. Zhang, “Improved Krill [31] S. Mouassa, F. Jurado, T. Bouktir, M. A. Z. Raja,
Herd Algorithm with Novel Constraint Handling “Novel Design of Artificial Ecosystem
Method for Solving Optimal Power Flow Optimizer for Large-Scale Optimal Reactive
Problems”, Energies, Vol. 11, No. 1, pp. 76, Power Dispatch Problem with Application to
2018. Algerian Electricity Grid”, Neural Computing
[22] M. Abdo, S. Kamel, M. Ebeed, J. Yu, and F. and Applications, Vol. 33, pp. 7467-7490, 2021.
Jurado, “Solving Nonsmooth Optimal Power [32] R. D. Zimmerman, M. Sa´nchez, and R. J.
Flow Problems Using a Developed Grey Wolf Thomas, https://www.pserc.cornell.edu/
Optimizer”, Energies, Vol. 11, No. 7, pp. 1692, matpower, Accessed 1 , 2018.
2018.
[23] T. Niknam, M. R. Narimani, M. Jabbari, and A. Appendix
R. Malekpour, “A Modified Shuffle Frog
Nomenclature and abbreviations
Leaping Algorithm for Multi-Objective Optimal PG,slack active power of the slack generator
Power Flow”, Energy, Vol. 36, No. 11, pp. 6420- VL magnitude voltage of the load bus
6432, 2011.
International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47
 Received: September 3, 2021. Revised: October 7, 2021. 538

𝑄𝐺 reactive power of the generators

Sl transparent power flow in line
PG active power of the generator
Tk tap changer of the transformer
QC reactive power of shunt compensator
VG magnitude voltage of the generator bus
N Toal number of buses
NL number of load buses.
NG number of generating units
NT number of regulating transformers.
Nl number of line.
Nc number of shunt compensators
Vi,Vj magnitude voltages of the buses i, j .
Pi active power injection into 𝑖 𝑡ℎ bus
Qi reactive power injection into 𝑖 𝑡ℎ bus
PGi active generated at bus i
QGi reactive power generated at bus i
PDj load active power at bus j
QDj load reactive power at bus i
PGi,min min. active power limit of generator
QGi,min min. reactive power limit of generator
PGi,max max. active power limit of generator
QGi,max max. reactive power limit of generator
VGi,min min. voltage of the generator bus i
VGi,max max. voltage of the generator bus i
VLi,min min. voltage of the load bus i
VLi,max max. voltage of the load bus i
Qci,min min. reactive power compensative devise at
load bus i
Qci,max max. reactive power compensative devise at
load bus i
Ti,min min. tap changer of the transformer
Ti,max max. tap changer of the transformer
Sl,max max. line flow limits of transmission line

International Journal of Intelligent Engineering and Systems, Vol.14, No.6, 2021 DOI: 10.22266/ijies2021.1231.47

View publication stats