Volume 04 Issue 12-2024
20
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
ABSTRACT
Optimal Power Flow (OPF) is a critical problem in power system operation and planning, aimed at determining the
most efficient operational conditions of the system while respecting various operational constraints. Genetic
Algorithms (GA), with their ability to solve complex optimization problems, have been increasingly employed to
address the OPF problem. This study focuses on performing a parametric analysis to investigate the impact of
selecting appropriate control and state variables on the efficiency and effectiveness of GA-based OPF solutions.
Various combinations of control variables (such as generator voltages, active power generation, and reactive power
generation) and state variables (such as bus voltages and branch power flows) are analyzed in this study. The results
highlight how the selection of control and state variables influences the convergence rate, computational time, and
solution accuracy of the genetic algorithm. A series of parametric studies are conducted to optimize the parameters
of the genetic algorithm, including population size, crossover rate, and mutation rate, to improve the overall
performance of the OPF model. The study demonstrates the significance of variable selection in achieving more
efficient and practical solutions for power system optimization. The findings suggest that the choice of control and
state variables plays a crucial role in balancing the trade-offs between solution quality and computational efficiency.
KEYWORDS
Optimal Power Flow, Genetic Algorithm, Control Variables, State Variables, Parametric Study, Power System
Optimization, Computational Efficiency, Variable Selection, Crossover Rate, Mutation Rate, Power System Operation.
INTRODUCTION
Research Article
A PARAMETRIC STUDY ON OPTIMAL POWER FLOW USING GENETIC
ALGORITHMS: SELECTION OF CONTROL AND STATE VARIABLES
Submission Date:
November 24, 2024,
Accepted Date:
November 29, 2024,
Published Date:
December 04, 2024
Rajat Raikwar
Lokmanya Tilak College of Engineering, Mumbai University, Navi Mumbai, India
Journal
Website:
https://theusajournals.
com/index.php/ajast
Copyright:
Original
content from this work
may be used under the
terms of the creative
commons
attributes
4.0 licence.
Volume 04 Issue 12-2024
21
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
Optimal Power Flow (OPF) is a fundamental problem in
power systems engineering, aiming to determine the
most efficient operation of a power grid while
satisfying a set of operational constraints. These
constraints typically include limits on generation,
voltage levels, transmission line capacities, and other
system parameters. The goal is to optimize an
objective function, often the cost of generation or
system losses, while maintaining system reliability and
performance. Solving the OPF problem has become
increasingly important with the growing complexity of
modern power grids, which involve large-scale
generation, renewable energy sources, and dynamic
demand patterns.
Traditional methods for solving OPF, such as linear
programming, nonlinear programming, and quadratic
programming, have limitations when dealing with
large, nonlinear, and non-convex optimization
problems that are common in real-world power
systems. As a result, alternative optimization
techniques, particularly heuristic algorithms, have
gained popularity in recent years. Among these,
Genetic Algorithms (GA) have emerged as a promising
tool due to their robustness in handling complex,
multidimensional optimization problems. GAs, inspired
by the process of natural selection, use evolutionary
strategies such as selection, crossover, and mutation
to explore the solution space and converge to an
optimal or near-optimal solution.
However, the performance of Genetic Algorithms in
OPF problems is heavily influenced by the selection of
control variables (such as generator outputs, reactive
power, and voltage levels) and state variables (such as
bus voltages, branch power flows, and other system
parameters). The choice of these variables can
significantly impact the algorithm’s convergence
speed, computational efficiency, and the quality of the
optimal solution. Despite its importance, systematic
studies addressing the effect of control and state
variable selection on the GA-based OPF solutions
remain limited.
This study aims to fill this gap by performing a
parametric analysis to explore how the selection of
control and state variables affects the efficiency and
performance of GA in solving the OPF problem. The
focus is on investigating different combinations of
these variables to determine the most effective set
that ensures an optimal balance between solution
quality and computational effort. Additionally, the
study evaluates the influence of key GA parameters,
such as population size, crossover rate, and mutation
rate, on the overall performance of the OPF model. By
systematically varying these parameters and variable
selections, the study seeks to provide insights into
optimizing the application of GAs for OPF in power
systems.
Through this analysis, the research aims to contribute
to the improvement of GA-based OPF models, offering
more efficient, reliable, and scalable solutions to the
increasingly complex challenges in modern power
system operation and optimization.
METHODOLOGY
This study employs a systematic parametric analysis to
investigate the impact of control and state variable
selection on the performance of Genetic Algorithm
(GA) in solving the Optimal Power Flow (OPF) problem.
The primary objective is to understand how varying
combinations of control and state variables influence
the convergence speed, computational efficiency, and
the quality of the OPF solution. The method consists of
four main phases: problem formulation, GA
parameterization, case study selection, and analysis.
Volume 04 Issue 12-2024
22
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
Problem
Formulation:
The
OPF
problem
is
mathematically formulated as an optimization
problem where the objective is to minimize a cost
function, typically the fuel cost or system losses, while
respecting a set of physical and operational
constraints. These constraints include generation
limits, voltage magnitude limits at buses, power flow
limits on transmission lines, and reactive power limits.
For this study, control variables include generator
active and reactive powers, and state variables include
bus voltages and branch power flows. These variables
are selected to reflect typical operational conditions
and to align with common practices in power system
operation.
Genetic Algorithm Parameterization: A standard
Genetic Algorithm framework is used to solve the OPF
problem. The GA operates through a series of
evolutionary steps, starting with an initial population
of potential solutions, followed by selection,
crossover,
mutation,
and
replacement.
The
parameters of the GA are critical to its performance
and are tuned as part of the parametric study. The key
GA parameters considered are:
Population Size: The number of candidate solutions in
each generation.
Crossover Rate: The probability of combining two
solutions to create offspring.
Mutation Rate: The probability of introducing random
changes in a solution to maintain genetic diversity.
Selection Mechanism: A method (such as tournament
or roulette-wheel selection) to choose parent solutions
for the next generation.
The GA is implemented with standard operators like
single-point or multi-point crossover, and bit-flipping
mutation. These parameters are varied systematically
to assess their impact on convergence speed and
solution quality.
Case Study Selection: To conduct the parametric
analysis, two well-known benchmark power system
networks are used: the IEEE 30-bus system and the
IEEE 57-bus system. These systems are selected
because they represent different levels of complexity,
making them suitable for evaluating the effectiveness
of the GA approach in both small and medium-sized
power networks. Each case study is formulated by
defining the control variables (such as generator
outputs and reactive power injections) and state
variables (such as bus voltages and branch flows),
along with operational constraints as discussed earlier.
The case studies allow for a detailed examination of
the performance of the GA when applied to practical
and scalable power system models.
Parametric Analysis and Data Collection: The study
investigates different combinations of control and
state variables by running multiple simulations for each
case study. The following steps are performed for each
simulation:
Control Variables: Various combinations of active
power generation, reactive power generation, and
voltage settings are chosen as control variables.
State Variables: Different sets of state variables,
including bus voltages, branch power flows, and line
currents, are considered to examine their influence on
the GA’s performance.
GA Performance: For each combination of variables,
the GA is executed, and performance metrics such as
convergence rate, computational time, and the final
solution quality are recorded. The convergence rate is
measured by the number of generations required to
Volume 04 Issue 12-2024
23
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
achieve a solution within a predefined error margin,
while computational time reflects the total time taken
to reach the optimal solution.
To ensure robustness and accuracy, each simulation is
run multiple times with different random seeds, and
the average results are recorded. This helps to mitigate
the impact of randomness in the GA process and
ensures that the results are statistically significant.
Analysis and Evaluation: After collecting the data, the
performance of different combinations of control and
state variables is evaluated using a set of performance
indicators:
Convergence Speed: The number of generations or
iterations required for the algorithm to reach a
satisfactory solution.
Solution Quality: The closeness of the obtained
solution to the theoretical optimal or best-known
solution.
Computational Efficiency: The total computational
time taken to find the optimal solution, which is an
important factor when implementing the GA in real-
time power system operations.
The results are analyzed to identify trends, such as
which combinations of control and state variables yield
the best balance between solution quality and
computational effort. The analysis also includes the
effects of varying GA parameters (population size,
crossover rate, and mutation rate) on the overall
performance, providing insights into the optimization
of GA settings for the OPF problem.
RESULTS
The results of the parametric study on the application
of Genetic Algorithms (GA) to the Optimal Power Flow
(OPF) problem show the following key observations:
Control and State Variables Impact: The selection of
control and state variables significantly influenced the
performance of the GA in terms of convergence speed
and solution quality. It was found that when a larger
set of control variables (e.g., both active and reactive
power generation) was used, the algorithm achieved a
more precise optimization of the system but at the cost
of increased computation time. For instance, when
only active power generation was considered as a
control variable, the solution was reached more
quickly, but the quality of the solution was suboptimal.
In contrast, including reactive power control improved
the solution's accuracy, albeit with a higher
computational burden.
Convergence Speed: The GA showed faster
convergence when fewer state variables (such as
fewer bus voltages and line flows) were included in the
optimization. However, limiting the number of state
variables often resulted in a less accurate
representation of the power system, leading to
suboptimal operational decisions. The most effective
combinations for fast convergence included moderate
numbers of state variables, balancing the need for
solution accuracy with computational efficiency.
Computational Efficiency: Computational time was
inversely related to the number of variables
considered. Smaller population sizes and fewer state
variables generally resulted in quicker solutions.
However, the trade-off was that the optimality of the
solutions was compromised, especially for more
complex systems like the IEEE 57-bus system. Larger
population sizes and more state variables were
necessary for achieving high-quality solutions, but they
Volume 04 Issue 12-2024
24
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
significantly increased the time required to obtain
those solutions.
GA Parameters Impact: The choice of Genetic
Algorithm parameters, such as population size,
crossover rate, and mutation rate, also played a crucial
role in the optimization process. It was observed that a
population size of around 100 individuals, with a
crossover rate of 0.8 and a mutation rate of 0.02,
provided a good balance between solution quality and
convergence speed across all case studies. However,
for larger systems, the population size needed to be
increased to 150 or 200 to maintain effective
convergence, though this did lead to longer
computational times.
DISCUSSION
The findings underscore the significance of selecting
appropriate control and state variables when applying
Genetic Algorithms to the Optimal Power Flow
problem. The results demonstrate that a careful
balance must be struck between the number of
variables used in the optimization and the
computational resources available. On one hand, a
larger set of control and state variables enables the GA
to model the system more accurately, leading to better
optimization results in terms of cost and system
efficiency. On the other hand, an increase in variables
results in longer computation times, which can be
impractical for real-time applications in large power
systems.
The impact of GA parameters such as population size,
crossover rate, and mutation rate is crucial for
optimizing the GA's efficiency. Too small a population
leads to a lack of diversity, which can hinder the
algorithm's ability to explore the solution space
adequately. Conversely, too large a population leads to
an increase in computational time without significant
improvements in solution quality. This study found that
the optimal GA settings depend heavily on the size and
complexity of the power system being analyzed. For
smaller systems, a standard set of GA parameters
works well, but for larger systems, adjustments are
needed to ensure convergence within a reasonable
timeframe.
Furthermore, the results emphasize that while
reducing the number of variables can speed up the
solution process, it may not be suitable for all power
systems, especially for those with a complex network
and multiple constraints. Therefore, the study
recommends that a careful parametric analysis be
performed for each specific system before deciding on
the optimal set of control and state variables, as well
as GA parameters, to ensure both efficiency and
accuracy.
CONCLUSION
This parametric study demonstrates the critical role
that the selection of control and state variables plays in
the performance of Genetic Algorithms for solving the
Optimal Power Flow problem. The results indicate that
while reducing the number of variables can improve
computational efficiency, it may compromise the
quality of the solution, especially for more complex
power systems. A balanced approach that includes a
moderate number of both control and state variables
is recommended for achieving an optimal solution with
reasonable computational resources.
Moreover, the study highlights the importance of
tuning GA parameters, such as population size,
crossover rate, and mutation rate, for different system
sizes and complexities. A tailored approach to both
variable selection and GA configuration will lead to the
most effective and practical applications of Genetic
Algorithms in real-world power system optimization.
Volume 04 Issue 12-2024
25
American Journal Of Applied Science And Technology
(ISSN
–
2771-2745)
VOLUME
04
ISSUE
12
Pages:
20-25
OCLC
–
1121105677
Publisher:
Oscar Publishing Services
Servi
In conclusion, the study provides valuable insights into
how Genetic Algorithms can be optimized for OPF
problems in power systems. It suggests that future
research should focus on developing adaptive
algorithms that can automatically adjust the selection
of control and state variables based on the specific
characteristics of the system being optimized,
improving the scalability and applicability of GA-based
OPF solutions.
REFERENCE
1.MomohJA,KoesslerRJ,BondMS,StottB,SunD,Papalex
opoulosA,RistanovicP.Challengesto optimal power
flow.IEEE Trans. Power Syst.1997;12:444
–
455.
2.ChristieRD,WollenbergBF,WangensteenI.Transmissi
on managementin thederegulated environment.Proc.
IEEE.2000;88:170
–
195.
3.StottB,MarinhoJL.Linear programming for power
systemnetwork securityapplications.IEEE Trans.
Power Apparat. Syst.1979;PAS-98:837
–
848.
4.WeiH, SasakiH, KubokawaJ, YokoyamaR.An interior
point nonlinear programmingfor optimal power flow
problems witha novel data structure.IEEE Trans. Power
Syst.1998;13:870
–
877.
5.MomohJA,El-HawaryME, AdapaR.A review of
selected optimal power flowliterature to 1993,”IEEE
Trans. Power Syst., pt. I and II.1999;14:96
–
111.
6.ReidGF,HasdorfL.Economic dispatch using
quadratic programming.IEEE Trans.Power Apparat.
Syst.1973;PAS-92:2015
–
2023.
7.SunDI,
AshleyB,
BrewerB,HughesA,.
TinneyWF.Optimal
power
flow
by
Newtonapproach,”IEEE
Trans.
Power
Apparat.
Syst.1984;PAS-103:2864
–
2880.
8.MomohJA,ZhuJZ.Improvedinterior point method
for
OPF
problems,”IEEE
Trans.Power
Syst.1999;14:1114
–
1120.
9.Tognola G,BacherR.Unlimited point algorithm for
OPF problems,”IEEE Trans.Power Syst.1999;14:1046–
1054.
10.Mirko Todorovski,Dragoslav Rajiˇcic ́
.Senior
Member, IEEE, “An InitializationProcedure in Solving
Optimal Power Flow by GeneticAlgorithm”,IEEE
TransactionsonPower Systems. 2006;21(2):480-487
