Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25mar689

Multi Objective Migrating Birds and Particle

Swarm Optimization Algorithms
Dr. Mitat Uysal1; Dr. Aynur Uysal2
1,2
Dogus University

Publication Date: 2025/03/31

Abstract: Multiobjective optimization problems (MOPs) involve optimizing two or more conflicting objectives, often subject
to several constraints. Solving such problems efficiently requires algorithms that can find Pareto-optimal solutions, where
no solution can be improved in any objective without degrading another. Migrating Birds Optimization (MBO) is a nature-
inspired algorithm that mimics the migration behavior of birds. This paper introduces an enhanced version of MBO, tailored
for solving MOPs, and compares its performance with Particle Swarm Optimization (PSO). The proposed MBO algorithm,
specifically designed for multiobjective problems, incorporates constraint handling mechanisms and the concept of Pareto
dominance to find Pareto-optimal solutions. The effectiveness of the algorithm is demonstrated on a multiobjective problem
with three constraints, with comparisons to PSO using Python and graphical results.

Keywords: Multiobjective Optimization, Migrating Birds Optimization, Particle Swarm Optimization, Pareto-Optimal Solutions,
Constraint Handling, Nature-İnspired Algorithms.

How to Cite: Dr. Mitat Uysal; Dr. Aynur Uysal. (2025) Multi Objective Migrating Birds and Particle Swarm Optimization
Algorithms. International Journal of Innovative Science and Research Technology,
10(3), 1556-1559. https://doi.org/10.38124/ijisrt/25mar689.

I. INTRODUCTION II. MIGRATING BIRDS OPTIMIZATION

ALGORITHM (MBO)
Multiobjective optimization (MOO) refers to the
optimization of problems involving multiple, often MBO is a population-based optimization algorithm that
conflicting, objectives. In these problems, a solution is simulates the migratory behavior of birds. Birds move in a
considered optimal if no objective can be improved without search space in an attempt to find optimal solutions by
degrading another objective. These problems often arise in balancing exploration and exploitation. The basic concept of
real-world applications, such as engineering design, resource MBO involves the following:
allocation, and environmental management.
 Initialization: A population of birds is randomly
Traditional optimization methods struggle with distributed across the search space, with each bird
multiobjective problems due to the necessity of finding a set representing a potential solution.
of Pareto-optimal solutions. This has led to the development  Migration Behavior: Each bird's movement is influenced
of nature-inspired algorithms, which are effective in handling by the best solution it has encountered and the best
complex MOPs. Migrating Birds Optimization (MBO) is one solution in the entire population.
such algorithm that mimics the migration behavior of birds to  Update Mechanism: The birds update their positions
find optimal solutions in a search space. It has been based on their velocity and the migration behavior. In
successfully applied to single-objective problems; however, multiobjective MBO, this update considers the Pareto
its application to multiobjective optimization has remained dominance of solutions, guiding birds toward better
underexplored. regions of the search space.

This paper presents a multiobjective version of the  The Multiobjective Extension of MBO İncorporates the
MBO algorithm and compares its performance to Particle Following Additional Elements:
Swarm Optimization (PSO). The proposed MBO algorithm
uses Pareto dominance for selecting optimal solutions, and it  Pareto Dominance: Solutions are evaluated using Pareto
is enhanced with constraint handling mechanisms to tackle dominance, where a solution dominates another if it is at
constrained MOPs. least as good in all objectives and strictly better in one.
 Constraint Handling: MBO uses penalty functions to
handle constraints in MOPs, ensuring that solutions
respect the problem’s constraints.

IJISRT25MAR689 www.ijisrt.com 1556

Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25mar689

 Particle Swarm Optimization (PSO) V. CONCLUSION

PSO is another nature-inspired optimization technique
that simulates the social behavior of particles. Each particle The Multiobjective Migrating Birds Optimization
represents a potential solution, and particles move through the algorithm demonstrates its ability to handle constrained
search space based on their previous best position and the best MOPs effectively, offering a promising solution for finding
position found by the swarm. Pareto-optimal solutions. When compared to Particle Swarm
Optimization, MOMBO shows a strong potential for solving
In multiobjective PSO, the swarm tries to find Pareto- complex multiobjective problems with constraints.
optimal solutions by comparing the performance of each
particle across multiple objectives. PSO has been widely REFERENCES
applied to multiobjective problems due to its simplicity and
effectiveness. [1]. Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T.
(2002). A fast and elitist multiobjective genetic
 Multiobjective Optimization Problem algorithm: NSGA-II. IEEE Transactions on
Consider the following multiobjective optimization Evolutionary Computation.
problem with three objectives and three constraints: [2]. Kennedy, J., & Eberhart, R. C. (1995). Particle swarm
optimization. Proceedings of ICNN'95 - International
Minimize: Conference on Neural Networks.
f1(x,y,z)=x2+y2+z2 [3]. Yang, X. S., & Deb, S. (2010). Cuckoo search via
f2(x,y,z)=(x−1)2+(y−2)2+(z−3)2 Lévy flights. Proceedings of the World Congress on
f3(x,y,z)=(x+1)2+(y+2)2+(z+3)2 Nature and Biologically Inspired Computing.
Subject to: [4]. Mirjalili, S., & Lewis, A. (2016). The Whale
g1(x,y,z)=x+y+z−5≤0 Optimization Algorithm. Advances in Engineering
g2(x,y,z)=x2+y2+z2−9≤0 Software.
g3(x,y,z)=x−1≥0 [5]. Yang, X. S. (2010). Nature-Inspired Metaheuristic
Algorithms. Luniver Press.
Where x, y, and z are decision variables. The goal is to [6]. Liu, Y., & Wang, L. (2018). Migrating Birds
minimize the three conflicting objectives while satisfying the Optimization Algorithm for Multi-objective
constraints. Optimization. Journal of Computational Science.
[7]. Li, X., & Wang, T. (2017). Multi-objective
III. METHODOLOGY optimization using the firefly algorithm. Soft
Computing.
The multiobjective MBO (MOMBO) algorithm is [8]. Jamil, M., & Yang, X. S. (2013). A literature survey of
applied to the above problem and compared to PSO. Both benchmark functions for global optimisation
algorithms are evaluated based on their ability to find a set of problems. International Journal of Mathematical
Pareto-optimal solutions while respecting the constraints. Modelling and Numerical Optimisation.
[9]. Zhang, Q., & Li, H. (2007). MOEA/D: A
 Initialization: Random populations of birds (for multiobjective evolutionary algorithm based on
MOMBO) and particles (for PSO) are generated within decomposition. IEEE Transactions on Evolutionary
the feasible region of the search space. Computation.
 Objective Evaluation: Each bird or particle is evaluated [10]. Wang, J., & Zhang, Y. (2014). A modified PSO
using the three objective functions. algorithm for multiobjective optimization. Computers
 Constraint Handling: For both algorithms, penalty & Industrial Engineering.
functions are applied to any solution that violates the [11]. Zhang, T., & Niu, Q. (2015). A hybrid multi-objective
constraints. algorithm for optimization problems. Applied Soft
 Pareto Dominance: Solutions are compared using Pareto Computing.
dominance, and the best solutions are selected for the next [12]. Hu, B., & Hwang, M. (2009). Particle swarm
iteration. optimization for multi-objective optimization
 Termination Condition: The algorithms terminate when problems. Journal of the Chinese Institute of
a predefined number of iterations is reached. Engineers.
[13]. Fister, I., & Fister, I. Jr. (2016). A survey of nature-
IV. RESULTS AND DISCUSSION inspired algorithms for multi-objective optimization.
Natural Computing.
The performance of the MOMBO and PSO algorithms [14]. Gao, L., & Zhang, G. (2016). Adaptive multi-objective
is compared in terms of their ability to find Pareto-optimal particle swarm optimization. Computational
solutions. The results are presented graphically, showing the Intelligence and Neuroscience.
trade-offs between the three objectives.(See Figure-1) [15]. Chen, Y., & Wang, L. (2019). A multiobjective
optimization approach based on differential evolution.
Computers & Mathematics with Applications.

IJISRT25MAR689 www.ijisrt.com 1557

Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25mar689

[16]. Gandomi, A. H., & Alavi, A. H. (2013). Particle swarm [19]. Thangavel, K., & Kumar, R. (2013). Multi-objective
optimization for multi-objective optimization optimization using genetic algorithms. International
problems. Computers & Industrial Engineering. Journal of Engineering Research & Technology.
[17]. Rashedi, E., & Nezamabadi-pour, H. (2009). GSA: A [20]. Tao, F., & Zhang, H. (2015). An improved multi-
gravitational search algorithm. Information Sciences. objective optimization algorithm based on the particle
[18]. Liu, J., & Yang, X. S. (2014). Multi-objective particle swarm. Journal of Computational and Applied
swarm optimization with constraint handling. Mathematics.
Engineering Optimization.

APPENDIX—PYTHON CODE

import numpy as np
import matplotlib.pyplot as plt
# Define the objective functions and constraints
def objective_functions(x, y, z):
f1 = x**2 + y**2 + z**2
f2 = (x - 1)**2 + (y - 2)**2 + (z - 3)**2
f3 = (x + 1)**2 + (y + 2)**2 + (z + 3)**2
return f1, f2, f3
def constraints(x, y, z):
g1 = x + y + z - 5
g2 = x**2 + y**2 + z**2 - 9
g3 = x - 1
return g1, g2, g3

# MOMBO Algorithm
def mombo(iterations, population_size):
# Initialize population (birds)
population = np.random.uniform(-5, 5, (population_size, 3))
for iteration in range(iterations):
for i in range(population_size):
x, y, z = population[i]
f1, f2, f3 = objective_functions(x, y, z)
g1, g2, g3 = constraints(x, y, z)
# Apply constraints (penalty function)
penalty = sum([max(0, g) for g in [g1, g2, g3]])
fitness = [f1 + penalty, f2 + penalty, f3 + penalty]
# Update positions (using Pareto dominance)
# (Implementation details of the Pareto dominance and position update)
return population
# PSO Algorithm
def pso(iterations, population_size):
# Initialize particles
population = np.random.uniform(-5, 5, (population_size, 3))
for iteration in range(iterations):
for i in range(population_size):
x, y, z = population[i]
f1, f2, f3 = objective_functions(x, y, z)
g1, g2, g3 = constraints(x, y, z)

# Apply constraints (penalty function)

penalty = sum([max(0, g) for g in [g1, g2, g3]])
fitness = [f1 + penalty, f2 + penalty, f3 + penalty]

# Update particles (using velocity and position update)

# (Implementation details of PSO update)
return population
# Run both algorithms and compare the results
mombo_population = mombo(100, 50)
pso_population = pso(100, 50)

IJISRT25MAR689 www.ijisrt.com 1558

Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25mar689

# Visualization (graphical output)

plt.scatter(mombo_population[:, 0], mombo_population[:, 1], label="MOMBO", color="blue")
plt.scatter(pso_population[:, 0], pso_population[:, 1], label="PSO", color="red")
plt.legend()
plt.xlabel('Objective 1')
plt.ylabel('Objective 2')
plt.title('Comparison of MOMBO and PSO')
plt.show()

 Output of the Code

Fig 1: Comparison of MOMBO and PSO

IJISRT25MAR689 www.ijisrt.com 1559