International Journal of Research and Engineering
ISSN: 2348-7860 (O) | 2348-7852 (P) | Vol. 5 No. 9 | September-October 2018 | PP. 500-507
Digital Object Identifier DOI® http://dx.doi.org/10.21276/ijre.2018.5.9.2
Copyright © 2018 by authors and International Journal of Research and Engineering
This work is licensed under the Creative Commons Attribution International License (CC BY).
creativecommons.org/licenses/by/4.0 | |
ORIGINAL
ARTICLE
Solving Vehicle Routing Problem using
Ant Colony Optimisation (ACO) Algorithm
Author(s):
1
*Wan Amir Fuad Wajdi Othman,
1
Aeizaal Azman Abd Wahab
1
Syed Sahal Nazli Alhady,
1
Haw Ngie Wong
Affiliation(s):
1
School of Electrical & Electronic Engineering,
Universiti Sains Malaysia, Malaysia
*Corresponding Author: wafw_othman@usm.my
Abstract - Engineering field usually requires
having the best design for an optimum
performance, thus optimization plays an important
part in this field. The vehicle routing problem
(VRP) has been an important problem in the field
of distribution and logistics since at least the early
1960s. Hence, this study was about the application
of ant colony optimization (ACO) algorithm to
solve vehicle routing problem (VRP). Firstly, this
study constructed the model of the problem to be
solved through this research. The study was then
focused on the Ant Colony Optimization (ACO).
The objective function of the algorithm was studied
and applied to VRP. The effectiveness of the
algorithm was increased with the minimization of
stopping criteria. The control parameters were
studied to find the best value for each control
parameter. After the control parameters were
identified, the evaluation of the performance of
ACO on VRP was made. The good performance of
the algorithm reflected on the importance of its
parameters, which were number of ants (nAnt),
alpha (α), beta (β) and rho (ρ). Alpha represents the
relative importance of trail, beta represents the
importance of visibility and rho represents the
parameter governing pheromone decay. The route
results of different iterations were compared and
analyzed the performance of the algorithm. The
best set of control parameters obtained is with 20
ants, α = 1, β = 1 and ρ = 0.05. The average cost and
standard deviation from the 20 runtimes with best
set of control parameters were also evaluated, with
1057.839 km and 25.913 km respectively. Last but
not least, a conclusion is made to summarize the
achievement of the study.
Keywords: Vehicle Routing Problem, Ant Colony
Optimization, ACO, VRP, Swarm Algorithm
I. INTRODUCTION
The study is about solving vehicle routing problem (VRP)
using ant colony optimization (ACO) algorithm. This is a
software-based project. VRP generalizes the well-known
travelling salesman problem (TSP). The study can be
divided into two parts, vehicle routing problem (VRP) and
ant colony optimization (ACO) algorithm.
The vehicle routing problem (VRP) has been an important
problem in the field of distribution and logistics since at
least the early 1960s [1]. VRP research accelerated during
the 1990s [2]. Researchers could develop and implement
more complex search algorithms due to the improvement of
microcomputer capability and availability. During this era
the term meta-heuristics was introduced to name a number
of search algorithms for solving these VRPs as well as other
combinatorial optimization problems [3].
The technical definition of vehicle routing problem (VRP)
states that m vehicles initially located at a depot are to
deliver discrete quantities of goods to n customers. The aim
of a VRP is to determine the optimal route used by a group
of vehicles when serving a group of users. The objective of
VRP is to minimize the overall transportation cost. The
solution of the classical VRP is a set of routes which all
begin and end in the depot, and which satisfies the
constraint that all the customers are served only once. The
transportation cost can be improved by reducing the total
travelled distance and by reducing the number of the
required vehicles.
Two important classes of population-based optimization
algorithms are evolutionary algorithms and swarm
intelligence-based algorithms [3]. In this research, swarm
intelligence-based algorithm is chosen to be applied on VRP.