A novel multiobjective optimization algorithm based on bacterial chemotaxis
Marı
´
a Alejandra Guzma
´
n
a,c,
, Alberto Delgado
b
, Jonas De Carvalho
c
a
Department of Mechanical and Mechatronics Engineering, National University of Colombia, Carrera 30 # 45-03 - Building 453, Bogota, Colombia
b
Department of Electrical and Electronics Engineering, National University of Colombia, Carrera 30 # 45-03 - Building 453, Bogota, Colombia
c
Department of Mechanical Engineering, EESC-University of Sao Paulo, Avenida do Trabalhador s
~
aocarlense 400, Sao Carlos, Brazil
article info
Article history:
Received 3 December 2008
Received in revised form
24 September 2009
Accepted 28 September 2009
Available online 5 November 2009
Keywords:
Multiobjective optimization
Bacterial chemotaxis
Bio-inspired techniques
Pareto Optimal Front
Chemotactical strategy optimization
abstract
In this article a novel algorithm based on the chemotaxis process of Echerichia coli is developed to solve
multiobjective optimization problems. The algorithm uses fast nondominated sorting procedure,
communication between the colony members and a simple chemotactical strategy to change the
bacterial positions in order to explore the search space to find several optimal solutions. The proposed
algorithm is validated using 11 benchmark problems and implementing three different performance
measures to compare its performance with the NSGA-II genetic algorithm and with the particle swarm-
based algorithm NSPSO.
& 2009 Elsevier Ltd. All rights reserved.
1. Introduction
Most real-world optimization problems require making deci-
sions involving two or more goals that typically are in contra-
diction with each other. When these goals are the minimization or
maximization of functions they are typically referred to multi-
objective optimization (MO). From the 1950s, in the area of
operational research, a variety of methods known as classical has
been developed for the solution of multiobjective optimization
problems (MOP). These methods are based on formal logic or
mathematical programming. Some of the most representative
classical methods are linear programming, the weighted sum
method and the goal programming method (Dantzig and Thapa,
1997). As an alternative to classical methods, a variety of
techniques inspired on natural processes has emerged in the last
two decades.
The emulation of nature has inspired scientists in various fields
through the history of mankind. Recently, due to advances in
computing and the emergence of new ideas based on the behavior
of living organisms and natural processes, the techniques inspired
in nature have gained increasing interest motivated by two basic
aspects (De Castro and Von Zuben, 2004):
(1) Traditional methods have proven to be unable to adequately
handling complex problems, characterized by the lack of
complete mathematical models and the manipulation of a
large number of variables.
(2) To a variety of engineering problems there is a similar version
in nature.
Among bio-inspired optimization techniques, the most known
are genetic algorithms (AGs). The pioneering work in the practical
application of the fundamentals of AGs to MO is the vector
evaluated genetic algorithm (VEGA) (Schaffer, 1984). At present,
the most popular genetic algorithm for solving MOP is the
nondominated sorting genetic algorithm II (NSGA-II) (Deb et al.,
2002). Another bio-inspired approach is the so-called particle
swarm optimization (PSO), which was recently implemented in
the solution of MOP using algorithms such as nondominated
sorting particle swarm optimizer (NSPSO) (Li, 2003), multi-
objective particle swarm optimization (MOPSO) (Coello et al.,
2004), and time variant multi-objective particle swarm optimiza-
tion (TV-MOPSO) (Kumar et al., 2007). Another interesting
biological process that has been already implemented as an
optimization technique is the bacterial chemotaxis. About this
novel technique Amos et al. (2007) exposed the potential of
implementing bacterial chemotaxis as a distributed optimization
process, recognizing that in natural colonies, it is the interaction
and communication between bacteria the mechanism that
enables them to develop biologically advantageous patterns.
2. Multiobjective optimization problems (MOP)
A multiobjective optimization problem is defined as the
problem of finding a vector of decision variables that satisfies
ARTICLE IN PRESS
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/engappai
Engineering Applications of Artificial Intelligence
0952-1976/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engappai.2009.09.010
Corresponding author at: Department of Mechanical and Mechatronics
Engineering, National University of Colombia, Carrera 30 # 45-03 - Building 453,
Bogota, Colombia. Tel.: +57 1 3165000 14107; fax: +55 16 3373 9402
E-mail address: maguzmanp@unal.edu.co (M. A. Guzma
´
n).
Engineering Applications of Artificial Intelligence 23 (2010) 292–301