
EUROGEN 2015 September 14-16, 2015, Glasgow, UK
Differential Evolution with Local Search and Re-Initialization
Lei Peng*
School of Computer Science,China University of Geosciences(Wuhan)
Advanced Space Concept Laboratory (ASCL)
University of Strathclyde, James Weir Building, 75 Montrose Street, Glasgow, G1 1XJ, UK
Email: penglei0114@gmail.com , lei.peng@strath.ac.uk
Massimiliano Vasile
Advanced Space Concept Laboratory (ASCL)
University of Strathclyde, James Weir Building, 75 Montrose Street, Glasgow, G1 1XJ, UK
Email:massimilano.vasile@strath.ac.uk
Guangming Dai
School of Computer Science,China University of Geosciences(Wuhan)
LuMo Road No.388,Hong Shan District,Wuhan,430074,China
Email: gmdai@cug.edu.cn
Huozhen Hu
School of Computer Science,China University of Geosciences(Wuhan)
LuMo Road No.388,Hong Shan District,Wuhan,430074,China
Email:iamasea12@126.com
Summary
In this paper,we present a hybrid Differential Evolution(DE) algorithm for solving global optimization problems.The
proposed approach,called DELR ,uses differential evolution as a global search algorithm and combines with local
search(LS) operator and periodic re-initialization to obtain an appropriate trade-off between the exploration and
exploitation.A new contraction criterion which is based on the roughness in objective space and maximum distance in
decision space is proposed to decide what time does local search is started.The proposed algorithm is evaluated and
compared to several well known evolutionary algorithms on twenty-one CEC2005 benchmark functions.The results show
that the proposed algorithm provides competitive performance with previous algorithms.
Keywords: Differential evolution,re-initialization,local search,roughness.
1 Introduction
Differential Evolution was first proposed by Storn
and Price [1]in 1995 as a highly competitive EA
designed to solve global numerical optimization
problems.It shares similarities with previous evolutionary
algorithms(EAs).For example,DE works with a population
of solutions, called vectors,it uses recombination and
mutation operators to generate new vectors and,finally,it
has a replacement process to discard the less fit vectors.DE
uses real encoding to represent solutions.Some of the
differences with respect to other EAs are the following:DE
uses a special mutation operator based on the linear
combination of three individuals and a uniform crossover
operator.It has several attractive features.Besides being an
exceptionally simple structure,it is significantly faster and
robust for solving numerical optimization problem and
is more likely to find the global optimum.In the past few
decades,DE has been successfully used in many real-world
applications,such as space trajectory design,linear antenna
arrays,engineering design,vehicle routing problem,and so
on.
Despite having several striking features and successful
applications to different fields,DE has sometimes been
shown slow convergence and accuracy of solutions when
the problem is hard to explore.Many efforts have been made
to improve the performance of DE and many variants of DE
have been proposed.
Brest [2]proposed a novel approach to the self-adapting
control parameter of DE,called jDE.Each individual in
the population is extended with parameter values.The
control parameters are adjusted by means of evolution.The
better control parameters are propagated by the better
1