An efficient modified differential evolution algorithm for solving constrained non-linear integer and mixed-integer glob
- PDF / 1,390,774 Bytes
- 19 Pages / 595.276 x 790.866 pts Page_size
- 36 Downloads / 182 Views
ORIGINAL ARTICLE
An efficient modified differential evolution algorithm for solving constrained non-linear integer and mixed-integer global optimization problems Ali Wagdy Mohamed1
Received: 12 August 2015 / Accepted: 9 December 2015 Springer-Verlag Berlin Heidelberg 2015
Abstract In this paper, an efficient modified Differential Evolution algorithm, named EMDE, is proposed for solving constrained non-linear integer and mixed-integer global optimization problems. In the proposed algorithm, new triangular mutation rule based on the convex combination vector of the triplet defined by the three randomly chosen vectors and the difference vectors between the best,better and the worst individuals among the three randomly selected vectors is introduced. The proposed novel approach to mutation operator is shown to enhance the global and local search capabilities and to increase the convergence speed of the new algorithm compared with basic DE. EMDE uses Deb’s constraint handling technique based on feasibility and the sum of constraints violations without any additional parameters. In order to evaluate and analyze the performance of EMDE, Numerical experiments on a set of 18 test problems with different features, including a comparison with basic DE and four state-ofthe-art evolutionary algorithms are executed. Experimental results indicate that in terms of robustness, stability and efficiency, EMDE is significantly better than other five algorithms in solving these test problems. Furthermore, EMDE exhibits good performance in solving two highdimensional problems, and it finds better solutions than the known ones. Hence, EMDE is superior to the compared algorithms.
& Ali Wagdy Mohamed [email protected] 1
Operations Research Department, Institute of Statistical Studies and Research, Cairo University, Giza 12613, Egypt
Keywords Evolutionary computation Global optimization Differential evolution Triangular mutation Mixed-integer non-linear programming (MINLP)
1 Introduction Generally, optimization is the process of finding the best result for a given problem under certain conditions. In real world problems, applications and different fields of science and engineering, the optimization problems are subject to different types of objective functions and constraints with different type of variables [1]. Thus, most of these problems can be formulated as mixed integer non-linear programming problems (MINLP) that involve continuous as well as integer decision variables. These problems are recognized as a class of NP complete problems plus due to their combinatorial nature, are considered difficult problems [2, 3]. This class of optimization problems frequently arise in various real-world problems and application fields such as mechanical design [4], Synthesis of chemical process flow sheets and design of materials [5], scheduling [6], network design [7], feature selection [8], vehicle routing [9], strategic planning [10], data classification [11] and many more [12, 13]. There are two types of MINLP according to the
Data Loading...
