Indexed by:期刊论文

Date of Publication:2016-01-01

Journal:International Journal of Hybrid Information Technology

Included Journals:EI

Volume:9

Issue:7

Page Number:191-206

ISSN No.:17389968

Abstract:Differential evolution (DE) algorithms have been extensively and frequently applied to solve optimizationproblems. Theoretical analyses of their properties are important to understand the underlying mechanismsand to develop more efficient algorithms. In this paper, firstly, we introduce an absorbing Markovsequence to model a DE algorithm. Secondly, we propose and prove two theorems that provide sufficientconditions for DE algorithm to guarantee converging to the global optimality region. Finally, we design two DE algorithms that satisfy the preconditions of the two theorems, respectively. The two proposed algorithmsare tested on the CEC2013 benchmark functions, and compared with other existing algorithms.Numerical simulations illustrate the converge, effectiveness and usefulness of the proposed algorithms. © 2016 SERSC.