Abstract
This paper proposes a memetic algorithm by integrating adaptively a local search approach with a recently proposed variant of differential evolution, reflected adaptive differential evolution with two external
archives (RJADE-TA). The main objective is to bring together the exploration factor of differential evolution and exploitative component of local
search to solve continuous optimization problems. A novel hybrid local
search mechanism is proposed and demonstrated leading to a crossbred
version of RJADE-TA. In other words, the best solutions after a regular
toll of global search are migrated to an archive, where Davidon Fletcher
Powell local search method is implemented to the migrated solutions. Afterwards, the population is updated with new reflected sol utions to prevent premature convergence. The proposed approach is novel in the sensethat most of the algorithms store only inferior or superior solutions in the
archives. None of the algorithms implemented the local search inside the
archive. Thus, this combination is a new one. To evaluate the merit of
developed meme, a benchmark suite of complex 28 functions from CEC
2013 test problems is selected and implemented. The experimental results demonstrate that this integration of local search strategy can further
improve the performance of RJADE-TA. They further reveal that the proposed meme outperforms differential evolution based algorithms on most
of the tested problems.
Rashida Adeeb Khanum, Muhammad Asif Jan, Wali Khan Mashwani, Hidayat Ullah Khan, Saima Hassan. (2019) RJADE/TA Integrated with Local Search for Continuous Nonlinear Optimization Ra, Punjab University Journal of Mathematics, Volume 51, Issue 4.
-
Views
488 -
Downloads
78
Article Details
Volume
Issue
Type
Language
