Differential Evolution Algorithm Based on a Competition Scheme

Abstract

Differential Evolution (DE) is a simple yet powerful population-based metaheuristic algorithm to solve global optimization problems. In this paper, a novel differential evolution is proposed based on the competition among candidate solutions. In the proposed algorithm, the candidate solutions are divided into two groups including losers and winners based on a competition among candidate solutions. Winners generate new candidate solutions based on the DE’s standard mutation and crossover operators, whereas losers learn from the winners. To this end, both crossover and mutation are changed for the loser ones. Competition is not performed in each iteration. In this regard, a new control parameter representing the competition period is introduced. We assess the performance of the proposed algorithm on CEC2017 benchmark functions with three dimensions of 30, 50, and 100. The experimental results verify the effectiveness of the proposed algorithm on the majority of the benchmark functions.