Differential evolution enhanced by combining group learning and elite learning

Abstract

Differential evolution (DE) is fully validated as a feasible algorithm for solving optimization problems. Additionally, for the complex optimization problems with high dimension, the traditional DE suffers from slow convergence. This paper proposes an enhanced DE algorithm that combines group learning and elite learning. The proposed algorithm improves the global search capability while guaranteeing a certain convergence speed. Through extensive experiments we confirm the superior competitiveness of the proposed DE algorithm compared to the traditional ones.