A binary differential evolution algorithm learning from explored solutions

Although real-coded differential evolution (DE) algorithms can perform well on continuous optimization problems (CoOPs), designing an efficient binary-coded DE algorithm is still a challenging task. Inspired by the learning mechanism in particle swarm optimization (PSO) algorithms, we propose a binary learning differential evolution (BLDE) algorithm that can efficiently locate the global optimal solutions by learning from the last population. Then, we theoretically prove the global convergence of BLDE, and compare it with some existing binary-coded evolutionary algorithms (EAs) via numerical experiments. Numerical results show that BLDE is competitive with the compared EAs. Further study is performed via the change curves of a renewal metric and a refinement metric to investigate why BLDE cannot outperform some compared EAs for several selected benchmark problems. Finally, we employ BLDE in solving the unit commitment problem (UCP) in power systems to show its applicability to practical problems.