Multi-population Based Search Strategy Ensemble Artificial Bee Colony Algorithm with a Novel Resource Allocation Mechanism

Abstract

Artificial bee colony algorithm (ABC) is a simple yet effective biologically-inspired optimization method for global numerical optimization problems. However, ABC often suffers from slow convergence due to its solution search equation performs well in exploration but badly in exploitation. Moreover, all food sources are assigned with almost equal computing resources so that good solutions are not being fully exploited. In order to address these issues, we propose a multi-population based search strategy ensemble ABC algorithm with a novel resource allocation mechanism (called MPABC_RA). Specifically, in employed bee phase, all food sources are divided into three subgroups according to their quality. Then each subgroup uses different search equations to find better solutions. By this way, better tradeoff between exploitation and exploration can be obtained. In addition, the superior solutions in onlooker bee phase are allocated with more resources to evolve. And onlooker bees fully exploit the area between the locations of the selected superior solutions and the current best solution by a novel search equation. We compare MPABC_RA with four state-of-the-art ABC variants on 22 benchmark functions, the experimental results show that MPABC_RA is significantly better than the compared algorithms on most test functions in terms of solution accuracy, convergence rate and robustness.