Multi-population artificial bee colony algorithm based on the nearest neighbour partition

Abstract

The artificial bee colony (ABC) has shown great potential among many swarm intelligence optimisation algorithms (SIOAs). However, ABC still shows deficiencies in some aspects. The weak exploitation ability makes the original ABC hard to achieve promising results when dealing with complex optimisation problems. The roulette selection method may not work at the late search stage. To make up for these deficiencies, a modified multi-population ABC with the nearest neighbourhood partition (namely NNPMABC) is proposed in this paper. Firstly, a novel partition method is used to divide the swarm into several subgroups. Then, three improved search strategies and a new selection method based on the nearest neighbour partition are designed. In addition, a new search strategy is constructed for the scout bee stage. To prove the effectiveness of NNPMABC, 22 benchmark problems are tested. Results show NNPMABC performs the best among six ABCs.