In order to solve the insufficient development ability of artificial bee colony algorithm, this paper proposed a multi-population artificial bee colony algorithm with variable population size(DMABCPS). It divided the whole population into several subpopulations with the individual mean position as the central point. It used three different strategies in the employer bee phase to ensure the exploitation of the good population, the balance of the intermediate population and the exploration of the poor population. In the onlooker bee, it selected two search strategies adaptively based on the selection mechanism of success rate. In addition, it established a new probabilistic selection model to select subpopulations and their internal individuals. Finally, it used 22 standard function test sets to verify that the proposed algorithm is superior to current algorithms.

[2021-06-05] Printed Article /doc/2020.07.0174/v2