Multi-Subpopulation Artificial Bee Colony Algorithm Based on Individual Classification

The artificial bee colony algorithm faces difficulties in insufficient search performance when tackling intricate optimization tasks. To improve this problem and enhance the algorithm's ability to effectively balance the exploration and exploitation, we propose a multi-subpopulation artificial bee colony algorithm that utilizes individual classification to enhance performance(DZABC). In DZABC, the whole bee population was segmented into three distinct sub-populations based on varying levels, and each sub-population adopted strategies with different characteristics as candidate strategies according to its unique population characteristics. This diversity of strategies enables algorithms to leverage the strengths of different strategies to optimize the overall search performance. The introduction of multi-subpopulation mechanism not only aids in preserving population diversity, but also ensures that the algorithm achieves good convergence performance in the search process. In addition, we design a new judgment mechanism in the scout bee phase to further enhance the algorithm's convergence efficiency. To validate the efficacy of the DZABC algorithm, we apply it to 16 test functions, and demonstrate its robust competitiveness through comparative analysis with other algorithms. Experimental outcomes indicate that the DZABC algorithm outperforms other comparative algorithms to a certain extent, which shows its potential and advantage in solving optimization tasks.