A multi-role steered artificial bee colony algorithm with variable dimensionality perturbation for multimodal optimization problems

Yaxin Kang, Haibo Yu, Li Kang, Gangzhu Qiao, Dongpeng Guo, Jianchao Zeng

Abstract: To address the issues of slow and premature convergence of artificial bee colony (ABC) for handling complex multimodal optimization problems, this paper proposes a multi-role steered variable dimensionality perturbation ABC called MrABC-vdp. MrABC-vdp assigns the nectar sources four distinct roles based on their quality at each iteration. To strike the global exploration and local exploitation equilibrium of the solution space, the bee colony's search behavior is tightly controlled and corrected based on the synergistic relationship established between various preference search equations under multi-role nectar source guidance and the employed bee, onlooker bee, and scout bee phases. To achieve the dynamic trade-off of the global exploration speed and local convergence accuracy of the bee colony across the iteration, a variable dimensionality perturbation is imposed on the candidate nectar sources by a simulated binary crossover operator with time-varying crossover probability. Experimental results on the CEC 2017 multimodal benchmarks show that the proposed MrABC-vdp has a significant performance advantage and strong robustness versus nine advanced homogenous and heterogenous swarm-inspired intelligent algorithms.