Abstract

The Artificial Bee Colony algorithm (ABC) is a swarm intelligence algorithm inspired by honey bee harvesting behavior. It boasts the benefits of minimal parameters and strong exploration capabilities. However, the ABC algorithm is still susceptible to local optima entrapment and lacks consideration of selection probability in the onlooker bee phase, leading to reduced convergence accuracy in later search stages.

人工蜂群算法（ABC）是受蜜蜂采蜜行为启发的群体智能算法，具有参数少、探索能力强等优点，但易陷入局部最优。本文提出分层学习人工蜂群算法（HLABC）：通过分层学习机制划分种群层级，改进雇佣蜂的开发能力与侦查蜂的探索能力。在 CEC2005/2010/2013/2022 四套基准的 69 个函数上，HLABC 平均以 89% 的获胜率优于 ABC 及其变体，并在无线传感器网络部署、智能家居功率调度与图像分割等应用中表现高效稳定。

Keywords