Abstract
The artificial bee colony (ABC) is a population-based metaheuristic that mimics the foraging behaviour of honeybees in order to produce high-quality solutions for optimisation problems. The ABC algorithm combines both exploration and exploitation processes. In the exploration process, the worker bees are responsible for selecting a random solution and applying it to a random neighbourhood structure, while the onlooker bees are responsible for choosing a food source based on a selection strategy. In this paper, a disruptive selection strategy is applied within the ABC algorithm in order to improve the diversity of the population and prevent premature convergence in the evolutionary process. A self-adaptive strategy for selecting neighbourhood structures is added to further enhance the local intensification capability (adaptively choosing the neighbourhood structure helps the algorithm to escape local optima). Finally, a modified ABC algorithm is hybridised with a local search algorithm, i.e. the late-acceptance hill-climbing algorithm, to quickly descend to a good-quality solution. The experiments show that the ABC algorithm with the disruptive selection strategy outperforms the original ABC algorithm. The hybridised ABC algorithm also outperforms the lone ABC algorithm when tested on examination timetabling problems.