A Self-adaptive Artificial Bee Colony Algorithm with Guard Stage for Global Optimization

Abstract

The artificial bee colony (ABC) algorithm is a heuristic optimization algorithm based on the behavior of honeybee swarms. Inspired by particle swarm optimization (PSO) and differential evolution (DE) algorithms, we propose an improved ABC algorithm, named SAG-ABC, which incorporates a self-adaptive employed bees and guard stage to construct a more efficient algorithm. This algorithm combines the advantages of ABC algorithm, which has good exploration capability and global search ability and ease of implementation with fewer control parameters, DE and PSO algorithm, which exchange information with several individuals and utilize the history best information. The searching strategies in these different swarm intelligent algorithms are presented. The information is exchanged among individuals or elements. For the new SAG-ABC algorithm, the self-adaptive employed bees are guided by the global history best bee to enable search in a wider area. Then the search results are adapted to a smaller area. The guard stage is applied to improve the search performance of the employed bees phase by controlling the frequency with which the employed bees abandon the food source. Comparisons between the PSO algorithm, DE algorithm and ABC algorithm are made based on 16 benchmark functions. The results demonstrate the good performance and searching ability of the proposed algorithm.