| Summary: | 碩士 === 亞東技術學院 === 資訊與通訊工程研究所 === 102 === In the traditional artificial bee colony algorithm, bees only use one dimension when moving to find food sources. This type of movement leads to the bees’ obviously diminished ability to exchange information. In order to improve this problem, the dynamic information exchange based artificial bee colony algorithm was put forth in this study.
In this paper, the dynamic information adjustment mechanism is proposed. A suitable number of dimensions will involve for bees’ movement in current generation according to solution searching status. It can be used to provide needed information for the bees’ movement, thereby enhancing their information exchange ability. In addition, in order to enhance bees’ search for better food sources when moving, the elite mechanism and the jumping mechanism were also proposed to improve accuracy of bees’ movement. The jumping mechanism, whose concept is similar to crossover mechanism of genetic algorithm, will force bees move to around the best bee. Further, the elite mechanism will make the best bee to provide its information with other bees and guide them toward to potential direction for following movement.
In the experiments, the CEC 2005 test functions with 10, 30 and 50 dimensions are adopted to test the proposed method and compared it with six ABC related works. From the results, it can be observed that the proposed method performed better performance than six variants of ABC approaches. The proposed method performs excellent results on unimodal, multimodal and hybrid composition functions.
|
|---|
