An Improved Artificial Bee Colony Algorithm: Particle Bee Colony
WANG Ji-chao,LI Qing,CUI Jia-rui,ZUO Wen-xiang,ZHAO Yue-fei
DOI: https://doi.org/10.13374/j.issn2095-9389.2018.07.014
2018-01-01
Abstract:With an aim to address the disadvantages of the artificial bee colony algorithm of slow convergence speed and ease of falling into the local optimum in the later period of the evolution process as well as to improve the traditional artificial bee colony algorithm, the concept of the "global optimum" in particle swarm optimization is introduced. Therefore, an improved artificial bee colony algorithm, called particle bee colony (PBC), is proposed herein. First, the concept of degree toward optimum is proposed for measuring the degree to which the leader approaches or is removed from the "global optimum" in a limited iteration process. The individuals' values of degree toward optimum denote their "development potentials." The individuals that have a low degree toward optimum require a great mutation extent to find a good solution. Second, a new colony of bees, initiated by the particle bee, is uniquely developed. In mutation period, the leader will be changed into the scout or the particle bee according to the value of the degree toward optimum. The appearance of particle bees can increase the population diversity and expand the search area to a large extent. Next, analysis reveals that the sequence of population of the PBC is a finite homogeneous Markov chain and the population evolution process is monotonous. On the basis of the above observations, it can be proved that the population sequence of the proposed algorithm converges to the global optimum solution set with probability 1. Last, the algorithm proposed in this study is applied to numerical simulations of several classical test functions. Furthermore, the proposed algorithm is compared with the traditional artificial bee colony algorithm and other improved bee colony algorithms. The simulation results show that PBC increases the population dispersion and broadens the search area, thereby allowing the proposed algorithm to achieve fast convergence rate and high optimization accuracy.