Research On Artificial Bee Colony Algorithm And Its Application

Artificial bee colony (ABC) algorithm is a swarm intelligence randomoptimization algorithm based on the particular intelligent behavior of honeybee swarms,and it provides a new method for solving global optimization problem that exists in thefield of science. Since the advantages of less control parameters, easily programmingand simple calculation, it has received more and more attention by researchers.However, there are still some weakness in the artificial bee colony algorithm, such aspremature convergence, easily trapped in local optimum and convergence slower inlater evolution. In order to effectively improve the performance of artificial bee colonyalgorithm, some works on artificial bee colony algorithm are carried out from variousaspects and simple applications are implemented in the paper. The main works are asfollows:(1) Inspired by the differential evolution algorithm’s variation equation andparticle swarm optimization algorithm’s evolution equation, two ABCalgorithms based on modified search equation are presented to overcome theweak exploitation of ABC algorithm’s search equation. The experimentalsimulations show that improved algorithm has a certain degree ofimprovement in the aspect of solution accuracy and convergence.(2) In order to enhance the overall optimization capability of the ABC algorithm,an ABC algorithm based on chaotic differential evolution search is proposedby combining with differential mutation idea and introducing the differentialevolution local search operator at the stage of onlooker bees. The experimentalanalysis indicates that the new algorithm introducing the operator is promisingin terms of convergence rate and has an advantage of solution accuracycompared to ABC algorithm.(3) By introducing the three-points quadratic interpolation at the scout bee stage,an ABC algorithm with quadratic interpolation search strategy is presented toovercome the disadvantages of ABC algorithm easily trapped in local optimum.The proposed algorithm has the ability to help the employed bee trapped inlocal optimal solution break away from the bondage, the numerical simulationexperiments illustrate that the algorithm introducing the quadratic interpolationworks better in the solution accuracy and convergence rate, and thus it showsmore powerful optimal capability of whole swarm. (4) Based on the constraint handling method and the characteristics of the ABCalgorithm, an ABC algorithm for constrained optimization problems is given,and the benchmark functions are used to validate its performance. In addition,as an application to one of our proposed algorithms, the ABC algorithm withquadratic interpolation search strategy proposed in (3) is applied to simpleimage segmentation problems. The simulation results indicate that thealgorithm can obtain the better threshold value and then it can be used toeffectively split images.