Research On Improving Of Artificial Bee Colony Algorithm

Since the 40 s of the last century,researchers have designed some novel algorithms,by means of inspirations from the natural world,such as evolution strategy,genetic algorithm(GA),particle swarm optimization(PSO),and artificial bee colony(ABC)algorithm.There are elegant theories for these algorithms that have been applied in a variety of areas.The ABC algorithm simulates the intelligence and social behavior of honeybee swarms,with four main concepts: employed bees,onlookers,scout bee and honey sources.Each employed bee carries information about a specific source;from all of the sources,each onlooker selects a relative better one;there is one scout who knocks out a relative poor source.The algorithm has the advantages of fewer control parameters,strong robustness,fast convergence,high flexibility and easy implementation.In this paper,combing the advantages of ABC,GA and PSO,an improved ABC algorithm is proposed,with several features: a design of merging two branches is used;a reverse learning strategy and a normal distribution idea are introduced into the initialization;to keep the equal number of sources in the swarm for each branch,two new sources are created for a crossover operation;the execution controlling of the operations of each branch makes use of the thought of dynamic equilibrium;to dynamically determine the aggregative distribution of sources,two parameters for the iterative number and the number of individuals are added to the variance equation.By experiments on a set of 12 benchmark functions,the results demonstrate that the convergence precision and convergence rate of the improved algorithm are definitely enhanced.