Application And Research On Artificial Bee Colony Algorithm

With the rapid development of science and technology, problems in research andengineering field become more and more complex. For its large amount of computation,non-traditional optimization methods have been able to compute. The intelligent bionicoptimization algorithms that rising in recent decades can efficiently solve those problems thattraditional methods difficultly solved and bring Gospel to complex problems.Artificial bee colony (ABC) algorithm is in the making in recent years. It is ameta-heuristic bionic swarm intelligence algorithm that inspired by the natural behavior ofhoney bees. Related experiments have confirmed that ABC algorithm has obvious advantagescompared with other intelligent bionic algorithm in solving the high-dimensional functionproblem. Because of Its simple, efficient and robust distinctive features, more and morescholars pay their attention and favor on this algorithm. This article will study ABC algorithmin depth, and applys it to the traveling salesman problem (TSP). The main contents are asfollows:First, this paper has discussed global search and local search capabilities and analyzedthe global search and local search capability of ABC algorithm. That provides a theoreticalbasis for improving the ABC algorithm.Second, to the phenomenon that employed bees’ evolution is out of sync that due to theonlooker bees’ preference in ABC algorithm, this paper proposed a collaborative evolutionaryopen artificial bee colony algorithm. This algorithm does several times additional search ope-ration in each iteration of ABC algorithm. These additional search operations optimize thecurrent poor solution using the information that provided by currently optimum solution. Itnot only improves the efficiency of a single search, but can lead the entire bee populations toevolve synchronously and also can reduce the search step length. Thus it can improve theoverall performance of the ABC algorithm.Third, in the co-evolutionary algorithm, this paper proposed four methods to using theinformation of current optimum solution to optimize the current poor solution. Method one:using a dimension that chose from one better solution directly replaces a dimension valuewhi-ch is in the bad solution. Method two: using relatively modest approach. The designedexperi-ment proved that these two methods are much better than ABC algorithm In terms ofthe rate of convergence and the accuracy of the solution. Method three and method fourmainly aim at functions that have a global optimum solution that its all dimensions value are equal or closed. The designed experiments show that for such functions the methods are betterthan the ABC algorithm in terms of optimizing results, and it can effectively solve some of theproblems that ABC algorithm is difficult to solve.Fourth, this paper proposed a new solution on the traveling salesman problem based onartificial bee colony algorithm. This method not only takes advantage of the foraging mecha-nism of honey bees, but also learns from the one-dimensional update strategy that used inartificial bee colony algorithm. During conducting path optimization, it only makes minimalchanges on the basis of candidate paths at a time, that it to say it only changes two basic linkpaths once. While most of the traditional solutions use integrated update mechanism.Experiments show that the proposed algorithm is more accurate and efficient than the basicant colony algorithm.